Short-term return gained by merger announcements in the airline industry
Calculating the abnormal return in an event study
Arie Brouwer
Student Number 10274006
Bachelor program Economics & Business Specialization Economics & Finance
Faculty of Economics and Business, Universiteit van Amsterdam Amsterdam, July 14th, 2014
Abstract
This thesis studies the effect merger announcements have on the stock price of the target and the acquiring firms in the North-American and European airline industry in 1998-2014 and if shareholder return is gained by this process. We study mergers in the airline industry between 1998 and 2014. By implementing an event study and calculating the abnormal return we can find out if there is a positive shareholder return gained by the targets, the acquirers and the combined companies.
During the periods of interest, (-1, 1) and (-10, 1) days around the merger announcement, the shareholders of the combined firms had an insignificant positive cumulative abnormal return of 1.031% and 2.078% , respectively. The shareholders of the acquiring firms had an insignificant negative cumulative abnormal return of -0.421% and -2.307%, respectively. We therefore cannot say that there was a positive return gained by the shareholders of the acquiring and combined firms during the pre-merger period. The shareholders of the target firms had an insignificant positive cumulative abnormal return of 2.483% during the period (-1, 1) and a significant positive cumulative abnormal return of 6.463% during the period (-10, 1). We can however say that on the announcement day the shareholders of the target airline companies and the combined firms have a statistically significant benefit from the pre-merger dealings. There was a positive abnormal return of 2.220% and 1.531% during the period (-1, 1) and 2.146% and 1.211% during the period (-10, 1), respectively. We may speak of statistically significant higher shareholders’ return on the announcement date of the merger.
Between 1986 and 2014, many laws and deregulation packages have been implemented to enforce competition in the airline industry. These laws and deregulation packages, together with our findings of abnormal returns, which are lower than the results from the studies of Singal (1996) and Knapp (1990), could indicate a more competitive airline market in comparison to 20 years ago.
Table of Contents
Abstract……… 1
Table of Contents……… 2
1. Introduction………..……….………. 3
1.1 The research question……….………. 3
1.2 Hypothesis……….……….……… 4
1.3 Structure………... 5
2. Literature review……….. 6
2.1.1 Mergers, acquisitions and their motives ……….………..………..…..……… 6
2.1.2 Reasons for a merger to be executed, while not giving any shareholder return………... 7
2.1.3 Measuring the success of a merger………..………..………. 8
2.1.4 Merger waves………..………. 9
2.2 The Airline Industry………..……….…………..………. 11
2.3 Mergers in the Airline industry………..……… 12
3. Method………..……… 14
3.1 Finding the event date, estimation period and the event window………..….……….. 15
3.2 Selection criteria for data………..……….……….. 16
3.3.1 Calculating normal market parameters and measuring abnormal return………....………. 17
3.3.2 Calculating the normal return………. 19
3.3.3 Measuring the abnormal return………..………..……….. 20
3.4 Statistical significance………..……….……… 21
4. Data………….……….……… 23
5. Results……….……….. 25
6. Conclusion………. 32
7. References……….…….. 34
8. Appendix……… 37
1. Introduction
2In the last few decades mergers and acquisitions have become a much debated subject. The first merger waves began in the early 1900s, when managers hoped to expand and improve their businesses by acquiring each other. It was widely held that, for example, synergies or economics of scale would make such deals profitable. However, Walsh (1988) researched the success of these mergers and found that target firm executives experience considerable acculturative stress. Almost 70% of all mergers depart in the five years following completion.
The existing studies, however, differ in their results which makes this topic an interesting one to expand on. Singal (1996) for example analyzed 14 airline mergers between 1985 and 1988 when there was a less serious threat of an antitrust challenge. He found that airline mergers during that period both enhanced market power and made the merging firms' operations more efficient. Gaughan (2010) on the other hand analyzed the fifth merger wave between 1998 and 2001 where shareholders lost $240 billion, because of short-term financially orientated deals.
1.1 The research question
Many of these studies were about mergers and acquisitions in general, like Agrawal and Jaffe (2000) who concluded that managers of acquiring firms report that only 56% of their acquisitions can be called successful compared to the original objectives they wanted to accomplish. There have also been studies about mergers and acquisitions in specific sectors of the economy, however almost none about the airline industry alone. Most studies about mergers and acquisitions in the airline industry are about changes in the price of flight tickets, customer satisfaction, market power or case studies about specific mergers.
Because of the Airline Deregulation Act in the United States in 1978, the American airline industry had to cope with intensified competition. The new low cost carriers that entered the market, as well as the terrorist attack on 9/11 and the still rising fuel costs (which presently make up 30% of their total costs) necessitated considerable cost-cutting throughout the airline industry (Morrison & Winston, 1995). During this period the European air transport market was heavily centered around the national airlines, which lead to little incentive for cost-cutting and high-ticket prices. Positive experiences with the more competitive airline industry in the United States opened the door to deregulation of the European air transport market. A package of deregulation measures was implemented in 1987, which was the first step toward a single European airline market (Burghouwt & Huys, 2003).
The fact that there has not been much research about the impact on the shareholders’ return, combined with the rising competition in the airline industry, makes research about mergers and acquisitions in this sector particularly relevant. This is why the research question of this thesis is:
What effect did merger announcements have on the stock price of the target and the acquiring firms in the North American and European airline industry in 1998-2014 and was shareholder return gained by this process?
