UNIVERSITY OF GRONINGEN • APRIL 2010
Inter-country conflict, stock returns and dividend policy:
The Korean case
JELJER DE WAGT ∗
ABSTRACT
This study examines the economic impact of events that happened between North- and South Korea for a sample of ten stock indices, between 1979 and 2009, using a GARCH model. It also investigates the impact that sentiment has on both stock returns and on dividend policy, for the firms listed on the South Korean KOSPI 50 index. We want to determine if there are misconceptions surrounding the Korean conflict, that adversely influence stock indices and firms’ dividend policies. Contrary to what we expected, the results suggest that the individual events do not have an unambiguous significant impact on returns. Sentiment, in the form of a series of positive or negative events, impacts stock returns in South Korea in a different way than in other countries; no such influence is found for these other countries. The impact of sentiment on dividend policy is found to be non-significant. Dividend policy in South Korea is significantly influenced by previous dividend payments and retained earnings, and by a firm’s debt-equity ratio. Excluding the period of the Korean crisis has no impact on our results. Overall, we find no significant evidence that the performance of South Korea’s stock index is suffering from misconceptions surrounding its stringent relationship with North Korea.
Keywords: Korean conflict; dividend policy; stock indices; sentiment; uncertainty JEL code: F51 - International Conflicts; Negotiations; Sanctions
G35 - Payout Policy
∗
Jeljer IJ. de Wagt; Master student at the University of Groningen; student number 1494503. The
INTRODUCTION
The Korean War marked the beginning of the stringent relationship that North and South Korea still have today. The conflict arose from the failed attempts of the two Korean powers to reunify after the country had been divided post World War II, when the Soviet Union took in North Korea and the United States occupied South Korea. Ever since the ending of the Korean War in 1953 several attempts have been done to reunify Korea, but all have been unavailing. In fact, the two countries are officially still at war with each other.1 According to Estrada and Park (2008), there is a wide and growing gap between the two Koreas in terms of political, social, economic and technological development, as well as in overall development. North Korea coped with famine in 2008, while South Korea was the 13th-largest economy in the world in that year. They compare the German unification, which took place in 1990, to a potential unification between the two Koreas, and suggest that Korean unification is likely to be a costly and disruptive process, entailing large adjustment costs. Besides this, Estrada and Park (2008) feel that the poor performance of the German economy since the unification has highlighted further potentially adverse effects of unification for the Korean economy. Reunification thus seems far away.
The communist North Korea, officially the Democratic People’s Republic of Korea, has a history of regional military provocations, proliferation of military-related items, and long-range missile development. It also has its Weapons of Mass Destruction programme and massive conventional armed forces, which far outnumber those of the democracy of South Korea. Moreover, North Korea has nuclear ambitions. The country started to build a nuclear reactor at Yongbyon in 1979 and has already conducted two nuclear tests. Consequently, any conflict that occurs between the two countries is immediately of major concern to the international community.2
The standoff continues to cause unrest. An alleged satellite launch by North Korea in April 2009, which was thought to be a cover-up for the testing of a long distance missile, led to a condemnation by the United Nations Security Council, not the first one for the North Korean regime. The turmoil causes thus not only tension between the two neighbours, but it in fact influences other countries too. South Korea also feels this influence in its own economy.
1
Korea in East Asian and World History
2
Its minister of finance in 2003, Kim Jin Pyo, said that foreigners’ misconceptions about tensions with North Korea were to blame for much of the South Korean stock market’s shaky performance around that time and that this may be scaring investors away from the country. 3 South Korea already receives less foreign investments than average in East Asia. The tension on the peninsula could thus well be a cause of bad performance of the stocks of firms in the south, as Kaun (1990) finds that war, and also the threat of war, tend to have an immediate and somewhat negative impact on general stock prices.
If rapprochements between the two Korea’s lead to improvements in stock market’s performance, more effort can be focused on creating peace. Therefore, it needs to be established whether the situation really has an impact on South Korea’s financial market. Furthermore, since the United Nations condemn the North Koreans almost on a regular basis, it is interesting to see if the underlying conflict impacts other countries’ indices as well. The stock markets of countries in the region, such as Japan, China, Thailand, and Taiwan, could be affected by the unrest between their neighbours. In addition, countries like the United States and Russia could also be affected by the conflicts in the region, as they have always supported the South Koreans and the North Koreans, respectively. Since China has always chosen the side of North Korea in many situations, they have almost always voted against strong measures from the United Nations against the North, they also have a tie to the conflict. China, the United States and Japan are also the three largest trading partners of South Korea.4 Six of the aforementioned countries are also participating in the so-called Six-Party International Talks. In this platform, the two Korea’s, the United States, China, Russia and Japan discuss North Korea’s nuclear weapons programme and try to find a solution for it. The main goal is to reach the verifiable denuclearisation of the regime in the North, in return for aid and security guarantees.5 Maybe one day, when this goal is reached, the two countries can once again become one and the threat of war is gone.
This research aims at establishing how developments in the stringent relationship between North and South Korea affect the stock indices of the countries involved and the dividend policies of firms in South Korea, reflected by the amount of dividend they pay. This paper follows a two-step approach. Firstly, stock market effects of twenty events in the
3
New York Times 4
Background Note: South Korea
5
relationship are examined for neighbouring countries, for those countries taking part in the Six-Party Talks and for Western core markets. The efficient market hypothesis that is put forward by Fama (1970) expects that markets use all available information efficiently and completely in their evaluation of markets, sectors, and individual firms, and the occurrence of each event should thus also have an immediate impact. After that, the focus is on these events influencing the dividend policy of firms in South Korea, where key events represent the start of either a negative or positive period of sentiment in the Korean relation. Events causing a negative or positive sentiment are, for example, the start of the build of a nuclear reactor in 1979 and the signing of the Nuclear Non-Proliferation Treaty by North Korea in 1985. In line with the previous discussion, one would expect firms to keep cash within the firm when the sentiment is negative and pay out more dividends when it is positive (Dittmar, 2008). The main goal of this paper is to examine if former minister of finance Kim Jin Pyo is correct in saying that South Korea’s shaky stock market performance is caused by the misconceptions surrounding the conflict. This is done, for example, by examining if there is such a thing as overreaction to consistent negative patterns of news as Barberis et al. (1998) suggested, and to establish to what extend this is the case. Besides this, if rapprochements lead to improvements in stock market’s performance, more effort should be focused on creating peace. There have not been many previous investigations that look into the impact of the relationship that the two Koreas have and the conflicts that arise between them on stock indices. The South Korean stock market is the main focus of this paper, as the effect of the conflict on firms’ dividend policies in this country is examined. The time period under investigation is the last thirty years, between late 1979 and August 2009. Since the Korean Crisis also occurred during this period, robustness tests are conducted to account for this.
