The effect of marketshare on the size of the ripple effect:
a case study on the global alcoholic drinks market between 2009 and 2014
Bachelor Thesis by Dennis Groot (10165142)
Faculty of Economics & Business Economics & Finance
Supervised by Mr M.A. Dijkstra MSc
2
Statement of Originality
I, Dennis Groot, declare to take full responsibility for the content of this thesis. I declare that the content is original and that no sources other than mentioned in the text are used for this thesis.
3
Abstract
This paper studies whether or not the marketshares of companies have an influence on the size of a ripple effect. The size of the ripple effects are calculated using cumulative abnormal returns, which have two different time windows. The first with 5 trading days before announcement date and 5 trading days after announcement date. The second with 3 trading days before announcement date and 3 trading days after announcement date . The announcement dates are found by selecting profit announcements companies which are
indicated as below investors expectations. The companies which are selected all operate in the alcoholic drinks market, have a marketshare larger than 0.75% in 2014 and are all publicly traded between 2009 and 2014. 13 companies are selected. 18 announcement dates are used which are announced by 10 different companies. Previous studies indicate that the contagion effect is larger than the competitive effect, so the ripple effect is expected to have a negative value. This is the case because the announcements also have a negative value effect. Also the marketshares are expected to influence the size of the ripple effect. The regression done in this paper, with a sample size of 211, indicates that the ripple effect is indeed of a negative value and the marketshares of the companies have a significant influence on the size of the ripple effect.
4
Table of contents
1. Introduction ... 5
2. Literature review ... 6
2.1 Ripple effect ... 6
2.2 Theory contagion effect ... 6
2.3 Theory competitive effect ... 7
2.4 Theory market structure hypothesis ... 9
2.5 Empirical evidence on contagion effect, competitive effect and market structure hypothesis ... 9
2.6 Conclusion on empirical evidence... 13
3. Methodology ... 13
3.1 variables ... 13
4. Data ... 16
4.1 Finding companies and announcements ... 16
4.2. Data and databases ... 17
5. Results ... 19
6. Conclusion ... 21
References ... 23
5
1. Introduction
Ripple effects could influence the stock price of a company without an underlining reason of the company itself. Companies’ stock prices are influenced by the performances of its competitors. These performances are shown for example in the profit announcements of these companies. But are all companies’ stock prices in an industry equally influenced by these profit announcements or do certain market characteristics influence the size of the ripple effect? The aim of this paper is to research whether or not the marketshares of the companies in an industry have an influence on the size of the ripple effect of a profit announcement. This is done by answering the research question of this paper, which is: Does the marketshare of a company, operating in the global alcoholic drinks market, influence the size of the ripple effect between 2009 and 2014?
To answer the research question, this paper will look at the cumulative abnormal return, CAR, of 13 companies which operate in the global alcoholic drinks market between 2009 and 2014. The CAR will be calculated by adding the abnormal returns of the stock price returns of 5 trading days before the announcement date and 5 trading days after the
announcement date. This will also be done with a timeframe of 3 trading days before the announcement date and 3 trading days after the announcement date. The announcements are profit related and are indicated as below investors expectations. This can be seen by a negative response of the announcing companies’ stock price directly after the announcing date. 18 announcement are used and the announcements are made by 10 different companies. All the announcements are made between 2009 and 2014. A regression is made using the variables: cumulative abnormal returns, marketshare of companies, total assets, leverage and stock price correlation between the announcing company and its competitors. Dummy variables for the companies which are competitors and for each year the announcement is made are also put in the regression to account for company or yearly factors which aren’t included in the regression. The sample size of this regression is 211.
The research finds that for the alcoholic drinks market between 2009 and 2014 the marketshare of both the announcing company and its competitors have a significant influence on the size of the ripple effect. So does the correlation of stock price returns of the
announcing company and its competitors, the CAR of the announcing company and the assets of the announcing company. The assets of the competitors and the degree of leverage of the announcing company and its competitors aren’t a significant influence on the size of the ripple effect.
6 In the next chapter a literature review is given on what a ripple effect is and what in theory determines the size of a ripple effect. In chapter 3 the method of doing the research is presented. After that the data will be mentioned in chapter 4. In chapter 5 the results of the research will be discussed and in the last chapter a conclusion will be made.
2. Literature review
2.1 Ripple effect
In this paper the term ripple effect will be used as the overall effect of a profit announcement that lowers the stock price of the announcing company on its competitors’ stock price. Cheng and McDonald (1996) state that the term ripple effect consists of two underlining effects. Namely, the contagion effect and the competitive effect. The contagion effect has the effect that the stock price of the competitors will go in the same direction as the announcing company. The competitive effect has the effect that the stock price of the
competitors will go in the opposite direction as the announcing company. Both the effects and their implications to the research done in this paper will be discussed in the next paragraphs.
2.2 Theory contagion effect
Lang and Stulz (1992) state that the contagion effect is the change in the equity value of competitors after an announcement that can’t be linked to the transfer of wealth from the announcing company to its competitors. The effect of the contagion effect is the part of the total effect, that has the same movement direction as the effect of the announcement has on the announcing company. So in case of a bankruptcy announcement, it will be the negative impact on competitors of the announcing company (Ferris, Jayaraman, & Makhija, 1997). The mechanism of the contagion effect is as follow. Each company in an industry is likely to have the same characteristics as other companies in the same industry. So when a company
announces a certain kind of news, there is a high probability that the other companies in the industry has a similar situation. Investors may see this in the same way and change their valuation in the same way as for the announcing company. This may lead to the same stock price movement of the competitors as for the announcing company (Chi, 2009). The impact of the contagion effect will be expected to be more severe for competitors with highly similar characteristics as that of the announcing company. This is the case because competitors with highly similar characteristics are likelier to experience the same results from the market
7 conditions as the announcing company than companies which don’t have highly similar characteristics (Lang & Stulz, 1992).