1.2 Hypothesis
In order to find out what effect North American and European merger announcements in the airline industry have on the stock price of the target and the acquiring firms in 1998-2014 we have the following two hypotheses.
For the targeted firms the hypothesis is:
H0: The return gained by shareholders of the target firm during the merger announcement is zero H1: The return gained by shareholders of the target firm during the merger announcement is positive
For the acquiring firms the hypothesis is:
H0: The return gained by shareholders of the bidding firm during the merger announcement is zero H1: The return gained by shareholders of the bidding firm during the merger announcement is positive
In addition, the hypotheses for shareholder return in total is:
H0: The combined return gained by the shareholders during the pre-merger announcement is zero H1: The combined return gained by the shareholders during the pre-merger announcement is positive
Shareholder return will be measured by changes in the stock price during the merger announcement, using an event study.
1.3 Structure
This thesis is structured as follows. In part 2, we give an overview of the motives for mergers and acquisitions and we briefly go through the history of the airline industry. Relevant studies and their empirical results will be discussed as well. Section 3 describes the method used to calculate the shareholders’ return and the selection criteria used for the analyzed data. Section 4 shows all the mergers and acquisitions used in this study. Section 5 shows the results gained by using the established method. The effect the merger announcement had on the stock prices will be analyzed, as well as their statistical significance. Section 6 will contain the conclusion of this thesis.
2. Literature review
In this section an overview will be given of mergers and acquisitions in general. The reasons why firms want to merge with each other and how to measure the success of this phenomenon will be discussed. The main empirical results of important studies about mergers and acquisitions and their effect on shareholders’ return will be reviewed. Finally, we look into the history of the European and American Airline industries and the merger waves which have come to pass.
2.1.1 Mergers, acquisitions and their motives
Over the last two decades, mergers and acquisitions have rapidly become a more discussed subject in the United States and especially in Europe. The volume of mergers and acquisitions in Europe picked up dramatically in 1995, when it doubled (Gaughan, 2010). At its peak in 1999, the value of European mergers and acquisitions equaled 38% of the total global M&A deal value. In the early 1980s the main center of big M&A deals was the United States (Gaughan, 2010). Europe started to become as important as North America when it came to merger deals, which made both continents important to analyze in this thesis. To answer the research question of this thesis however, we first need to understand what a merger exactly is. A merger is not a physical transportation of assets or stocks of one corporation (the target) to the other (the acquirer). Instead the two corporations that are going to merge become a unified operation by law. In this study mergers and acquisitions are considered the same for the sake of simplicity. This strategy of merging is used to achieve for example corporate growth, diversification,
economies of scale or vertical integration (Kwall, 2006)
Trautwein (1990) researched theories of merger motives and distinguished three cases in which it would be profitable for shareholders to participate in a merger:
1. Efficiency theory
This theory views mergers as being executed to achieve synergies. There are three types of synergies which can be distinguished. Lower costs of capital can be achieved through financial synergies. This can be accomplished by reducing the systematic risk of a company's investment portfolio by investing in unrelated businesses. Operational synergies may lower the costs of the involved business units or can enable the company to offer unique products and services by combining operations. Managerial synergies may be obtained when the managers of the acquirer are superior in planning and monitoring to the managers of the target, which benefits the target’s performance.
2. Monopoly theory
The Monopoly theory views mergers as being executed to achieve market power. To obtain market power, firms can cross-subsidize products. The profits gained from their strong market position are used to sustain a fight for market share in another market. The firm can also compete in more than one market, through tacit collusion with competitors or by restricting entry.
3. Valuation theory
The Valuation theory argues that mergers are executed by managers who have superior information about the value of the target. In this situation managers of the bidding company have unique information, which the stock market does not have, about possible advantages that may be obtained by combining the two companies.
2.1.2 Reasons for a merger to be executed, while not giving any shareholder return
The three theories above predict that mergers should have a positive effect on shareholders’ return. Empirical research has however documented that the bidding firms consistently pay large premiums for target firms. This result is on par with many studies on mergers, which show that shareholders of the bidding firms do not obtain statistically significant gains from all the mergers they undertake (Asquith, Bruner & Mullins, 1982). During the fifth merger wave, from 1998 to 2001, the shareholders of the acquiring firms lost 12 cents around acquisition announcements per dollar spent on acquisitions, which accounted for a total loss of $240 billion (Moeller, Schlingemann & Stulz, 2005).
These examples show that mergers are not only done because of the Efficiency, Monopoly and Valuation theory above. According to Trautwein (1990) there are two more evidence-based theories why a merger is executed while not giving any immediate return to the shareholders.
4. Empire-building theory
The Empire-building theory states that mergers are executed by managers who maximize their own utility instead of maximizing shareholder return. Most managers’ salary is based on the size of the firm they manage, which gives them motivation to expand the company instead of doing what best is for the shareholders.
5. Process theory
The Process theory argues that mergers are not executed as comprehensively rational strategic decisions. This theory states that individuals have limited information processing capabilities. This means that the search for information and alternatives by individuals is simplified and not complete, which gives incentives for wrong decisions.
Mergers can lead to a positive abnormal return for shareholders, depending on which strategy the manager of the bidding firm used as reason to merge. When a manager has inadequate information or wants to maximize his own utility, mergers may be unprofitable. To calculate what kind of effect the merger has on shareholder return, MacKinlay (1997) uses an event study to calculate the abnormal return.