Contrary to what we expected to find, individual events do not have an unambiguous significant influence on the stocks indices under investigation. The dividend policies of the fifty largest firms listed on the KOSPI 50 index are also not significantly influenced by sentiment, nor are the ten indices. South Korea does stand out. We find no evidence that the indices in our sample are suffering from misconceptions surrounding the Korean conflict.
finally the effects of sentiment are discussed. Section II present the data and methodology used in this paper, followed by the empirical results in Section III. Finally, section IV concludes.
I. LITERATURE REVIEW
1. Korean unification
Estrada and Park (2008) evaluate the prospects of unification between South Korea and North Korea from a multidimensional perspective, encompassing the political, social, economic and technological dimensions. They find that the gap between the two Koreas is growing on all areas, and that unification is likely to be a very costly and hard process. South Korea has transformed from a typical poor developing country into an economic powerhouse which is one of the world’s twelve biggest economies, whereas decades of autarky and central planning have reduced North Korea to one of the poorest countries in the world. When looking at the political situation, they observe that South Korea is a thriving democracy whereas North Korea is communist dictatorship. Estrada and Park (2008) conclude that there are thus many parallels between the German unification which took place in 1990 and a potential unification between the two Koreas.
Lim (1997) also notes that the reunification of Germany provides Korea with many lessons since it occurred in a post-Cold War environment. He states that prior to the German reunification, most scholars had not anticipated the sudden fall of the East German regime, nor the eventual reunification. The same situation could now be true for North and South Korea, and Lim (1997) presents his own plan with five steps to reunification. He argues that economic issues, rather than the political ones, could prove to be of the greatest concern to a smooth reunification between the peninsula’s long-divided neighbours.
example for the current Korean situation. Effects of events in the relationship or the emergence of news could be similar today to what happened in 1990.
2. The economic impact of political news
Brooks et al. (2005) examine if the German reunification in 1990 and events leading up to that day had an impact upon country returns, across different nations. They use twenty different major events that took place between the 40th anniversary of the creation of East Germany in October 1989 and December 1990, when the Christian Democratic Party emerged as the winner in the first unified Germany election. The economic impact of each of these events is measured. This impact is quite significant for some events, which can for instance be news releases. For example, on the day that East Germany called a special sitting of Parliament to decide the date of unification with West Germany, the European stock markets increased significantly. The results of Brooks et al. (2005) show a stronger stock market reaction to the selected events in European countries, particularly for those countries that have closer economic links to Germany. This German unification can be seen as an example for the Korean situation, as is mentioned above. Brooks et al. (2005) show that political events can influence the stock market and the economy. This indicates that the countries with an economic link to either North or South Korea, with South Korea being the 13th-largest economy in the world in 2008, could also be disproportionately affected, similar to the countries during the German unification.
Sultan (1995) also examines the stock markets’ reaction to both optimistic and pessimistic news about the German reunification. His results indicate that both types of news led investors to revise their expectations of future returns from Germany related investments, although pessimistic news seemed to have a greater effect. There is evidence that shocks to the German economy were transmitted to the global equity portfolio. This, according to Sultan (1995), suggests that in the absence of barriers to capital flows, it is possible that country-specific risk may be priced in the global financial market. This is consistent with the notion that when country-specific risk becomes part of the global financial risk, such risk cannot be ignored.
Hang Seng Index and on China-related stocks as measured by the Red-Chip Index. For the Hang Seng Index, Chan and Wei (1996) find that favourable political news is correlated to positive returns, and unfavourable news is correlated to negative returns. Besides that, the results from nonparametric tests give strong evidence of the impact of political news on stock market volatility. For the Red-Chip Index, the volatility is affected by political shocks, but the returns are not. They conclude that some stocks can be more influenced by political shocks than others.
Fernandez (2007) analyzes the effect of political conflicts in the Middle East on stock markets worldwide. She shows that political unrest and the ongoing Middle East conflicts have had an effect primarily on the stock markets of countries in that region and in emerging Asian countries. Besides that, the political instability in the Middle East has had a heterogeneous effect on the sensitivity of stock returns in those countries. Fernandez (2007) thus proves that political conflicts also affect stock returns.
The overall effect that political news and news in general can have on stock markets and indices is clear. Developments change the information level of the stock market. Traders who do not use relevant information will make losses. We can therefore expect that some political events alter the beliefs about the future development of a firm, or all the stocks traded in an equity market. This reasoning is in line with the efficient market hypothesis put forward by Fama (1970), who expects that markets use information efficiently and completely in their evaluation of markets, sectors, and individual firms. The literature shows that the German reunification has impacted stock markets around the world, and that political news can indeed have an effect on different stock markets, even when that stock market is in a neighbouring country. It also shows that news influences investors’ expectations about future returns. Finally, both political instability and political unrest also affect stock returns. For the Korean case this proves that news concerning the situation, whether it is good or bad, can also influence the returns on the stock exchanges in the countries under investigation. Besides that, political instability could also be affecting the stocks.
3. The economic impact of war and conflicts
immediate and somewhat negative impact on general stock prices, with the exception being one sample of defence-related firms. Kaun’s (1990) own analysis suggests the same negative reaction for both defence and commercial firms during the Korean War, where increases in hostilities caused significant negative reactions from investors. During the Vietnam War, however, investor attitudes regarding defence firms brightened with increased hostilities. Kaun (1990) sees this as a case of ‘war in a distant land’, where the country that produces the weapons is the only one who gains.
Schneider and Troeger (2006) also study how war affects the economy. They find that war has a negative impact on stock market returns. The authors examine the influence that the political developments within three war regions had on global financial markets; on the CAC in France, the Dow Jones in the United States and the FTSE in the United Kingdom, from 1990 to 2000. Using daily stock market data, the authors show that the conflicts affected the interactions at these financial markets in the Western world negatively. Conflictive events influenced the volatility of the stock market much more strongly than cooperative ones. When it comes to examining the ups and downs of the financial sectors in the conflict regions themselves, Schneider and Troeger (2006) expect the overall negative effect to be more pronounced.