Lang and Stulz (1992) also state that the announcement could also affect the market value of competitors by changing their dealings with suppliers, customers and regulators. This could for instance be the case when an announcing company can’t meet their liabilities. The suppliers may be reserved with supplying products to other companies in the industry,
because the suppliers may fear that the other companies may also face the same distress as the announcing company does (Lang & Stulz, 1992).
Companies that are dominant in their market are in theory expected to have a more intra-industry effect, because dominant companies are seen by investors as a mayor indicator for the wellbeing of the general industry. Dominant companies usually have a larger
marketshare than other companies in the same market. This means that for this research, a larger marketshare would in theory lead to a larger contagion effect (Yu, Zhang, & Zheng, 2010).
2.3 Theory competitive effect
The competitive effect is the change in the equity value of competitors after an announcement, that can be linked to the transfer of wealth from the announcing company to its competitors. To illustrate the working of the competitive effect look at an industry with imperfect competition and where the announcements are related to an unexpected decrease in demand. This could be the case for instance when the products of the announcing company are suddenly decreased in popularity in comparison to its competitor’s product. When the announcement about a decrease in demand has been made the competitors would gain a positive reaction out of the announcement. This is because the competitors can experience an increase in demand. The competitors would raise their marketshare at the expense of the announcing company (Lang & Stulz, 1992). In industries which aren’t completely competitive companies sell slightly above cost price, which means increased demand will increase profits. These increasing profits will lead to a positive effect for the competitors of an announcing company. However, in a perfectly competitive industry competitors do not gain from extra sales. This is because products are sold against cost price. So only when an industry isn’t a perfectly competitive industry the competition effect will play a role. The degree of
competition also has an influence on the size of the competitive effect, because competitors can profit more from their output when the level of competitiveness goes down. This leads to
8 the notion that the importance of the competitive effect should be linked to the degree of competition (Lang & Stulz, 1992).
Ferris, Jayaraman and Makhija (1997) state that suppliers and consumers may become cautious of the announcing company, if the announcement is bad news for the announcing company. The suppliers may be unwilling to supply on credit and the consumers may be unwilling to pay before delivery, because of the fear that the announcing company may became unable to meet their liabilities. This may lead to opportunities for competitors to expand their business and to negotiate lower prices from their suppliers, because the suppliers and consumers will transfer their business away from the announcing company to the
competitors (Ferris, Jayaraman, & Makhija, 1997).
A drop in production efficiency could result in higher marginal costs for the
announcing company. This could lead to higher prices and lower output. The higher prices and lower output of the announcing company could have the effect that competitors raise their prices. This is because their products would be substitutes for the products of the announcing company and the demand for the products of the competitors will rise (Lang & Stulz, 1992). In case of a bankruptcy announcement the competitive effect could work in two ways. First the non-announcing competitors in the industry are positively affected, because the competitors could increase their marketshare at the cost of the announcing company (Chi, 2009). Secondly, because of a bankruptcy announcement it may be difficult for the bankrupt company to respond to takeovers. The announcing company may require additional
investments to block the takeover. Parts of the company could be taken over or even the whole company. Companies with financial problems are more often a target for takeovers because they are usually cheaper to acquire than financial healthy companies. Usually the aim for the takeover is to implement the whole company or parts of the company into another company, or to restructure the company which is taken over and to resell the company with a profit. The bankruptcy announcing company may have difficulties raising the funds quickly because of the bankruptcy announcement, and management’s attention may be scattered by the bankruptcy process. This may lead to lower activity in the market by the announcing company, which create opportunities for competitors to gain a stronger market position (Lang & Stulz, 1992).
The magnitude of the competitive effect will, in theory just like the contagion effect, be related to the marketshare of the announcing company. This is the case because troubled companies with a large marketshare will give a larger part of the market to gain for the competitors than companies with a small marketshare. So, in theory, the larger the
9 marketshare of the announcing company the larger the competitive effect will be for its
competitors (Yu, Zhang, & Zheng, 2010).
2.4 Theory market structure hypothesis
Cheng and McDonald (1996) construct the market structure hypothesis: for a contestable market a bankruptcy filing of a member company should not lead to any abnormal price performance of the surviving companies in the same market. A contestable market in this paper is a market in which there are almost no barriers to enter. If the market structure is not contestable, there are probably barriers to entry and the industry would have market power. If the market power argument is in play, then the bankruptcy of a company increases the market power of the surviving companies in the market or industry. This results in a stock price appreciation of the surviving companies, because greater market power should lead to a higher market value. For a non-competitive market in which market power exists, the market structure hypothesis states that a bankruptcy announcement of a company should lead to a positive ripple effect for the surviving companies in the same market or industry (Cheng & McDonald, 1996).
The theory of the market structure hypothesis indicates that profit announcements that are below the expectations of investors, will have the outcome that the competitors benefit from the announcement. This is only the case if the alcoholic drinks market isn’t a contestable market.
2.5 Empirical evidence on contagion effect, competitive effect and market structure hypothesis Lang and Stulz (1992) investigate 59 bankruptcy filings within 41 different industries between 1970 and 1989. They use abnormal returns to calculate the size of the ripple effect. Lang and Stulz (1992) find that on average, bankruptcy announcements decrease the value of a value-weighted portfolio of competitors by 1%. This means that the contagion effect is more severe by 1% than the competition effect. They also find that this effect is larger by 3.22% for highly levered industries1. Lang and Stulz (1992) also find that a bankruptcy announcement has a significant positive effect for highly concentrated industries2 with low leverage. The
1
The highly levered industries are industries that have higher leverage than the median of all the tested industries.