2.1.3 Measuring the success of a merger
Empirical research on mergers has revealed that when mergers create shareholder return, most of the gains are accrued by the target company (Andrade, Mitchell & Stafford, 2001). Andrade, Mitchell & Stafford (2001) investigated whether mergers create return for shareholders with traditional short-window event studies. They did this by examining changes in the stock price around the merger announcement compared to what the stock price would have been had there not been a merger, which is called the abnormal return. They calculated a statistically significant abnormal return of 13% for the target firm, when examining the period from one day before to one day after the merger announcement. The bidding firm had a statistically significant negative abnormal return of -1.5% (Andrade, Mitchell & Stafford, 2001)
Moeller, Schlingemann & Stulz (2005) examined the influence of the fourth merger wave on the shareholder return of the acquiring firm and compared it to the impact of the merger wave in the 1980s. They calculated the average cumulative abnormal return over the specific event window for the value-weighted portfolio of the target and the bidder. They found that merger announcements in the 1990s are very profitable for the acquiring-firm shareholders until 1997, resulting in a total shareholder return of $ 8 billion. In the fifth merger wave however, between 1998 and 2001, there was a total loss of $240 billion (Moeller, Schlingemann & Stulz, 2005). This was because of a rather small percentage of extremely unprofitable mergers. Were one to exclude just over 2% of the observations, shareholder return would have increased with acquisition announcements (Moeller, Schlingemann & Stulz, 2005).
The two studies above both used an event study to calculate the abnormal return when researching the return gained by a merger announcement. Dyckman, Philbrick & Stephan (1984) also performed an event study, where they calculated the abnormal return of a merger announcement in order to find out which model best predicts the return gained during a merger announcement. There was a statistically significant difference that gives a slight preference to the Market Model (Dyckman, Philbrick & Stephan, 1984), which will be further discussed in section three.
2.1.4 Merger waves
To decide which time span to use for this thesis and which continents to look at, we used the history of mergers in the past. There have been six merger waves throughout history, which all had a different influence on the economy and all had a different impact on shareholders’ return. Gaughan (2010) analyzes all six merger waves, the first four of which only took place in the United States. The first merger wave started in 1897 and ended in 1904. The mergers of this first wave were 78.3% horizontal, which resulted in monopolistic market structures (Gaughan , 2010). After World War I, an economic boom gave companies a lot of investment capital, which lead to the second merger wave, starting in 1916. Government structures were worried by the power that monopolies had gained, so they started to enforce antitrust laws1 (Henry, 1965). Therefore mergers in the second wave were predominantly vertical, which ended when the stock market crashed in 1929 (Gaughan , 2010).
During the 1940s there were neither technological breakthroughs nor dramatic changes in infrastructure, so the increase in mergers was relatively small. Larger firms did acquire privately held smaller firms for motives of tax relief. During this period of high estate taxes, the transfer of a firm within the family was very expensive2 (Eistenstein, 1956).
The third merger wave dates from 1965 to 1969 and occurred during another economic boom in which 80% of the mergers that took place were conglomerate mergers3 (Gaughan , 2010). In 1962, the Supreme Court provided a stringent initial interpretation to section 7 of the Clayton Act (Harlan, 1973). As a consequence, horizontal and vertical acquisitions of any size became relatively rare, which
1Congress passed the Clayton Act in 1914, an antitrust law that reinforced the antimonopoly provisions
of the earlier Sherman Act.
2 On June 19th, 1935, the President of the United States recommended, in addition to the estate tax, an inheritance succession and a legacy tax in respect to all large amounts received by any legatee. After this Congress increased the taxes with 2% on the first tax bracket of $10000 and ended with 70% over the last bracket above $50 million.
3 A conglomerate merger is a merger between two firms, that are involved in totally unrelated business activities.
9
gave more incentive to invest in conglomerate mergers. However, most of these mergers were not profitable. 60% of the cross-industry mergers were sold or divested, because the acquisitions were followed by poor financial performance (Gaughan , 2010). The unique characteristic of the fourth merger wave from 1984 to 1989 was the significant role of hostile takeovers. This was the wave of the megamergers and billion dollar deals, for which a lot of debt was used instead of capital owned (Gaughan , 2010).
The fifth merger wave from 1998 to 2001 was a truly international one. The volume of mergers and acquisitions in Europe picked up dramatically in 1995, when it doubled (Gaughan, 2010). At its peak in 1999, the value of European mergers and acquisitions equaled 38% of the total global M&A deal value. During the fourth merger wave and in the mid-1990s there were a lot of profitable long-term strategic deals, which lead to $8 billion return gained for shareholders (Moeller, Schlingemann & Stulz, 2005). However, between 1998 and 2001 shareholders lost $240 billion, because of short-term financially orientated deals. This was because of a rather small percentage of extremely unprofitable mergers. Were one to exclude 2% of the observations, shareholder return would have increased with acquisition announcements (Moeller, Schlingemann & Stulz, 2005). The last merger wave, the 6th one, was in 2004-2007, where new potential targets and bidders entered the market as result of increased privatization4. This merger wave was very much like the 5th merger wave where the value of the deals that were made is concerned. Both merger waves had a total annual value of M&A’s between $750.000 and $2 million, whereas in the first four merger waves the highest total value reached was $500.000 in 1988 (Gaughan, 2010).