In her 2008 article, Fernandez analyzes how the U.S.’ declaration of the war on terror and the subsequent invasion of Iraq has impacted long-term volatility of stock markets around the world, in four geographic regions. She shows again, just like in her 2007 paper, that political instability in the Middle East had its greatest impact on the volatility of financial markets around the beginning of the war in Iraq. It mostly hit developed markets like the United States, the United Kingdom, and Japan. For the period after 2006, that volatility declined for most indices, except for two oil-related indices. The main conclusion is that stock markets around the world are affected by the war in Iraq.
Saddam’s fall, which may also have indicated the probability of a costly war breaking out, lowered stock prices. These prices adjusted gradually to this information. Results are not all significant for the pre-war period. Amihud and Wohl (2003) found that the ending of the war was associated with lower cost on the economy of the United States and the realization of the benefits it produces, such as lower oil prices and lower risk of terror attacks. Besides that, they observed that when media attention increased, effects on stock prices became quite strong and stock prices adjusted more prompt to information.
Sweeney and Zhang (1999) investigate the impact of India’s nuclear test on the economies of India, China, and Pakistan, and the effects of Pakistan’s nuclear test on the other countries. They find that the nuclear tests caused important economic damage to India and its neighbours. Stock markets fell between 2,70 and 10,59 percent when India conducted its test, and losses were even greater after Pakistan’s nuclear test. Sweeney and Zhang (1999) concluded that in the financial market’s opinion, these tests caused major reductions in both countries’ economic security. The nuclear tests had no important effects on the stock markets of a group of ten countries, including amongst others the Netherlands and Japan, but the tests did have economic effects for countries in southern Asia. However, the main economic damage was done on the two countries that conducted the test and on China, who supported their nuclear programs. China lost its economic security and 6,63 percent of its wealth.
Caplan (2002) examines how war affects the growth of an economy. He finds that wars fought exclusively on foreign soil do have marginally higher real output growth than peacetime periods, in line with Kaun (1990). Real growth during all other wars, however, is sharply below peacetime levels. Caplan (2002) concludes that real output growth clearly declines substantially during domestic wars, even though it slightly increases during foreign wars.
All in all, the existing literature is quite clear that war, the threat of war and conflicts between countries on average all have a negative influence on the economy. An economy is influenced by political instability in a region, and this influence is also felt on major stock markets. The Korean situation could have similar effects. North Korea’s nuclear program and its nuclear tests could also have a negative effect on the region.6 Besides that, China has
6
always supported North Korea, and thus could again be harmed if North Korea were to conduct more nuclear tests. Increased hostilities between the two Korea’s could also have a negative impact on stock markets. Caplan (2002) supports the claim made by the South Korea’s minister of finance in 2003, that its economy could be hurt by the image that South Korea is still at the break of war with North Korea, since that causes investors to be more negative. In anticipation of war, investors could avoid investing in South Korean stocks and firms. The probability of a war actually breaking out could indeed also affect stock prices negatively. The incidents that occur between the two Koreas could get investors to think chances of war are increasing and lower prices. Positive events, such as a peace treaty being signed or an improvement in the relation between the two countries, could have a positive effect. They also find that effect could also be stronger for ‘large’ events with more media-coverage, as stock prices reactions then become more direct.
These results lead to an expectation of finding events that take place between North and South Korea that do actually affect the stock markets of other countries too. If these effects prove to be severe, when a nuclear test for example causes stock markets to decline steeply, more effort should be aimed towards finding a solution for the stressed situation between the two Korea’s. If positive events cause investors to believe there will not be a war in the region and the unrest will go away and this causes stocks to go up, that is another stimulant to improve the stringent relationship the two countries have.
4. Dividend policies
decline. For South Korea this would mean that we should see a smoothed dividend policy, which then would confirm these findings of Lintner (1956).
Bradley et al. (1998) examine the link between cash-flow volatility and dividend payout. They argue that the volatility of cash available for dividends is affected by both market-wide and firm-specific factors. They find that higher levels of expected uncertainty are associated with lower payout levels of dividend. When there is a stock-price penalty associated with dividend cuts, managers rationally pay out lower levels of dividends when future cash flows are less certain. Bradley et al. (1998) also find that payout ratios are lower for firms with higher expected cash flow volatility, and that higher levels of uncertainty are associated with lower pay-out ratios.
Chay and Suh (2009) reach similar conclusions. They show that across countries, cash-flow uncertainty, as measured by stock return volatility, has a negative impact on the amount of dividends as well as the probability of paying dividends. Their main finding is that cash-flow uncertainty has a significant impact on the amount and probability of dividends in each of the examined countries. This is because in general, external financing is more costly than internal financing and could be even more so for firms with unpredictable cash flows because these firms may be financially constrained. Thus, firms with high cash-flow uncertainty will be more reliant on internal funds and will pay low dividends. Furthermore, dividends are known to be sticky and a decision to decrease dividends may trigger a severe decline in firm value. The impact of cash-flow uncertainty on dividends is generally stronger than the impact of other potential determinants of payout policy, such as the earned/contributed capital mix, agency conflicts, and investment opportunities. Chay and Suh (2009) also find that the effect of cash-flow uncertainty on dividends is distinct from the effect of a firm’s financial life-cycle stage.
markets dislike economic uncertainty and better long-run growth prospects raise equity prices. They conclude with their finding that dividend yields predict future returns, and that there is a significant negative correlation between price-dividend ratios and consumption volatility.
Dittmar (2008) examines how business conditions affect corporate decisions on how much cash to hold, and whether to distribute any excess cash in the form of stock repurchases or dividends. She discusses that the dramatic increase in corporate cash holdings between 1980 and the present has been driven mainly by an increase in the risk of companies. She argues that companies have become riskier partially due to macro-economic factors and that because of the possibility of encountering tough economic times, many companies hold large amounts of cash. Dittmar (2008) argues that reserves of cash are especially valuable during periods of financial trouble because they provide a buffer against shortfalls in operating profits. Such buffers can be used to avoid financial distress or provide funding for valuable projects that might otherwise have to be put off. Later, when industries stabilize, corporate cash requirements can fall and one option is to pay out excess cash by initiating or increasing cash dividends. She concludes that in times of uncertainty, firms may be better of repurchasing stock with their excess cash, because that is a smaller corporate ‘commitment’ than paying dividend.
Opler et al. (1999) examine the determinants and implications of holdings of cash and marketable securities by publicly traded U.S. firms in the 1971-1994 period. They find that firms with strong growth opportunities and riskier cash flows hold relatively high ratios of cash to total non-cash assets. Firms that have the greatest access to the capital markets, such as large firms and those with high credit ratings, tend to hold lower ratios of cash to total non-cash assets. Opler et al. (1999) find little evidence that excess non-cash has a large short-run impact on payouts to shareholders. They state that the main reason that firms experience large changes in excess cash is the occurrence of operating losses.