2Highly concentrated industries are industries that have higher concentration than the median
10 significant positive effect for highly concentrated industries with low leverage would suggest that in such industries competitors would benefit from the bankruptcy filing company (Lang & Stulz, 1992).
Ferris, Jayaraman and Makhija (1997) studied 274 companies who announced a bankruptcy. The small companies are listed at the NASDAQ and the larger companies are listed at the NYSE/AMEX. Ferris, Jayaraman and Makhija (1997) find that large company bankruptcies generate a contagion effect. On average for every dollar lost on the stock price by the bankrupted company, the competitors had a combined loss of 3.32 dollar on their stock prices. In addition, they find that small company bankruptcies generate a contagion effect among smaller sized competitors. But it appears that the contagion effect doesn’t go from small bankruptcy companies to large competitors. These findings of the difference in size may imply that the assets of smaller companies are not representative of the assets of the larger companies. An explanation that the researchers give is that larger companies are more
diversified, so that the contagion effect of a small company bankruptcy only affects a smaller part of the business of large competitors (Ferris, Jayaraman, & Makhija, 1997).
Cummings, Wei and Xie (2012) and Akhigbe, Martin and Whyte (2005) find that the contagion effect is larger than the competition effect. Cummings, Wei and Xie (2012) study the market value impact of operational risk3 events on non-announcing companies in the U.S. banking and insurance industries. They studied 415 bank events and 158 insurer events in which a operational loss occurred. The data is collected between 1978 and 2010. They find that operational loss announcements have a strong negative spillover4 effect on
non-announcing competitors within the financial industry (Cummins, Wei, & Xie, 2012).
Akhigbe, Martin and Whyte (2005) study the effect of the bankruptcy of WorldCom in 2002 on different stakeholders. They study a sample containing 64 institutional investors, 22 creditors and 96 competitors. These are also the different stakeholders. The finding of this research is that the large competitors5 are negatively affected by the announcements leading to the bankruptcy of WorldCom. Thus this means that the contagion effects appears to dominate the competitive effect (Akhigbe, Martin, & Whyte, 2005).
Yu, Zhang and Zheng (2010) research the contagion effect of corporate scandals on other companies with a different quality level of corporate governance. They study 330
3
Operational risk is categorized between different categories like internal fraud, external fraud, employment practice & workplace safety, damage to physical assets, business disruption & system failure and execution delivery & process management.
4Significant at least at the 5% level. 5
11 company-scandals events between 1997 and 2008. The level of corporate governance is measured by looking at four regional Chinese indices6. Each index is divided into two groups: High index group and Low index group7. Yu, Zhang and Zheng (2010) find that a higher level of corporate governance reduces the stock volatility of a company and with that reducing the contagion effect of a scandal in comparison with a company with a low level of corporate governance. This is the case because a higher level of corporate governance reduces the change of a corporate scandal. So the contagion effect will be lower with companies with a high level of corporate governance. Another finding in the paper is that the contagion effect is larger in industries with a low herfindahl index and that the competition effect is larger in industries with a high herfindahl index. This could be the case because in industries with a low herfindahl index the competitors could be forced to also conduct in a illegal way or they could be outperformed in a highly competitive market (Yu, Zhang, & Zheng, 2010).
Chi (2009) and Chi and Tang (2008) study the contagion and competitive effect of reorganization filings. Both researches investigate reorganization filings in the Taiwan stock market. Chi (2009) looks at 59 announcements in 15 different industries between 1987 and 2006. Chi and Tang (2008) study 66 announcements in 15 different industries between 1990 and 2006. In both the researches they find evidence that the stock price movement of portfolio rivals is negatively correlated with announcements of reorganization. Competitors experience significant negative effects from announcements of reorganization. Chi (2009) find that the competitive effect is stronger than the contagion effect. 52.11% of all the companies experience a negative stock price reaction on the filings. Chi and Tang (2008) find that the competitive effect is also stronger than the contagion effect. In the first day after the announcement the stock price of the competitors declined on average with 1.846%. This indicates that negative competitive effects outweighs positive contagion effects for other companies in the same industry (Chi, 2009), (Chi & Tang, 2008).
Slovin, Sushka and Polonchek (1999) research whether a negative event at one bank generates externalities for the banking industry. They study 62 dividend reduction
announcements and 61 regulatory enforcement8 actions announcements between 1975 and 1992. They find that dividend reductions at money center banks generate industry wide negative effects. Money center banks are banks that process mainly large international and
6
Marketization index, legal index, openness and GPD per capital.
7Based on the median of each index. Above median is high and below median is low.
8Examples of regulatory enforcement actions are: order of constraint on specific managerial
12 domestic financial flows through global networks. The negative effect means that a contagion effect is present, but it is not present when regional banks reduce dividend. Regional banks are banks that service only on a domestically bases. They find that a dividend reduction at regional banks have a positive competitive effect on banks in the same geographic area. These competitive effects are more severe when the markets have a higher four-firm concentration ratio. They also find that regulatory enforcement actions by United States government autorities against money center banks have no effect on its competitors, but enforcement actions against regional banks have positive competitive effect on competitors (Slovin, Sushka, & Polonchek, 1999).
Wu (2002) researches the stock-return behaviour around earnings restatement
announcements. The study uses hand-collected data from between 1977 and 2001 and consists of 1068 earning restatement announcements. All announcing companies are listed in the United States. This study finds that between this timeframe there are no spillover effects on companies with similar company characteristics9. This implies that there is no ripple effect of a declining confidence among investors. A declining confidence is measured by looking at a decrease of the earnings response coefficient (Wu, 2002).