The period of the 5th and 6th merger waves is interesting to analyze, because the two waves are similar where total M&A deal value is concerned. Only 2% of the merger observations during the 5th merger wave was responsible for the loss of $240 billion in shareholder return (Moeller, Schlingemann & Stulz, 2005). This fact would make the stock data during the 5th merger wave still usable for our analysis, when combining it with the stock data of the 6th merger wave. During the 5th merger wave the value of European mergers and acquisitions equaled 38% of the total global M&A deal value . This makes not only North America a continent to consider for this thesis, but also Europe.
4Especially Eastern Europe, Asia and Central and South America.
10
2.2 The Airline Industry
Because of the American Airline Deregulation Act of 1978, the American airline industry had to cope with intensified competition. These laws reduced regulatory restrictions and widened the
opportunities for airlines to set air fares and choose capacity and frequency5 (Moore, 1986). Because of this, many low cost carriers entered the market. Big companies like Pan America and Midway could not handle this intensified competition and went bankrupt or were liquidated (Morrison & Winston, 1995). Furthermore, the terrorist attacks of 9/11 caused the American airline industry’s profits to decrease6 (Goodrich, 2002). The increased competition, the terrorist attack and the rising fuel prices, which made up to 30% of the total airline costs in the nineties, created a lot of pressure to cut costs (Morrison & Winston, 1995).
The American Airline Deregulation Act in 1978 also had a great impact on the airline industry in Europe. During this period the European air transport market was heavily centered around the national airlines. The airlines were more or less the offspring of the governments. Most flag carriers, like KLM, British Airways and Air France were heavily subsidized or owned by their government. This lack of competition resulted in little incentive for cost cutting and high-ticket prices (Burghouwt & Huys, 2003). Positive experiences with the more competitive airline industry in the United States opened the door to deregulation in the European air transport market. The first package of deregulation measures was implemented in 1987, which was the first step toward a single European airline market7. By implementing a second and third package in 1990 and 1993 respectively, the European airline market was further deregulated (Marin, 1995).
Due to these changes, major European airlines adopted new network strategies to deal with this intensified competition. The formation of global strategic alliances and mergers in the 1990s, the adoption of hub-and-spoke networks and the low-cost concept are the most important of these new strategies (Burghouwt & Huys, 2003).
5 The Civil Aeronautics Authority (CAB) moved to reduce entry barriers and to permit considerable pricing
freedom. The major provisions of the American Airline Deregulation Act in 1978 called for an open-entry plan under which carriers could enter one new market a year until 1981. Furthermore, the bill instructed the CAB to grant operating rights to any air carrier seeking to serve a route on which only one other carrier was actually providing service. The Act also permitted carriers to lower prices 50% or raise them 5%.
6 It is estimated that the US airline industry lost between $1 and $2 billion during the first week after the
terrorist attack on 9/11. The Federal Government shut down the US airline industry for two full days. During that time, the airlines lost over $100 million in sales revenue. They continued to suffer millions of dollars in lost revenue as much of the traveling public was too scared to fly. Planes kept flying half-empty three months after the tragedy.
7 Freedom of market entry to any European airline. No restrictions on capacity or frequency. Freedom to
offer any fare, except if it is disapproved by the two countries involved.
11
2.3 Mergers in the Airline industry
Most studies about mergers in the airline industry are about changes in the price of flight tickets, customer satisfaction, market power or case studies about specific mergers. For example the merger between KLM and Air France (Friesen, 2005). An event study showed that stockholders of the acquirer, Air France, earned insignificant positive abnormal returns of 0.24% on the announcement day. The target, KLM, experienced a significant positive abnormal return of 1.67% (Friesen, 2005). This agrees with what Andrade, Mitchell & Stafford (2001) concluded in their study. When mergers create shareholder return, most of the gains are accrued by the target company. Friesen (2005) used 19 important events, from the moment that the merger was announced, till the delisting of KLM shares. All 19 events had an event window of 7 days (-3, +3). However, such an extensive period of analysis is not used in studies that analyze multiple mergers, like Moeller, Schlingemann & Stulz (2005).
There are some studies which are more relevant to this thesis. Singal (1996) analyzed airline mergers in the United States between 1985 and 1988. Of the 27 airline mergers identified, 8 small mergers were excluded because of non-available stock data. The identified mergers were found by using the Wall Street Journal index and trade journals such as Air Transport World and Aviation Week And Space Technology. By doing an event study and analyzing the significance of abnormal stock returns around the announcement of a merger proposal, Singal (1996) studies the effect of airline mergers on market power and efficiency gains. For this research he uses four different event periods, which are (-1, 0), (-1, 1), (-3, 1) and (-5, 1). The first is a 2-day event period, one day before the merger announcement and the day of the merger announcement. The second is a 3-day event period, one day before the merger announcement until one day after the merger announcement, etc. Singal (1996) obtained the announcement date from the Wall street Journal and based the normal market parameters on the 270 days prior to the beginning of the event period.