(2008) do find, however, that outside of the US there is little evidence of a systematic positive relation between relative prices of dividend paying and non-paying firms and the propensity to pay dividends. They also do not find any link between the propensity to pay dividends and share repurchases.
Nacuer et al. (2006) study the dividend policy of 48 firms listed on the Tunisian Stock Exchange during the period 1996–2002. Their study outlines the main determinants that may drive the dividend policy of Tunisian quoted firms. The results clearly demonstrate that Tunisian firms rely on both current earnings and past dividends to fix their dividend payment. However, the study shows that dividends tend to be more sensitive to current earnings than prior dividends. Also, profitable firms with more stable earnings can afford larger free cash flows and thus pay larger dividends. Furthermore, firms distribute larger dividends whenever they are growing fast.
Bhattacharya (1979) assumes that outside investors have imperfect information about a firms’ profitability and that cash dividends are taxed at a higher rate than capital gains. He shows that under these conditions, such dividends function as a signal of expected cash flows. Bhattacharya (1979) states that the signalling cost structure he develops is not only realistic, dividends are linked only to expected cash flows, but also the only simple structure consistent with the assumption of an exogenously costly dividend-signalling equilibrium.
DeAngelo et al. (1992) examine corporate dividend policy. They find that dividend policy has information content in that knowledge that a firm has reduced dividends improves the ability of current earnings to predict future earnings. Their evidence indicates that current income is a critical determinant of dividend changes, and that managers’ reluctance to reduce dividends should lead them to do so only when earnings are especially poor. The former implication is supported by the finding that a loss is essentially a necessary condition for dividend reductions by firms with established earnings and dividend records. The latter implication is supported by the findings that not all firms with losses reduce dividends and those that do have deeper and more persistent earnings problems.
down into the value from assets in place and the value from future growth opportunities or growth options. As a growth firm becomes mature it has fewer options to grow, and assets in place play a bigger role in determining its value. This leads to a decline in systematic risk. Thus, according to Grullon et al. (2002), mature firms generate large free cash flows. Abstracting from agency conflicts, older firms are likely to pay out these cash flows in the form of dividend or stock repurchases. They refer to this as the maturity hypothesis.
financial variables affecting dividend policies at firm level also need to be examined. It is expected that events that increase chances of a speedy positive solution of the Korean conflict will cause more cash to be distributed again by firms and draw more investors to South Korea.
Up to now, firms that keep cash aboard and decrease dividend payments are expected to have poor financial performance or less future cash flow, which both decrease stock value and make a firm less attractive to invest in. Lintner (1956) also assumes that firms keep dividend payments at the level of the previous payment. This investigation aims at showing that it is not to be automatically assumed that firms that pay out less cash are by definition financially distressed. If it can be proven that South Korean firms keep cash aboard or cut dividends because of an increase in uncertainty following a period of negative sentiment due to the Korean conflict, investors have a reason to look beyond that dividend cut and give a firm a thorough financial screening. Removing the prejudice on a cut of dividend payments could make international investment in South Korea become more likely again.
Table I
Foreign Direct Investment
The table presents the foreign direct investment (FDI) overview for South Korea, for East Asia and the World. FDI Flows is the value of all investments in the home country made directly by residents and companies of other countries as a percentage of gross fixed capital formation. Since direct foreign investment excludes investment through the purchase of shares, FDI Stocks is the investment through the purchase of shares in the home country made directly by residents and companies of other countries as a percentage of gross domestic product.
1990 2000 2007 2008
FDI Flows ― as a percentage of gross fixed capital formation South Korea Inward 2,1 1,8 0,9 2,8 Outward 2,1 3,0 5,2 4,7 East Asia Inward 9,8 8,6 8,2 8,4 Outward 5,9 5,4 6,1 6,1 World Inward 8,2 13,5 16,6 12,8 Outward 8,3 13,0 18,0 14,6
FDI Stocks ― as a percentage of gross domestic product South Korea Inward 2,0 7,1 11,4 9,8 Outward 0,9 5,0 7,1 10,3 East Asia Inward 25,9 31,8 33,3 23,1 Outward 5,4 23,0 26,6 20,3 World Inward 9,1 18,1 29,1 25,0 Outward 8,5 19,3 29,8 27,3
5. Sentiment
Brown and Cliff (2005) show that their direct survey measure of investor sentiment predicts market returns over the next 1 to 3 years and that this measure has the ability to explain deviations from intrinsic value as measured by other researchers’ models of stock prices. They propose two ways to interpret these findings. The more conservative interpretation is that they have identified some new factor related to asset valuation. The bolder interpretation that Brown and Cliff (2005) give is that they actually use an accurate measure of investor sentiment and this measure is related to the level of stock prices. This finding has several important implications. First, the results support the important yet controversial behavioural theories that predict the irrational sentiments of investors do in fact affect asset price levels. Second, this suggests asset pricing models should consider the role of investor sentiment. Third, market regulators and government officials should be concerned about the potential for market bubbles or ‘irrational exuberance’ if a sudden change in sentiment translates into a negative wealth shock that depresses economic activity.
Barberis et al. (1998) examine under reaction of stock prices to news such as earnings announcements, and overreaction of stock prices to a series of good or bad news. Their overreaction evidence shows that over longer horizons of perhaps 3 to 5 years, security prices overreact to consistent patterns of news pointing in the same direction. Moreover, over similar horizons, some measures of stock valuation, such as the dividend yield, have predictive power for returns in a similar direction: a low dividend yield or high past return tend to predict a low subsequent return (Campbell and Shiller, 1988). Both findings mentioned by Barberis et al. (1998) could be of influence for South Korea. Consistent patterns of news are noticeable in the Korean conflict; when North Korea conducts a series of tests to improve its military strength, such a process could bring forth a string of bad news. This could then lead to an overreaction in the security prices, further weakening the economic situation of South Korea.
activity is substantially amplified as the level of psychosocial effects increases. According to Drakos (2009) it provides further empirical support for the sentiment effect.
Kling and Gao (2008) measure investors’ sentiment in China. They uncover that share prices and investor sentiment do not have a long-run relation; however, in the short-run, the mood of investors follows a positive-feedback process. Hence, institutional investors are optimistic when previous market returns were positive. Contrarily, negative returns trigger a decline in sentiment, which reacts more sensitively to negative than positive returns. Kling and Gao (2008) also show that previous market returns influence investors’ sentiment, but not vice versa. They find that sentiment does not predict future stock returns.