The results of the research done by Cheng and McDonald (1996) suggest that the market structure influences the effect of the announcement effects. The study looks at 7 bankruptcy announcement of airlines and 5 bankruptcy announcements of railroads between 1962 and 1991. They looked at 65 airlines and 73 railroad companies. The railroad market is stated as a non-contestable market and the airline market as the contestable market. The airline market is perceived to be contestable because the assets are mobile and for the railroad market it is not. The level of mobility of assets is used by Cheng and McDonald (1996) to determine whether or not there are barriers to enter. The findings of the positive ripple effect for airlines and the negative ripple effect for railroads are consistent with the market structure hypothesis. These findings conclude that, under the market structure hypothesis, if a market isn’t a contestable market, a rise in market power should lead to a higher valuation of a company (Cheng & McDonald, 1996).
9Characteristics like company size, restated amount of earnings, number of restated quarters
13
2.6 Conclusion on empirical evidence
The overall empirical evidence suggest that the contagion effect will be more severe than the competitive effect. Although some studies find that the competitive effect is more severe than the contagion effect. The empirical evidence also suggest that the market concentration plays a role in the size of the ripple effect and that companies with a larger marketshare generate a larger ripple effect than companies with a smaller marketshare. The market structure hypotheses does indicate that, when the market is a contestable market, the marketshare of the company won’t play a role in the size of the ripple effect. The empirical findings indicate that the announcements made by the alcoholic companies will lead to a negative response to the competitors and the marketshare of the companies will have an influence on the size of the ripple effect.
3. Methodology
The following regression model will be used to answer the research question:
The i stands for the competitor of the announcing company. The j stands for the announcing company and the t stands for the specific date that is linked to the data variable. This model is comparable to the model used by Wu (2002), Yu, Zhang and Zheng (2010) and Chi and Tang (2008).
3.1 variables
The variable CARi,t means the Cumulative Abnormal Return of the competitor. This variable is computed by first using the CAPM described in the paper of Strong (1992) to find the beta of each company. This is done by calculating the returns of the end of the day stock prices and the S&P500 index between 2007 and 2008. Then the rate of the US treasury
14 return of the S&P500. With these returns a regression is done to estimate the beta of all the companies. The regression looks like:
ri,t stands for the return of the company on each date. rf,t stands for the US treasury rate on each date and rs&p500,t stands for the return of the S&P500 on each date.
With each beta the CAPM is used to calculate the expected return on each day
between 2009 and 2014. For the risk free rate the US treasury constant maturity 10 year bond is used. The formula used to compute the expected returns looks like:
To compute the abnormal return, the expected return of stock of each company is subtracted from the actual return. To compute the CAR, the abnormal returns of 5 trading days before the announcement date until 5 trading days after the announcement date are summed. Also, the CAR is calculated using a timeframe of 3 trading days before the announcement date until 3 trading days after the announcement date (Strong, 1992), (Pruitt, Wei, & White, 1988). The average value of the variable CARi,t is expected to be negative, because in the literature review it is found that the contagion effect outweighs the competition effect. This study looks at negative news of the announcing company, so the competitors will also experience a negative effect. This is the case, because as found in the literature review the contagion effect outweighs the competition effect.
The variable ln(MSj,t) stands for the natural logarithm of the marketshare of the announcing company and the variable ln(MSi,t) stands for the natural logarithm of the marketshare of the competitor. Lang and Stulz (1992) predict that the larger the marketshare of the announcing company the larger the ripple effect will be. This is the case, because companies with a large marketshare are seen by investors as a good indicator for the whole sector. When the competitor has a large marketshare the ripple effect on the competitor is expected to be smaller than with a competitor with a smaller marketshare. This is also because the larger company by marketshare are perceived to be a better indicator for the whole market than a smaller company by marketshare (Lang & Stulz, 1992).
15 The variable CARj,t is in the same way calculated as the variable CARi,t. This variable is put in the regression, because it is necessary to know the initial amount that starts the ripple effect. If the cumulative abnormal returns of the announcing company are more negative, then the ripple effect will also be more severely negative. This is the case because Yu, Zhang and Zheng (2010) state that the size of the initial announcement determines the degree of the ripple effect (Yu, Zhang, & Zheng, 2010).
The variables ln(ASSETSj,t) and ln(ASSETSi,t) are the natural logarithm of the total assets of the announcing company and its competitors. Yu, Zhang and Zheng (2010) suggest that companies which have larger total assets as its competitors, are more likely to be seen as a market leader in the industry. Market leaders are expected to be a mayor indicator for the industry in which they operate. This means that an announcing company with large total assets will generate a larger ripple effect than an announcing company with lower total assets. Non-announcing competitors with large total assets compared to other companies in the industry will be expected to experience a lower ripple effect than competitors with low total assets (Yu, Zhang, & Zheng, 2010). But companies in the alcoholic drinks market with total assets which are higher than the average of the industry tend to be more diversified in their operations. This means that not all the assets which the company possess, are used in the alcoholic drinks market. For companies, which are more diversified, the ripple effect is expected to be smaller than for companies which are less diversified. This leads to the prediction that companies with larger assets tend to have a smaller ripple effect.
The variables LEVj,t and LEVi,t stand for the leverage of the announcing company and its competitors. Lang and Stulz (1992) indicate that leverage magnifies the contagion effect but not the competitive effect. This is the case because a drop in the equity value would be more severe percentage wise when the company has a higher leverage than with a lower percentage. It is expected that when a company has a high leverage rate, the ripple effect will be larger on that company than when a company has a small leverage rate (Lang & Stulz, 1992). The contagion effect will be influenced by the level of leverage but the competitive effect won’t. So this means that the higher the level of leverage is, the larger the ripple effect will be for the companies in the alcoholic drinks market.