On average, the target firms earn statistically significant positive cumulative returns ranging from 18.43% to 22.00% depending on the event period. The event periods (-1, 0) and (-1, 1) had the highest t-values here, with 17.02 and 14.24 respectively (Singal, 1996). The bidding firms earn statistically significant positive cumulative returns ranging from 1.32% to 1.88% depending on the event period. The event periods (-1, 1) and (-1, 0) had the highest t-values, with 3.16 and 2.36 respectively (Singal, 1996). The combined firms earn statistically significant positive cumulative returns ranging from 3.157% to 4.193% depending on the event period. Singal (1996) concluded that airline mergers during that period both enhanced market power and made the merging firms' operations more efficient, by 12
also investigating the effect the mergers had on the stock prices of the rivals of the airlines that merged. However, this goes beyond the scope of this thesis.
One of the studies that comes closest to this thesis was conducted by Knapp (1990). Knapp (1990) uses an event analysis to study the motivation and effects of horizontal mergers in the airline industry of the United States. Their study examined 9 air carrier acquisitions announced in 1986. This group of mergers was unique, because the relevant antitrust agent at that time, the Department of Transportation, offered minimal resistance to the mergers in question. According to Knapp (1990), the Market Model is generally used for an event analysis. However, standard event methodology is inappropriate in this study, because only one year is tested and so the merger announcements are close together in calendar time. This situation causes Market Model residuals to be correlated across firms or the variance to be heteroscedastic. To correct this cross-sectional dependence, Knapp (1990) used the independent draw method.
Knapp (1990) based the normal market parameters on the 200 days prior to the beginning of the event period and used 7 event periods around the merger announcement, namely (-1, 0), (-1, 1), (-3, 3), (-10, 5), (-20, 0) and (-20, 10). The target firms earn statistically significant positive abnormal returns of 25% for the 20 days before and 10 days after the merger announcement. Most of the gains for the target firms were gained in the 20 days before the merger announcement (Knapp, 1990). The abnormal return became however slightly negative, but not significant, after the event of the merger announcement. The event period (-20,0) possessed the highest t-value (t=3.65) for the targets. Bidding companies experienced a significant abnormal return between 6% and 12%, where the event period (-10,5) possessed the highest t-value (t= 2.32) for the acquirers (Knapp, 1990). According to Knapp (1990) these significant positive abnormal returns for acquirers demonstrate that the airline merger market was less than competitive.
The studies by Singal (1996) and Knapp (1990) are most in line with this thesis. Therefore the method of this thesis will be based mostly on their decisions. The Market Model will be used for the event analysis, because this study is over a period of 15 years, which makes it unlikely that residuals will correlate across firms or the variance to be heteroscedastic. Singal (1996) found his most significant results with the event periods (-1,0) and (-1,1) and Knapp (1990) with the event periods (-20,0) and (-10,5). Because of the large difference between the event periods of the two studies, this thesis will use two event periods, namely (-10, 1) and (-1,1). The method will be further discussed in section three.
3. Method
To empirically test our research hypothesis and find out what the short-term effects of mergers in the airline industry on shareholders’ return are, we need to use an event study. An event study measures the impact of a specific event on the total value of a firm. Effects of an event will be shown in the market prices in a relative short time period, given the market is rational. To calculate the impact of this event we require a measure of the abnormal return, which is the actual ex-post return minus the expected return of the security, during the event. These results should be tested for their statistical significance (MacKinlay, 1997).
To look at stock market reactions to merger announcements and its effect on shareholders’ return, Andrade, Mitchill & Stafford (2001) chose an event-study as their method. They used the abnormal returns to calculate the shareholders’ return gained during the 3 day event period around the merger announcements.
According to MacKinlay (1997), the following steps have to be taken to conduct an event study:
• First define the event of interest and choose the period over which the share price of the designated firm will be examined. Due to price effects of announcements that occur after the stock market closes on the announcement day or because there is information known prior to the announcement, it is customary to expand your period of interest to multiple days. The event of interest in this study is the merger announcement and the period of interest are the days around the merger announcement.
• The estimation window needs to be determined. It is customary, while having access to daily data, to estimate the normal/expected market model parameters over the days prior to the announcement. The days around the announcement themselves should not be included in the estimation, to exclude the event from influencing the normal market model parameters.
• The third step consists in choosing selection criteria for which sample set of firms is included in the analysis. We are looking at European and American mergers in the publicly listed Air industry since 1998.
• The abnormal return should be calculated for all the stock data of the period of interest for the sample set of firms that are included in the analysis. This can be done by subtracting the expected return from the actual ex-post return of the involved firm.
• Finally, the results must be tested for statistical significance.
3.1 Finding the event date, estimation period and the event window
To calculate the abnormal return, we first need to determine which days exactly to look at. For this we need to choose our event date, estimation period and event window. The event day, is the date the specific event takes place. The period of interest are the days which are going to be examined, before and after the event date, because of prior knowledge and late reactions. The estimation window consists of the days before the event date, to calculate the normal market model parameters.
Event date
The event date that will be analyzed in this study is the announcement date of the merger. Knapp (1990) and Singal (1996) both use the announcement date, to calculate the abnormal return for mergers in the airline industry. According to Brown and Warner (1980) the best and accurate existing technique to determine the event date is to read old issues of the Wall Street Journal. Singal (1996) also obtained the announcement date out of the Wall Street Journal. Because of this, the Wall Street Journal will also be used to determine the event date in this thesis. In the figure below the event date/merger announcement is T0.