Lee et al. (2002) test the impact of noise trader risk on both the formation of conditional volatility and expected return. They find that shifts in sentiment are negatively correlated with the market volatility; that is, volatility increases (decreases) when investors become more bearish (bullish). The significance of sentiment on conditional volatility implies that conventional measures of temporal variation in risk omit an important factor. Besides this, Lee et al. (2002) find that that higher (lower) excess returns are associated with a decrease (increase) in conditional volatility resulting from larger bullish (bearish) shifts in sentiment for both small as well as large capitalization stocks. According to them, the permanent effect of noise trading on expected return is through its impact on the market’s formation of risk. They conclude that sentiment is not an individual investor phenomenon that affects only small capitalization stocks.
sentiment and psychosocial effect can have a serious effect on stock market returns. For the Korean case, the effect that sentiment has is examined.
On top of this, Opler et al. (1999) find that firms that have the greatest access to the capital markets tend to hold lower ratios of cash to total non-cash assets. When resources are scarce, this ratio will be higher. If investors really avoid investing money in South Korean firms due to the conflict, like Kim Jin Pyo said, firms will be more inclined to hold higher cash-reserves and consequently pay out less to stockholders.
II. DATA & METHODOLOGY
1. Data
Table II
The sample of countries
The table presents the ten countries that are included in the sample, the index under investigation and the base date that is used for each index. The end-date for each index is August 1st 2009.
Country Index Base date
Australia S&P/ASX 100 May 29, 1992
China Shanghai SE Composite Index January 2, 1991
Germany DAX 30 January 1, 1979
Japan NIKKEI 225 January 1, 1979
Russia RTS Index September 1, 1995
South Korea Korea Composite Stock Price Index (KOSPI) 200 January 3, 1990 Taiwan Taiwan Capitalization Weighted Stock Index January 1, 1979
Thailand Bangkok SET Index January 1, 1979
United Kingdom FTSE 100 January 1, 1984
United States NASDAQ 100 January 3, 1983
World MSCI World Price Index January 1, 1980
World MSCI All Countries World Price Index February 2, 1988
The second part of the study examines the amount of dividend that is paid in South Korea between 1979 and 2009. Major firms listed on the KOSPI 200 are selected, specifically all 50 firms listed on the South Korean KOSPI 50. The fifty largest firms, measured by market capitalization, are listed on this index. A list of the firms under investigation is provided in Table III. From these firms, financial data is gathered. All financial data is gathered from DataStream, and therefore the amount of years collected is limited to what DataStream provides. For every company data the following data is collected: the amount of cash; total cash dividends paid; enterprise value; total assets; earnings before interest but after taxes; retained earnings; company beta; net sales or revenues; total debt as a percentage of common equity; market to book value excluding intangibles. The year in which the company was founded is found either from Businessweek7 or from the company’s own website, also found on DataStream.
Seven companies listed on KOSPI 50 are excluded from the sample, due to data limitations. From the remaining 43 firms, a total of 458 firm years with full data is collected. Of these 458 firm years, there are only 72 observations in which a firm does not pay dividend. This is in contrast with what Fama and French (2001) find; they conclude that the proportion of firms paying cash dividends has fallen steeply to about one-fifth in 1999. They also find a lower incidence of dividend payers among large firms; this is not in line with what we show for South Korea.
7
Table III
Companies listed on KOSPI 50
Companies listed on KOSPI 50 as of January 30, 2010. A total of seven companies marked with an * are excluded from the sample used, due to data limitations. The total sample under investigation thus consists of 43 firms.
Company name
1 Doosan Heavy Industries & Construction Company Limited 2 Doosan Infracore Company Limited
3 Daewoo Shipbuilding & Marine Engineering Company Limited 4 Daewoo Engineering and Construction Company Limited 5 Daewoo International Corporation
6 Daewoo Securities Company Limited 7 GS Holdings Corp.
8 GS Engineering & Construction Company Limited 9 * Hana Financial Group
10 Hyundai Heavy Industries Company Limited 11 Hyundai Steel Company
12 Hynix Semiconductor Incorporated 13 Hyundai Development Company Limited
14 Hyundai Engineering & Construction Company Limited 15 Hyundai Merchant Marine Co., Ltd.
16 Hyundai Motor Company Limited 17 * Industrial Bank of Korea 18 Korean Air Lines Company Limited 19 * KB Financial Group Incorporation 20 KCC Corporation
21 Korea Electric Power Corporation 22 KT Corporation
23 KT & G Corporation 24 Kangwon Land Inc. 25 Kia Motors Corporation 26 Korea Gas Corporation 27 * Korea Exchange Bank 28 LG Corporation
29 LG Display Company Limited 30 LG Chem Limited
31 LG Electronics Incorporated 32 Lotte Shopping Company Limited 33 Mobis
34 NHN Corp
35 OCI Company Limited
36 POSCO
37 S-Oil Corporation 38 Samsung Card Co Limited 39 SK Holdings Company Limited 40 * SK Energy Co., Limited 41 SK Networks Company Limited 42 SK Telecom Company Limited 43 Samsung C & T Corporation
44 Samsung Electronics Company Limited 45 Samsung Fire & Marine Insurance
46 Samsung Heavy Industries Company Limited 47 Samsung Securities Company Limited 48 * Shinhan Financial Group Company Limited 49 Shinsegae Company Limited
2. Methodology
Following the procedure of Brooks et al. (2005), for each stock index in Table II, daily closing prices are collected from DataStream for the period between 1979 and 2009, or from the base year available in DataStream. The same is done for both world indices. Continuously compounded returns are then computed from these daily closing prices for each of the ten individual indices and the two world indices. The continuously compounded return for index i is the value of Rit that satisfies,
=
=
=
=
− − − − 1ln
it it itP
P
R
(1)where Pit is the closing price of index i at day t.
2.1. Impact of events
For each of the ten selected countries a dummy variable augmented market model of the following form is estimated,
it kt ki t WOR i i it
R
D
R
kε
γ
β
α
+
+
+
+
+
+
+
+
+
+
+
+
=
=
=
=
∑∑∑∑ = = = = 20 1 . (2)where Rit is the return on country index i, RWOR,t is the return on the world index, Dkt are the
20 dummy variables that refer to the 20 events listed in Table VI, (αi, βi, γi ) are parameters to
be estimated and εit is a random shock assumed to be distributed IN(0, σi2 ) (Brooks et al.,
Table IV
Distribution of the error terms from the OLS model
The table presents the summary statistics of the error terms using the OLS model from equation (2) for each individual country. The number of observations, mean, median, maximum, minimum, standard deviation, skewness, kurtosis, and the Jarque-Bera statistic are presented. a indicates that the Jarque-Bera statistic is larger then 5,99 and consequently non-normality is assumed.