The variable CORi,t stands for the correlation between the stock returns of the announcing company and its competitor. CORi,t is calculated by first selecting the stock returns of the companies 110 trading days before the announcing date until 10 trading days before the announcing date. These returns are put into stata and the correlation between the return of the announcing company and its competitors are calculated. The time window of 110
16 trading days until 10 trading days before the announcement date is used, because this gives a precise measurement for the correlation between the announcing company and its competitors on the announcing date. A longer time window will give an correlation which isn’t
representative with the correlation around the announcement date and a shorter time window could lead to an imprecise correlation, because of a calculation with fewer data points. In theory, the ripple effect will be largest for the companies which are highly correlated with each other. So with a larger correlation it will be expected that the ripple effect will be higher than when the correlation is smaller (Yu, Zhang, & Zheng, 2010). Lang and Stulz (1992) find that contagion effect is larger for industries in which the stock price returns of the announcing company and its competitors are severely correlated.
The variables ABIi,t, SABi,t, HEIi,t, CARLi,t, CREi,t, MCBi,t, BYBi,t, KHi,t, Di,t, AGHi,t, CBi,t and SMCi,t are all dummy variables for the competitors. These dummy variables are put in the regression to measure fixed effects. Fixed affects are used to control for unobserved variation. The variables are abbreviations of the companies’ name. The variables are in order of the companies’ name in table A1 in the appendix. The sixths company, Tsingtao Brewery Co Ltd, isn’t put in the regression as a dummy variable , because this would lead to the dummy variable trap. This company is chosen to be left out of the regression, because Tsingtao Brewery is always a competitor. Tsingtao Brewery isn’t an announcing company in this paper.
The variables 2010t, 2011t, 2012t, 2013t and 2014t are dummy variables for the year in which the announcement is made. These variables are also in the regression to measure for fixed effects. The dummy variable 2009 isn’t put in the regression, because this would also lead to the dummy variable trap.
4. Data
4.1 Finding companies and announcements
This study focuses on the alcoholic drinks market. This market has the SITC code 112. The focus on the alcoholic drinks market is done because in this market, there are globally over 70 companies active. In this market at least 15 companies are publicly traded and the marketshares can be found in Euromonitor/Passport. Publicly traded companies and the marketshares are necessary to conduct the research. All the companies that have been selected operate in this market, are publicly listed and have an annual marketshare larger than 0.75%
17 in 2014. With these criteria 13 companies are selected. The companies are selected using Euromonitor/Passport for the marketshare criteria and using Datastream for the listed criteria.
This paper looks at profit announcements which are indicated as below investors expectations. This can be seen by a negative response of the announcing companies’ stock price directly after the announcing date. The announcements are found in the database Lexis Nexis. In “search the news” the companies name is searched for. The source type is
“newspapers”. Then under “result groups” the part “subject” is chosen. Under “subject” the word “profits” is chosen. Within “profits” announcements are handpicked which contain the phrase “under/below investors expectations”. All the announcements are between 2009 and 2014.
4.2. Data and databases
Datastream is used to find the stock prices of each company, the US treasury constant maturity 10 year bond and the S&P500 between 2007 and 2014. The total assets and the leverage rate of each company between 2009 and 2014 is also found in Datastream. The marketshares of all the companies are found in Euromonitor/Passport. All the companies, the announcement dates and the values of the variables can be found in the appendix.
Table 1 and table 2 are an overview of the descriptive stats of the variables. Table 1 is with CAR5 and table 2 is with CAR3. CAR5 has the timeframe of -5, +5 trading days around the announcement date and CAR3 has the timeframe of -3, +3 trading days around the announcement date.
Table 1. Descriptive stats of the variables with CAR5 (n=211)
variable mean std. Dev. min max
CARi,t (%) -18.40 9.74 -41.62 5.87
MSj,t (%) 4.32 3.55 0.20 14.50
MSii,t (%) 3.99 3.86 0.20 16.30
CARj,t (%) -19.01 9.82 -40.97 6.07
ASSETSj,t (millions) 382.00 881.00 7.10 2900.00
ASSETSi,t (millions) 462.00 830.00 7.10 2920.00
LEVi,t (%) 30.02 13.64 2.01 57.04
LEVj,t (%) 33.14 9.03 14.01 47.57
18
Table 2. Descriptive stats of the variables with CAR3 (n=211)
variable mean std. Dev. min max
CARi,t (%) -12.33 6.86 -31.40 2.73
MSj,t (%) 4.32 3.55 0.20 14.50
MSii,t (%) 3.99 3.86 0.20 16.30
CARj,t (%) -13.74 6.77 -27.89 3.79
ASSETSj,t (millions) 382.00 881.00 7.10 2900.00
ASSETSi,t (millions) 462.00 830.00 7.10 2920.00
LEVi,t (%) 30.02 13.64 2.01 57.04
LEVj,t (%) 33.14 9.03 14.01 47.57
CORi,t 0.17 0.21 -0.30 0.71
In table 1 and table 2, both the CAR of the announcer and the competitor are on average negative. This would indicate that the contagion effect is more severe than the competitive effect. The standard deviation of both the CAR’s in the two tables are large. This indicates that the size of the different CAR’s are severely different in magnitude. This can also be seen by the different in the minimum value and the maximum value of the CAR’s. In both the tables the average CAR of the announcer and the average CAR of the competitor is slightly different. This could indicate that a few variables, other than the CAR of the announcer, have an impact on the ripple effect. The tables also show that the companies are severely different in the size of the company and the marketshare is also severely different between the
companies between 2009 and 2014. This indicates that there are a few large companies and a few smaller companies.
19
5. Results
Several regressions have been done to determine the effect of each variable. The result are shown in table 3 and 4. Table 3 are the regressions done with CAR5 and table 4 with the regressions of CAR3.