Period of interest
Due to price effects of announcements that occur after the stock market closes on the announcement day and because there is information known prior to the announcement, it is customary to expand your period of interest to multiple days (MacKinlay, 1997). Moeller, Schlingemann & Stulz (2005) used a three day event study to prove that in the 1980s the shareholder gains of the target firm exceeds that of the acquiring shareholders. To look at stock market reactions because of a merger announcement and its effect on shareholders’ return, Andrade, Mitchill & Stafford (2001) also chose a three day event period. The most significant results by Singal (1996) were also in a three day event period. Knapp (1990) however concluded that most gains for the target firms were made in the 20 days before the merger announcement. Because of the large difference between the event periods, which both give statistically significant results, this thesis will use two event periods. This study will analyze the event periods 10 days before the merger announcement until a day after the merger announcement (-10, 1) and a day before the merger announcement until a day after the merger announcement (-1,1). In the figure below the periods of interest will be between T-1 and T1 and between T-10 and T1.
Event window
According to MacKinlay (1997), the days around the merger announcement should not be included in the estimation, to exclude the event from influencing the normal market model parameters. In an example MacKinlay uses 250 to 20 days before the announcement, where other studies like Moeller, Schlingemann & Stulz (2005) use an event window from 205 to 6 days before the merger announcement. Armitage (1995) concluded that an estimation period of 100-300 days is sufficient for an assessment of the normal market model parameters. For this reason this study will use an event window of 200 to 20 days prior to the merger announcement. This is also in line with the event windows of Knapp (1990) and Singal (1996), who had event windows of 200 and 270 days prior to the beginning of the event period, respectively. In the line below the event window will be between T-20 and T-200.
3.2 Selection criteria for data
To find out what kind of an effect merger announcements in the airline industry have on the stock price of the target and the acquirer, this study is going to look at mergers between American and European airlines from 01-01-1998 till 01-01-2014. This period has been chosen, because 1998 was the first year of the 5th merger wave. According to Gaughan (2010) the 5th merger wave from 1998 to 2001 and the 6th merger wave from 2004-2007 were very much alike in the value of deals that were made. Because these merger waves are similar in this regard, both waves are used in this study to have as much data as possible and still be relevant for mergers nowadays.
By using the databases DataStream and Yahoo-Finance we can find all the security prices of and before the merger announcement. To select mergers which are suited for this study we use the selection criteria below:
• Both acquiring and target firm should be American or European air transportation companies. • The companies involved in the merger/acquisition should be publicly traded.
• Before the announcement date the acquirer controls less than 50% of the total shares of the target and after the merger is completed controls 100% of the total shares of the target.
• All information of the M&A about announcement and completion dates should be publicly accessible.
• Daily stock returns for the acquiring and target firm can be accessed by DataStream or Yahoo finance, at least 200 days before the announcement date.
• Daily stock returns of the corresponding market index can be accessed by DataStream, at least 200 days before the announcement date.
• There is at least one year difference between the last merger completion and the coming merger announcement8.
• All announcement dates of the M&A should be between 01-01-1998 and 01-01-2014. • All M&A should be completed.
3.3.1 Calculating normal market parameters and measuring abnormal return
Before measuring the abnormal return, the benchmark of normal returns should be obtained first. These benchmarks should be calculated over a period in which it is certain that the merger announcement will not have any influence over the normal market parameters. There are a number of approaches to calculate the normal return. The empirical results which will be gathered in this study are based on the following model:
(1) Rit = αi + βi*Rmt + εit (2)
Rit = The actual return on security i for day t
8 There needs to be a timespan between mergers involving the same firm. When two mergers which involve the same firm happen too close to each other the normal market parameters of the second merger could be influenced by the first merger. For example SAS Sverige AB and SAS Norge ASA
announced their merger on 4-20-2001, while also merging with SAS Danmark AS. These two mergers at the same time would influence the analysis
17
Rmt = The market return for day t εit = the error term for security i for day t αi , βi = Constants which are firm-specific
Dyckman, Philbrick & Stephan (1984) and MacKinlay (1997) discussed 3 models, which can be used to calculate the normal return using the formula above:
1. Mean-Adjusted return model
This model is labeled as the most naïve and simple model, because market-wide factors and risk are not accounted for. The expected return for a security is equal to the firm-specific constant, by calculating the average of a series of past returns. To get this estimation from the equation above (1), αi is set equal to the average return over the estimation period and βi is set equal to zero.
2. Market- adjusted Return Model
The estimated firm return is equal to the market return for that specific period. Estimated returns are not constant across time, but are constant across securities. To enforce this in formula (1) αi is set equal to zero and βi is set equal to one. The market-adjusted model is mostly used when there is limited data. Especially when it is the case that a pre-event estimation period is not possible.
3. Market Model
This model relates the return of a given security to the return of the market portfolio. The expected return of the firm is a linear function of the market return with the use of an OLS coefficient. This way αi and βi in equation (1) are calculated over the estimation period by using OLS. By using this method the variance of the abnormal return is reduced by excluding the part of the return that is related to variation in the market's return. This can lead to an increased ability to detect event effects. The error term εit of the Market Model will be assumed to be zero, because the relation between the security and market return remains unchanged.
According to Dyckman, Philbrick & Stephan (1984) the abilities of the three models above to measure the abnormal return correctly are quite similar. There is however a statistically significant difference that gives a slight preference to the Market Model.
According to Knapp (1990), the Market Model is generally used for an event analysis. However, standard event methodology is inappropriate in a study when merger announcements are close together in calendar time. Because this study is over a period of 15 years, it is unlikely that residuals will correlate across firms or the variance to be heteroscedastic. For this reason the Market Model, which relates the return of a given security to the return of the market portfolio is chosen to measure the normal return.