Austr-alia China
Ger-many Japan Russia
South Korea Tai-wan Thai-land UK United States Obser-vations 4480 4847 7719 7719 3630 6934 7719 7719 6674 5107
Mean 1,55E-20 8,4E-19
-5,12E-19 4,81E-20 -7,46E-20 -1,68E-19
-8,99E-20 4,16E-19 1,51E-19 2,76E-20 Median 0,00014 -0,00036 0,00010 0,00013 0,00026 -0,00007 -0,00017 -0,00009 0,00000 0,00000 Maximum 0,0515 0,7193 0,0641 0,1117 0,1536 0,1453 0,1959 0,1061 0,0511 0,1468 Minimum -0,0713 -0,1807 -0,1329 -0,1063 -0,2191 -0,1246 -0,1974 -0,1603 -0,0606 -0,0820 Standard deviation 0,0090 0,0254 0,0108 0,0118 0,0264 0,0188 0,0177 0,0147 0,0084 0,0134 Skewness -0,505 5,620 -0,475 -0,120 -0,327 0,085 0,051 -0,107 -0,133 0,254 Kurtosis 9,54 149,21 9,42 9,58 9,88 7,63 11,93 11,99 6,44 9,76 Jarque-Bera 8170 a a 4343116 13530 a 13953 a 7215 a 4575 a 25625 a 26025 a 3301 a 12797 a
Table V
Heteroskedasticity and Autocorrelation test
The table presents the summary statistics of the error terms using both the OLS model and the GARCH (1,1) model from regression (2), for each country individually, with each event spanning one trading day. The ARCH LM heteroskedasticity test statistic, and the Ljung-Box Q-statistics are calculated using 2 and 10 lags. * indicates 10% significance level, ** indicates 5% significance level, and *** indicates 1% significance level. The table also presents the regression statistics R-Squared, Adjusted R-Squared, and the Durbin-Watson statistic.
ARCH-LM Heteroskedasticity test
Australia China Germany Japan Russia South
Korea Taiwan Thailand
United Kingdom United States 2 lags OLS GARCH 295,68*** 7,51** 9,29*** 0,24 371,27*** 0,09 841,72*** 9,99*** 250,38*** 0,20 198,35*** 2,42 1693,35*** 20,25*** 753,51*** 0,26 705,59*** 26,47*** 569,35*** 6,05** 10 lags OLS GARCH 605,35*** 16,12* 24,20*** 1,24 566,71*** 2,88 1177,08*** 12,58 418,01*** 5,57 536,54*** 11,68 1830,11*** 76,19*** 926,82*** 0,93 900,28*** 34,12*** 1021,59*** 13,69
Ljung-Box Autocorrelation test
Australia China Germany Japan Russia South
Korea Taiwan Thailand
the GARCH (1,1) model, the statistics significantly improve. For half the countries under investigation the problems of heteroskedasticity and autocorrelation even completely disappear. For the other countries, a significant improvement is also noticeable, with both a reduction in the statistics and in significance levels. This leads to the conclusion that the choice for the GARCH (1,1) model is justified here. Student’s t-distribution is used instead of the normal distribution for this model, because it is able to provide a more robust interpretation of the t-statistics in the presence of leptokurtic data (de Jong et al. 1992).
To expand the analysis of the events, equation (2) is also estimated with each dummy comprising three trading days. According to MacKinlay (1997), even if the event being considered is an announcement on a given date, it is typical to set the event window length to be larger than one. This facilitates the use of abnormal returns around the event day in the analysis. Information coming out of North Korea is scarce; news from within the country also finds its way out only slowly. To account for the possibility that this information becomes known only gradually in the public place and on the stock markets, we also perform this additional test of the impact of each event. To do so, we include in the event period the two following trading days after the event date in Table VI. All data used here are as previously outlined.
2.2. Impact of sentiment
Table VI
Key events in the Korean relationship
Events are numbered under ‘event’; event’s dates of occurrence are presented under ‘Date’; ‘Event description’ describes the event. A ‘+’ in the fourth column indicates an expected increase in stock markets; a ‘-’ an expected decrease. Events 1, 2, 3, 4, 5, 8, 9, 10, 11, 13 and 16 are key-events.
Event Date 8 Event description Expected stock
market impact 1 Late 1979 North Korea starts to build a 5-megawatt nuclear reactor at
Yongbyon, aided by years of Soviet nuclear help -
2 December
12, 1985 Pyongyang signs up to Nuclear Non-Proliferation Treaty +
3 March 15, 1994
North Korea quits the International Atomic Energy Agency, 20 years
after it first became a member state -
4 October 21, 1994
US brokers an agreement to freeze plutonium production in North
Korea, and halt all work at the Yongbyon facility in exchange for aid +
5 August 31,
1998 North Korea tests a Taepadong long-range missile -
6 January 29, 2002
Most international diplomacy with North Korea ceases after President
George Bush lists North Korea in the “Axis of Evil” -
7 December 28 (30), 2002
North Korea restarts its nuclear facilities at Yongbyon and expels
officials from the International Atomic Energy Agency -
8 August 27-29, 2003
Six-Party Talks formed with China, Japan, North Korea, Russia and
the United States +
9 February 10, 2005
North Korea admits publicly for the first time that it has nuclear
weapons -
10 September
19, 2005 North Korea agrees to rejoin the Nuclear Non-Proliferation Treaty +
11 July 4, 2006 Multiple missile tests -
12 October 9,
2006 North Korea conducts first successful underground nuclear test -
13 February 13, 2007
North Korea agrees to start shutting its reactor and allow UN nuclear
inspectors back into the country in exchange for aid +
14 May 17, 2007
A pair of passenger trains crossed the border between North and
South Korea for the first time in more than 50 years 9 + 15 October 11
(13), 2008
The US says it will take North Korea off its state sponsors of
terrorism list, following verbal agreement on dismantlement +
16 April 5 (6),
2009 North Korea launches a multistage rocket -
17 April 14, 2009
North Korea's foreign ministry says the country will quit Six-Party
Nuclear Talks, and restart Yongbyon reactor -
18 May 25 and 26, 2009
North Korea conducts second successful underground nuclear test and
test fires two short-range missiles -
19 June 18, 2009
Russia and China - the country's traditional allies - call for North
Korea to return to the negotiating table +
20 July 2, 2009 North Korea test fires two short-range missiles off its east coast -
8
Global Security.org
9
a positive event that happens during a negative period and does not start a positive one; event 20, where North Korea test fires two missiles off its coast, does not bring peace on the peninsula closer. An event that caused a negative sentiment is, for example, the start of the build of a nuclear reactor in 1979. The signing of the Nuclear Non-Proliferation Treaty by North Korea in 1985 causes positive sentiment, because it reduced the threat of war and brings peace on the peninsula closer. Three of the events in Table VI take place on a day in the weekend, when stock markets are closed. In order to calculate their effect on the stock indices, their date is altered to the nearest following weekday on which the markets are open. This is the case for events number seven, fifteen and sixteen, for which the altered date is added in brackets in Table VI. Because the Bangkok SET Index was closed on December 30, 2002, the new date for event number seven, January 2, 2002 is used instead there. Barberis et al. (1998) find that security prices overreact to consistent patterns of news pointing in the same direction. Using the eleven key events from Table VI, the timeline in Table VII is created, in similar fashion as the one that Schneider and Troeger (2004) use, to be able to examine if this is also the case for South Korea. From the thirty years under investigation, seventeen years have been labelled positive, during a period totalling thirteen years a negative sentiment prevailed.