Table 3. Summary of regression with CAR5
CARi,t CARi,t CARi,t CARi,t CARi,t CARi,t CARi,t
ln(MSj,t) 0.67 (0.43) 0.71 (0.43) 2.75*** (0.53) 2.59*** (0.52) 2.35*** (0.57) 2.00*** (0.60) ln(MSi,t) 4.53* (2.49) 4.71* (2.48) 4.28* (2.30) 4.89** (2.34) 4.93** (2.33) 4.94** (2.32) CARj,t 0.44*** (0.08) 0.57*** (0.08) 0.55*** (0.09) 0.55*** (0.08) ln(ASSETSj,t) 1.22*** (0.37) 1.15** (0.37) 1.30*** (0.38) ln(ASSETi,t) -3.49 (3.80) -6.12 (4.25) -5.49 (4.24) LEVi,t 0.16 (0.12) 0.17 (0.12) LEVj,t -0.06 (0.05) -0.08 (0.05) CORi,t 4.38* (2.49) Constant -25.94*** (1.88) -30.20*** (3.21) -30.47*** (3.20) -14.83** (4.02) 21.81 (63.03) 66.86 (70.14) 54.45 (70.10) Company
fixed effects Yes Yes Yes Yes Yes Yes Yes
Year fixed
effects Yes Yes Yes Yes Yes Yes Yes
N 216 211 211 211 211 211 211
R2 0.63 0.62 0.63 0.68 0.70 0.71 0.71
Adjusted R2 0.60 0.59 0.59 0.65 0.67 0.67 0.67
20
Table 4. Summary of regression with CAR3
CARi,t CARi,t CARi,t CARi,t CARi,t CARi,t CARi,t
ln(MSj,t) 0.63* (0.33) 0.70*** (0.34) 2.09*** (0.43) 1.93*** (0.44) 1.94*** (0.48) 1.51*** (0.50) ln(MSi,t) 3.49* (1.95) 3.67* (1.94) 3.34* (1.84) 3.61* (1.91) 3.60* (1.92) 3.62* (1.89) CARj,t 0.48*** (0.10) 0.51*** (0.10) 0.51*** (0.10) 0.50*** (0.10) ln(ASSETSj,t) 0.38 (0.27) 0.39 (0.28) 0.56** (0.28) ln(ASSETi,t) -1.42 (3.11) -2.90 (3.50) -2.14 (3.46) LEVi,t 0.09 (0.10) 0.10 (0.09) LEVj,t 0.00 (0.04) -0.03 (0.04) CORi,t 5.24** (2.03) Constant -18.07*** (1.48) -21.42*** (2.51) -21.69*** (2.49) -9.68*** (3.43) 6.35 (51.64) 30.20 (57.79) 15.19 (57.22) Company
fixed effects Yes Yes Yes Yes Yes Yes Yes
Year fixed
effects Yes Yes Yes Yes Yes Yes Yes
N 216 211 211 211 211 211 211
R2 0.54 0.53 0.54 0.59 0.60 0.60 0.61
Adjusted R2 0.50 0.49 0.50 0.55 0.55 0.55 0.56
(*, ** and *** mean that the coefficient are significant at a 10%, 5% and 1% level)
The regression output in table 3 and 4 shows that the coefficients of variables ln(MSj,t) and ln(MSi,t) are significantly different from 0 on different significant levels in most of the regressions. Only the coefficient of ln(MSj,t) isn’t significant in the first and third regression of CAR5. The coefficients are stronger significantly with the CAR5 regression than with the CAR3 regression. The significant coefficients of the marketshares indicates that the
marketshares of both the announcing company and the competitors are an influence on the size of the ripple effect, even though the significant level is weak sometimes. The finding that the marketshare has an influence on the size of the ripple effect is in line with the empirical findings discussed in the literature review. All the coefficients of the marketshares are also positive. This means that companies with a larger marketshare generate a larger ripple effect. This is also in line with the theory and the empirical findings stated in the literature review.
21 In both the regressions of CAR5 and CAR3 the coefficients of ln(ASSETi,t), LEVi,t and LEVj,t aren’t significant on any level. These findings aren’t in line with what is expected. It is expected that these variables would have an influence on the size of the ripple effect.
ln(ASSETSj,t) is in the regression with CAR5 significant in all the regressions but in the regressions with CAR3 only in the last regression. But it can stated that the assets of the announcing company does influence the size of the ripple effect. The coefficients of CARj,t and CORi,t are all significant in both the regressions with CAR5 and CAR3. That the assets of the announcing company, correlation of the stock prices and the CAR of the announcing company have an influence on the size of the ripple effect is in line with the expectations. The positive coefficients are also in line with the expectations. The contagion effect is predicted to be more severe than the competitive effect and that can be seen by the positive coefficient of CARj,t. It is also predicted that the larger the assets of the announcing company the larger the ripple effect will be. This can be seen by the positive coefficient of ln(ASSETSj,t). The more the return of the stock prices of the announcing company and its competitors are correlated the larger the ripple effect will be. This can also been seen by the positive coefficient of CORi,t.
In all the regression of both CAR5 and CAR3 it is found that there are company fixed effects and yearly fixed effects. This indicates that there are company and yearly factors which influence the size of the ripple effect which aren’t included in the regression. The dummy variables in the regression are used to control for these unobserved variation.
All expectations that are mentioned above are all explained in chapters 2 and 3.
6. Conclusion
The purpose of this paper is to research whether or not the marketshare of the announcing company and its competitors have a significant impact on the size of the ripple effect. To be able to research this a research question is formulated. The research question of this paper is: Does the marketshare of a company, operating in the global alcoholic drinks market, influence the size of the ripple effect between 2009 and 2014? To answer this question a regression model is used.
The literature review shows that a ripple effect consist of two parts which are the contagion effect and the competitive effect. Empirical evidence indicates that the contagion effect is more severe than the competitive effect. The empirical evidence also show that the marketshare of a company has an overall influence on the size of the contagion effect and the competitive effect. The larger the marketshare of a company the larger the ripple effect will
22 be. But the market structure hypothesis formulated by Cheng and McDonald (1996) indicates that for a contestable market the marketshare doesn’t have an influence on the ripple effect.