3.3.2 Calculating the normal return
We use the Market Model in this study to measure the normal return. The Market Model relates the return of a given security to the return of the market portfolio. This model will be used 200 to 20 days prior to the merger announcement. In this case. the formula for the Market Model will be:
(2) Rit = αi + βi*Rmt + εit (t= -200 ; t= -20)
The normal return R from security i on day t must be calculated. In order to calculate this for each firm, the daily returns over the total period are used to estimate the regression equation. The Market Model assumes that the normal return is jointly normally distributed and independent through time.
In order to calculate Rit we need to find the return of the market portfolio Rmt. Therefore we use main stock indices. We use the databases ‘DataStream’, which shows the main stock index of each specific country of the firm and Bloomberg9, a respected financial software, data and media company, to determine a suitable main stock index. When a suitable main stock index is determined we use DataStream to get the stock data from the specific main stock index. Aside from the return of the market portfolio, the firm-specific constants αi and βi must be calculated as well. αi and βi can be determined by using an ordinary least square (OLS) regression in the given estimation window (t = -200; t=-20). The error term εit of the Market Model will be assumed to be zero, because the relation between the security and market return remains unchanged (MacKinlay, 1997).
9 13-6-2014: http://www.bloomberg.com/- markets/indexes/
19
3.3.3 Measuring the abnormal return
The abnormal return is the actual ex-post return minus the expected return of the security, during the event. When the Market Model is used to calculate the normal return, the sample of the abnormal return will be:
(3) ARit = Rit – (αi + βi*Rmt) (t = -1 ; t = 1) and (t = -10 ; t =1)
Where Rit is the actual daily return and (αi + βi*Rmt) is the expected daily return of the firm for that specific event. This will be calculated over the periods of interest (-1, 1) and (-10, 1). αi and βi can be determined by using an ordinary least square (OLS) regression. The Market Model assumes that the abnormal return is jointly normally distributed and independent through time.
All observations should be aggregated through time and across the different firms to be able to draw any conclusions. The observations of the abnormal return can be aggregated across firms by calculating the daily average abnormal return across N securities, which is the average of the abnormal returns of all N firms on one specific day. To calculate this, we use the following formula:
1
1
(4)
N t i t tAR
AR
N
==
∑
Where N is the number of securities used in the study, ARtis the daily average abnormal return across N
securities and ARit is the abnormal return of one security for one specific day of the period of interest. When the period of interest which is being studied is more than one day, we need the concept of ‘cumulative average abnormal return’ to aggregate through time. The cumulative average abnormal return is the sum of the daily average abnormal return across N securities from the total days in the period of interest (MacKinlay, 1997):
2 1 2 1 ( , )
(5)
t t t t t tCAR
AR
==
∑
20Where CAR is the cumulative average abnormal return for the event window t1 to t2 andARtis the daily
average abnormal return across N securities. To ensure that the abnormal average returns and cumulative average abnormal returns are statistically significant, we conduct a t-test and check the null hypothesis.
3.4 Statistical significance
To ensure statistically significant results, a hypothesis test is conducted by checking the null hypothesis. When the null hypothesis is correct, shareholders have no benefit from the merger announcement. To analyze each individual day of the period of interest and check if there is a positive abnormal return we use the following hypothesis test:
0 1
:
0(
1;1)(
10;1)
:
0(
1;1)(
10;1)
t tH
AR
t
t
H
AR
t
t
=
= −
= −
>
= −
= −
To analyze if the period of interest has a positive abnormal return we used the hypothesis test below:
0 1
:
0(
1;1)(
10;1)
:
0(
1;1)(
10;1)
t tH
CAR
t
t
H
CAR
t
t
=
= −
= −
>
= −
= −
According to Brown and Warner (1980) a one day t-test for abnormal returns can be calculated with the following formula: 2 20 200
(6)
1
180
180
t t t t tAR
t
AR
AR
=− =−=
−
∑
∑
WhereARt is the daily average abnormal return across all merger announcements. And the formula
below is the standard deviation derived from the estimation period.
2 20 200
1
(
)
180
180
t t t tAR
s AR
AR
=− =−
=
−
∑
∑
According to MacKinlay (1997) this standard deviation can be calculated by first calculating the variance of the cumulative average abnormal return by using the following formula and take the square root of the outcome after:
1 2 2 ( , ) 1 2 2
1
(7)
Var CAR
(
t t)
( , )
t t
N
σ
=
∑
To perform a t-test on the cumulative average abnormal return the formula below is suggested:
( 1,1) 2 20 200
(8)
1
180
180
t t t tCAR
t
AR
AR
T
− =− =−=
−
∑
∑
Where CAR is the cumulative average abnormal return for all merger announcement for all three days and T is the number of days of the period of interest.
By using these estimates we can analyze the abnormal return for any period. When the test value for
t
AR or CAR is greater than 1.96, the null hypothesis is rejected and it can be concluded that the
abnormal return is significantly different from zero at a 5% significance level. When the test value is greater than 1.64, the null hypothesis can be rejected at a 10% significance level.