Here, yt is the index return that is considered to be linearly related to the explanatory dummy
variables (Dt) and an error term εt. The dummy variables Dt represent the sentiment during the
period. The dummy variable for a positive sentiment, D1t, is one if observation t falls in a
positive period and zero otherwise, D2t is a dummy variable for a negative period and is one if
observation t falls in a negative period and zero otherwise. The coefficients δ in equation four represent the size and the direction of the return during a positive or a negative period for the index y (Choudhry, 2000). ht is the conditional variance. It is a one-period ahead estimate for
the variance calculated based on any past information thought relevant. c is the constant, which substitutes the dummy variable for a positive sentiment, D1t. Significance of αj implies
the existence of the ARCH process in the error term. One would expect a significant positive coefficient for the positive period. Correspondingly, the coefficient γ2 represent the size and the direction of volatility, measured by the variance, for the negative period. The volatility for the negative sentiment period is measured compared to the constant, which is a period with positive sentiment. In accordance with the examined literature, it would be expected that a period with negative sentiment brings forth more uncertainty and thus a higher variance (Lee et al., 2002). Also, in line with the previous discussion, one would expect firms to keep cash within the firm when the sentiment is negative and pay out more dividends when it is positive (Dittmar, 2008).
2.3. Impact on dividend policy
payment. Equation (6) is used to estimate the influence of the different factors on the amount of dividend paid in a year.
t t t t t t t t t t t t it t t t t it
D
S
S
S
DE
MTB
S
RE
A
C
S
DIV
A
E
V
AGE
S
DIV
ε
δ
δ
δ
δ
δ
β
δ
δ
δ
δ
δ
δ
δ
+
+
+
+
+
+
+
+
−
−
−
−
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
=
=
=
=
− − − − − − − − − − − − 1 12 1 1 11 10 9 8 7 6 1 5 4 3 2 1)
/
)
((
)
(
)
(
)
/
(
)
/
(
)
/
(
)
/
(
)
(log
)
/
(
(6)Here, DIVt is the dependent variable, the amount of dividend paid by firm i in year t, that is
considered to be linearly related to the explanatory variables and an error term εt. AGE, Vt , At,
St, Ct, β, REt and Et are the firm’s age in years, its market value in year t, its total assets, its
sales, cash, company beta, retained earnings and its earnings before interest, respectively. DIVit-1 is the amount of dividend firm i paid in year t-1. MTBt is the market to book value of
the firm, without intangibles, as acquired from DataStream. DEt is the total debt the company
has, as a percentage of common equity.
((
S
t−
−
−
−
S
t−−−−1)
/
S
t−−−−1)
is the net sales growth a firm has in year t. The dummy variable, D1t, takes on the value of one if observation t falls in anegative period and zero otherwise, thus when sentiment is positive. The explanatory variables are AGE, measured by the number of years the firm is old, the logarithm of Vt, the
firm’s market value in year t, Et / At, the firm’s profitability (Denis and Osobov, 2008). The
amount of dividends paid is scaled by the firms’ sales, as are the retained earnings and the amount of cash the firm has. εt is an error term. The coefficients δ in equation six represent
the influence of each factor, δ1 is the intercept (Brooks, 2008). The equation is tested using the Ordinary Least Squares method. The tables report OLS parameter estimates and standard deviations in parentheses.
Table VII Timeline of sentiment
Event date is presented as a calendar date; the number in brackets represents the corresponding event from Table III. A ‘+’ indicates a positive sentiment, i.e. an expected increase in dividend payments; a ‘-’ a negative sentiment and an expected decrease in dividend payments.
(1) (3) (5) (9) (11) (16) 1979 3/15/1994 8/31/1998 2/10/2005 7/4/2006 4/05/2009
--- +++++++ --- +++++++ --- +++++++ --- +++++++ --- +++++++ ---
Since dividend data is predominantly available annually, issues arise when the sentiment changes within that period of one year preceding the ex-dividend date. As a guideline, to prevent such a change of sentiment from influencing the results, the period between the ex-dividend date and the previous sentiment change, as seen in Table VII, can not be longer than one month. If this period is longer than one month, the dividend payment falls in the category of the new sentiment. A period of one month is chosen, because Gelper and Croux (2010) report that the European Economic Sentiment Indicator (ESI), a survey-based indicator that aims to get insights into the beliefs of economic agents, both from the demand and supply sides of the economy, is also published monthly by the European Commission. A period of one month thus is appropriate and sufficient to let the sentiment change in. So, for example, if a stock goes ex-dividend annually on January 1st, and something happens that changes the sentiment from negative to positive on November 2nd, then the payment on the first of January is considered to reflect that positive sentiment. This is because Dittmar (2008) argues that when industries stabilize, corporate cash requirements can fall and one option is to pay out excess cash by initiating or increasing cash dividends, and vice versa. The timeline in Table VII is used as a guideline for the sentiment prevailing at the ex-dividend date.
2.4. Robustness test
2.4.1. Korean crisis
between August 27, 1997 and April 2, 1999, when the index reached a value 75,38 for the first time after late 1997.