The results of the regressions of both CAR5 and CAR3 indicates that the marketshares of the announcing company and its competitors have a significant influence on the size of the ripple effect. The results also show that the assets of the competitors and the degree of
leverage of both the announcing company and its competitors doesn’t have a significant influence on the size of the ripple effect. The cumulative abnormal return and the assets of the announcer have a significant influence on the size of the ripple effect and so does the
correlation of the return of the stock prices of the announcing company and its competitors. In all the regressions there are company fixed effects and yearly fixed effects. This indicates that there may be company and yearly factors that influence the size of the ripple effect which aren’t included in the regression.
The findings of the influence of the marketshares, cumulative abnormal return of the announcer, assets of the announcer and correlation of the stock price returns are in line with the theory and the empirical evidence discussed in the literature review. The findings that the assets of the competitors and the degree of leverage doesn’t significantly influence the size of the ripple effect aren’t in line with the theory and the empirical evidence discussed in the literature review.
So to answer the research question. Marketshare does influence the size of the ripple effect for the global alcoholic drinks market between 2009 and 2014.
23
References
Akhigbe, A., Martin, A. D., & Whyte, A. M. (2005). Contagion effects of the world's largest bankruptcy: the case of WorldCom. The Quarterly Review of Economics and Finance ,
45 (1), 48-64.
Cheng, L. T., & McDonald, J. E. (1996). Industry Structure and Ripple Effects of Bankruptcy Announcements. The Financial Review , 31 (4), 783-807.
Chi, L.-C. (2009). Contagion and competitive effects of plan confirmation of reorganization filings: Evidence from the Taiwan Stock Market. Economic Modelling , 26 (2), 364-369.
Chi, L.-C., & Tang, T.-C. (2008). The response of industry rivals to announcements of reorganization filing. Economic Modelling , 25 (1), 13-23.
Cummins, D. J., Wei, R., & Xie, X. (2012, january 31). Financial Sector Integration and
Information Spillovers: Effects of Operational Risk Events on U.S. Banks and Insurers.
Retrieved december 2015, from SSRN:
http://papers.ssrn.com/sol3/papers.cfm?abstract_id=1071824
Ferris, S. P., Jayaraman, N., & Makhija, A. K. (1997). The response of competitors to announcements of bankruptcy: An empirical examination of contagion and competitive effects. Journal of Corporate Finance , 3 (4), 367-395.
Lang, L. H., & Stulz, R. M. (1992). Contagion and competitive intra-industry effects of bankruptcy announcements: An empirical analysis. Journal of Financial Economics , 32 (1), 45-60.
Pruitt, S. W., Wei, J. K., & White, R. E. (1988). The impact of union-sponsored boycotts on the stock prices of target firms. Journal of Labor Research , 9 (3), 285-289.
Slovin, M. B., Sushka, M. E., & Polonchek, J. A. (1999). An analysis of contagion and competitive effects at commercial banks. Journal of Financial Economics , 54 (2), 197-225.
Strong, N. (1992). MODELLING ABNORMAL RETURNS: A REVIEW ARTICLE. Journal
24 Wu, M. (2002, September). Earnings Restatements: A Capital Market Perspective. Retrieved
December 2015, from SSRN:
http://papers.ssrn.com/sol3/papers.cfm?abstract_id=1844265
Yu, X., Zhang, P., & Zheng, Y. (2010, May). Intra-Industry Effects of Corporate Scandal
Announcements: Evidence from China. Retrieved December 2015, from CAPANA:
25
Appendix
Table A1. Companies researched and their stock name used.
Anheuser-Busch InBev NV ANHEUSER-BUSCH INBEV (euronext)
SABMiller Plc SABMILLER
Heineken NV HEINEKEN
Carlsberg A/S CARLSBERG 'B'
China Resources Enterprise Ltd CHINA RES.BEER (HDG.)CO. Tsingtao Brewery Co Ltd TSINGTAO BREWERY 'A' Molson Coors Brewing Co MOLSON COORS BREWING 'B' Beijing Yanjing Brewery Co Ltd BEIJING YANJING BREW.'A' Kirin Holdings Co Ltd KIRIN HOLDINGS
Diageo Plc DIAGEO
Asahi Group Holdings Ltd ASAHI GROUP HOLDINGS Constellation Brands Inc CONSTELLATION BRANDS 'A'
26
Table A2. Announcement dates and title of article.