2.4.2. Cash flow influence
In order to be able to determine the influence of the amount of cash a company has or is able to acquire to pay its dividends, equation seven is estimated.
t t t t t t it t it
S
DIV
S
C
A
DE
D
DIV
/
)
=
=
=
=
δ
1+
+
+
+
δ
2(
−−−−1/
)
+
+
+
+
δ
3(
/
)
+
+
+
+
δ
4(
)
+
+
+
+
δ
5 1+
+
+
+
ε
(
(7)Here, DIVt is the dependent variable, the amount of dividend paid by firm i in year t, that is
considered to be linearly related to the explanatory variables and an error term εt. At, St, and
Ct, are the firm’s total assets, its sales and cash, respectively. DIVit-1 is the amount of
dividend firm i paid in year t-1. DEt is the total debt the company has, as a percentage of
common equity. The dummy variable, D1t, takes on the value of one if observation t falls in a
negative period and zero otherwise, thus when sentiment is positive. According to Gatchev et al. (2008), firms on average fund their dividends primarily by equity issues. They find that to pay 1 dollar of dividends, the average firm raises more than 1 dollar of external funds and uses the extra amount to build-up its cash balances and to distribute additional funds to its shareholders by repurchasing shares. It is thus interesting to see how the debt-equity structure of a firm influences dividend payments, next to how the normal availability of cash is of influence, as Dittmar (2008) finds that cash is of influence on dividend policy when uncertainty increases.
2.4.3 Outliers
III. RESULTS
1. Impact of events
The results of the GARCH (1,1) estimation of equation (2) are reported in Tables VIII and IX. Specifically, the results where the event date comprises one day are reported in Table VIII, while the results for the estimation where the event window consists of three trading days are reported in Table IX. The tables report GARCH (1,1) parameter estimates and standard errors in parentheses. γ1i corresponds to event one in Table V, γ2i to event two etcetera. The most important result Table VIII shows is that when the event window only consists of one trading day, from the twenty events under investigation, not one has a significant impact on the stock index of any of the ten countries under investigation. After that, what stands out is the impact that the world index has. Except for China, all countries’ stock indices have a highly significant positive relation with the world index, all at one percent significance. Besides this, measured over all countries together, almost half of the events have the impact that was expected ex ante. When we look at the individual countries, Taiwan rates highest; eleven out of nineteen events have the sign that was predicted beforehand. After that, in China, ten out of eighteen events show the predicted sign, followed by Thailand and the United States, where the score is ten out of nineteen. South Korea and Australia both show half the predicted signs. In the four remaining countries, Germany, Japan, Russia and the United Kingdom, less than half of the events have the anticipated sign. Japan and the United Kingdom seem to be influenced the least by South Korean events; they only react to seven and eight out of nineteen events in the predicted manner, respectively. This is not what we expected to find. Sultan (1995) finds that pessimistic news’ influence is greater than the influence of positive news, and Sweeney and Zhang (1999) find that nuclear tests cause important economic damage to the conducting country’s neighbours. Combining these findings would lead us to expect a significant negative response to events number twelve and eighteen, where North Korea conducts a nuclear test. We find no such response, however, so we can not confirm the findings of Sultan (1995) and Sweeney and Zhang (1999). The impact of the world index is much stronger than that of the individual events.
Table VIII
Event analysis ― one trading day
The table presents regression results from the GARCH (1,1) estimation of equation (2)with Student’s t-distribution, with each event encompassing one trading day. The dependent variable is the return from the different indices in each country. The independent variables are βi , which is the world index, and the coefficients γ1i - γ20i, which correspond to the
twenty events listed in Table V. αi is the constant. The expected impact of each event ex ante is also presented. Standard errors are reported in brackets below the coefficients. *
indicates 10% significance level, ** indicates 5% significance level, and *** indicates 1% significance level. The table also presents the regression statistics Squared, Adjusted R-Squared, and the Durbin-Watson statistic. ‘South Korea ― controlled’ presents the results for the same regression, but with data after winsorizing. NA means not available.
Expected
impact Australia China Germany Japan Russia
South Korea
South Korea ― controlled
one. This facilitates the use of abnormal returns around the event day in the analysis. The following tests account for this. The last factor is the fact that there can be outliers in the data. However, after controlling for them using winsorizing, the results for South Korea do not change. We do see a large reduction in standard deviations for South Korea, while the R-squared statistic remains roughly the same. Outliers could thus be an issue in our data.
Table IX extends our investigation of the twenty different events under investigation by expanding the event window. Instead of just one day, the event date, Table IX shows the GARCH (1,1) estimation of equation (2) when the event window consists of the event date and the following two trading days. Again, what that stands out is the impact that the world index has on the ten individual indices. Except for China, all countries’ stock indices yet again have a highly significant positive relation with the world index, all at one percent significance. The number of events that have the expected signs has increased by more than ten percent; now 102 instead of 90 out of 183 events across all countries show the predicted signs. This could point towards a better fit of the approach that uses an event window of three days, as MacKinlay (1997) already argued. There are three countries in Table IX in which the index stands out when it comes to its sensitivity to the events with their event window of three trading days. Where South Korea’s KOSPI 200 had nine out of eighteen signs as predicted in Table VIII, here that number has increased to fifteen. Only events number five, nine and twenty do not have the impact that was expected beforehand. Events five and twenty were also already positive in Table VIII, but event number nine, where North Korea admitted publicly for the first time that it had nuclear weapons, did have the predicted negative sign when the event window was just one day. That leaves just two unexpected returns for South Korea. The second country where the index does show extra sensitivity to the events when we use the larger event window is Australia. Here, thirteen out of eighteen events now show the predicted return, where before this was only nine. Thirdly, Japan proves to be also influenced more now; twelve instead of seven events now show the predicted sign. Taiwan now only has nine expected signs instead of eleven. The other six countries do not show major changes.
Table IX
Event analysis ― three trading days
The table presents regression results from the GARCH (1,1) estimation of equation (2) with Student’s t-distribution, with each event encompassing three trading days. The dependent variable is the return from the different indices in each country. The independent variables are βi , which is the world index, and the coefficients γ1i - γ20i, which correspond to the
twenty events listed in Table V. αi is the constant. The expected impact of each event ex ante is also presented. Standard errors are reported in brackets below the coefficients. *
indicates 10% significance level, ** indicates 5% significance level, and *** indicates 1% significance level. The table also presents regression statistics Squared, Adjusted R-Squared, and the Durbin-Watson statistic. ‘South Korea ― controlled’ presents the results for the same regression, but with data after winsorizing. NA means not available.
Expected
impact Australia China Germany Japan Russia
South Korea
South Korea ― controlled