Anheuser-Busch InBev NV november 3 2010
Sales decline hit inbev’s Q3 profits; InBev Profits
SABMiller Plc may 20, 2010 Sabmiller slides on cautious outlook
SABMiller Plc may 24, 2012
Emerging economies drive brewer’s sales as mature markets stagnate
Heineken NV august 25, 2011
Heineken shares hit after profit warning
Heineken NV august 21, 2013
Heineken H1 profits fall on lower sales volumes
Heineken NV october 24, 2013
Heineken forecasts single-digit profits fall
Carlsberg A/S august 21, 2014 Carlberg’s Russia woes worsen China Resources Enterprise Ltd may 17, 2012
Taxing times for mainland supermarket chain
China Resources Enterprise Ltd may 17, 2013
China resources is hit by extravaganza clampdown
Molson Coors Brewing Co november 8, 2012 Corporate earnings
Molson Coors Brewing Co may 30, 2013 Czech mate hurts Molson’s bottom line Kirin Holdings Co Ltd november 6, 2009
Kirin Holdings downgrades FY net profit forecast 25% to JPY45b
Kirin Holdings Co Ltd october 17, 2012
Kirin profits sent reeling by discount milk prices
Diageo Plc february 10, 2011
Diageo leaves investors thirsty for more
Diageo Plc january 30, 2014
FTSE falls as emerging-market weakness hits Diageo
Constellation Brands Inc january 7, 2010
Wine and liquor marketer Constellation Brands 3Q profit falls on weaker US wine sales
Constellation Brands Inc october 8, 2012
Constellation Brands Q2 profit down 23%
27
Table A3. Companies’ Beta
ANHEUSER-BUSCH INBEV 0.334
SABMILLER 0.534
HEINEKEN 0.398
CARLSBERG 'B' 0.529
CHINA RES.BEER (HDG.)CO. 0.476 TSINGTAO BREWERY 'A' 0.064 MOLSON COORS BREWING 'B' 0.614 BEIJING YANJING BREW.'A' -0.016
KIRIN HOLDINGS 0.107
DIAGEO 0.407
ASAHI GROUP HOLDINGS 0.139 CONSTELLATION BRANDS 'A' 0.802
28
29
30
Table A6. Marketshare
Global Brand Owner 2009 (%) 2010 (%) 2011 (%) 2012 (%) 2013 (%) 2014 (%)
Anheuser-Busch InBev NV 14.6 14.5 14.2 14.1 15.6 16.3
SABMiller Plc 7.0 7.3 7.7 7.6 7.6 7.6
Heineken NV 5.3 6.8 6.9 7.3 7.2 7.3
Carlsberg A/S 4.6 4.6 4.4 4.4 5.0 4.8
China Resources Enterprise Ltd 3.5 3.9 4.2 4.2 4.8 4.7
Tsingtao Brewery Co Ltd 2.5 2.6 2.9 3.2 3.5 3.7
Molson Coors Brewing Co 2.1 2.2 2.1 2.6 2.5 2.5
Beijing Yanjing Brewery Co Ltd 2.0 2.1 2.2 2.1 2.3 2.2
Kirin Holdings Co Ltd 1.6 1.5 2.1 2.1 2.0 1.9
Diageo Plc 1.4 1.4 1.4 1.4 1.4 1.7
Asahi Group Holdings Ltd 1.2 1.1 1.0 1.0
Constellation Brands Inc 0.4 0.4 0.2 0.2 0.8 0.9
San Miguel Corp 0.8 0.8 0.8 0.8 0.8 0.8
Table A7. Total assets and leverage
Name 2009 2010 2011 2012 2013 2014
ANHEUSER-BUSCH - TOTAL ASSETS
in millions 77.9 84.9 86.3 92.3 102.3 117.0
ANHEUSER-BUSCH - TOTAL DEBT
% TOTAL ASSETS 44 39.52 35.94 36.4 34.97 36.13
SABMILLER PLC - TOTAL ASSETS in
millions 22.1 24.6 24.1 34.4 37.0 32.2
SABMILLER PLC - TOTAL DEBT %
TOTAL ASSETS 30.57 25.21 21.73 34.62 32.99 31.78 HEINEKEN N.V. - TOTAL ASSETS in
millions 19.6 26.1 26.7 35.4 32.8 34.2
HEINEKEN N.V. - TOTAL DEBT %
TOTAL ASSETS 41.99 33.41 34.45 37.72 37.07 34.41 CARLSBERG A/S - TOTAL ASSETS in
millions 133.0 142.9 146.5 152.8 150.0 135.6
CARLSBERG A/S - TOTAL DEBT %
TOTAL ASSETS in millions 29.61 25.57 24.73 26.22 26.81 29.9 CHINA RESOURCES - TOTAL ASSETS
in millions 75.1 88.8 112.9 126.5 153.7 179.1
CHINA RESOURCES - TOTAL DEBT
% TOTAL ASSETS in millions 16.16 13.85 13.76 14.01 14.77 16.14 TSINGTAO BREWERY CO - TOTAL
ASSETS in millions 14.6 17.4 21.2 23.2 26.7 26.3 TSINGTAO BREWERY CO - TOTAL
31 MOLSON COORS BREW - TOTAL
ASSETS in millions 11.8 12.5 12.3 16.1 15.5 13.9 MOLSON COORS BREW - TOTAL
DEBT % TOTAL ASSETS 14.46 15.67 15.98 29.02 24.45 22.86 BJ YANJING BREWERY - TOTAL
ASSETS in millions 11.9 14.7 16.8 18.2 18.9 18.9 BJ YANJING BREWERY - TOTAL
DEBT % TOTAL ASSETS 11.18 18.34 21.57 21.29 8.19 9.29 KIRIN HOLDINGS CO - TOTAL
ASSETS in millions 2802.1 2603.3 2812.0 2903.8 2851.9 2916.5 KIRIN HOLDINGS CO - TOTAL DEBT
% TOTAL ASSETS 34.35 33.01 40.71 35.5 30.12 29.93 DIAGEO PLC - TOTAL ASSETS in
millions 17.4 18.9 19.3 22.0 24.8 22.7
DIAGEO PLC - TOTAL DEBT %
TOTAL ASSETS 49.33 46.63 42.96 40.23 41.8 41.84 ASAHI GROUP - TOTAL ASSETS in
millions 1412.6 1374.9 1501.0 1714.0 1782.4 1927.1 ASAHI GROUP - TOTAL DEBT %
TOTAL ASSETS 31.3 22.81 29.98 26.33 23.95 23.77 CONSTELLATION BRANDS - TOTAL
ASSETS in millions 8.0 8.1 7.2 7.1 7.6 14.3
CONSTELLATION BRANDS - TOTAL
DEBT % TOTAL ASSETS 55.17 47.57 45.15 44.02 43.28 49.09 SAN MIGUEL CORP - TOTAL ASSETS
in millions 429.6 822.7 882.3 1029.1 1154.5 1202.4 SAN MIGUEL CORP - TOTAL DEBT
% TOTAL ASSETS 30.2 54.88 57.04 55.45 55.94 55.67
32