4. Data
Between 01-01-1998 and 01-01-2014 there were 13 mergers between airlines that were publicly traded in the United States, 2 mergers in other parts of America and 11 European mergers. We found these mergers by screening for the Selection criteria in section 3.2. Based on these selection criteria many firms that merged were not suitable for this study, because they were not publicly traded, merged with another company to soon after the last merger or the merger was not completed yet. There is however no official list of airline mergers because neither the U.S. Department of Transportation nor A4A maintains official records of airline M&A activity. This is why we used the combination of an unofficial compilation of completed mergers and acquisitions10 and a thesis written by Guo (2011) about European Airline mergers to obtain our data.
By using the databases DataStream and Yahoo-Finance we obtained all the stock data needed for our analysis. The stock data of all target firms in appendix 3 could however not be found with the two databases used. Also the stock data from the acquirer American Airlines in appendix 3 could not be used, because this data could only be obtained after the 5th of October in 2000. This made seven of the 26 mergers unusable for our analysis. The stock data for the other 19 mergers were completely obtained, which are shown in table 1.
Table 1: Acquirers and targets used in the analysis
Acquirer Target Date announced
Delta Airlines Comair 18-10-1999
Air Canada Canadian Airlines 21-12-1999
Air France Regional Airlines SA 19-1-2000
Air France Brit Air SA 19-6-2000
Austrian Airlines Lauda Air Lufhtfahrt AG 18-8-2000 American Airlines Trans world airlines 10-1-2001 British Airways PLC British Regional Airlines Grp 8-3-2001
SAS AB Braathens ASA 21-5-2001
Lufthansa Air Dolomiti SpA 6-3-2003
Air France Klm 30-9-2003
Alitalia Gandalf SPA 25-3-2004
Lufthansa Fraport AG 9-8-2005
Lufthansa Austrian Airlines AG 12-3-2008
10 12-6-2014: http://www.airlines.org/Pages/U.S.-Airline-Mergers-and-Acquisitions.aspx
12-6-2014: http://thepointsguy.com/2013/02/past-and-future-airline-mergers-a-brief-history-and-predictions/ 11-6-2014: http://en.wikipedia.org/wiki/List_of_airline_mergers_and_acquisitions
23
Delta Airlines Northwest Airlines 26-9-2008 Republic Airways holdings Frontier Airlines 14-8-2009 United Airlines Continental Airlines 3-5-2010
Southwest Airlines AirTran 27-9-2010
LAN TAM 19-1-2011
American Airlines US Airways 14-2-2013
All stock data from the firms in table 1 were obtained by using DataStream, except for the companies United Airlines and American Airlines. The stock data for those two firms were obtained by using Yahoo-Finance.
By using DataStream and Bloomberg11 we found a list of suitable main stock indices for the countries of the airlines used in this study. These main stock indices and their codes used for DataStream are shown in appendix 4.
In the fifth merger wave, between 1998 and 2001, there was a total loss of $240 billion (Moeller, Schlingemann & Stulz, 2005). This was because of a rather small percentage of extremely unprofitable mergers. Excluding just over 2% of the observations, shareholder wealth would have increased with acquisition announcements (Moeller, Schlingemann & Stulz, 2005). This fact would make the stock data during the 5th merger wave still usable for our analysis, when combining it with the stock data of the 6th merger wave. It would however still be interesting to look at the mergers from the 5th merger wave separately and compare them with the results in total. This is what we are going to do for the mergers between 1998 and 2001 in Table 1.
11 13-6-2014: http://www.bloomberg.com/- markets/indexes/
24
5. Results
By using the formulas explained in section 3.3.3 we calculated the normal return, the abnormal return and the cumulative abnormal return for 19 mergers in the airline industry. We got the data of the stock price of each company from DataStream. The codes we used in DataStream for each specific company can be seen in appendix 1 (acquirers) and appendix 2 (targets). The data for American Airlines could however not be obtained completely. The data for American Airlines could only be obtained from October 5th 2000 onwards, which gave an event window of only 90 days for the merger between American Airlines and Trans World Airlines for the event period (-1, 1). The merger between American Airlines and Trans World Airlines had an event window of 80 days for the event period (-10, 1). This event window should however still give an adequate understanding of the normal market parameters for American Airlines.
Next we calculated the Slope, Intercept, R2, Standard Error and Variance of the normal market parameters for the 38 companies. We did this for the event window (-200, -20), as can be seen in appendix 5. We however found an exceptionally large standard error for Frontier Airlines, for no direct reason. Where other firms had a standard error between 0.03 and 0.07, Frontier Airlines had a standard error of 0.25. Because such a high standard error would influence the analysis we removed the merger between Republic Airways Holdings and Frontier Airlines from the analysis. The average abnormal return and the cumulative average abnormal return will now be calculated over 18 mergers.
By using the slope and intercept of the normal market parameters on the market model we were able to calculate the normal return, abnormal return and the cumulative abnormal return during the period of interest (-1, 1) and (-10, 1) for the acquirers and targets. Because the returns for the days -1, 0 and 1 are the same in both periods of interest we only show the event period (-10, 1) in appendix 612. There was however the problem that not all the stock data that was needed could be obtained. For the period of interest (-1, 1) all stock data was obtained. However, between day -2 and -10 there were some stock data missing, which is shown in appendix 6. This will not present a problem with calculating
12 Because the event window for the merger between American Airline and Trans World Airline is different between the two periods of interest, these are the only two companies that have a different variance and normal return during the days -1, 0 and 1. For this reason, the results for the period (-1, 1) will be given in Appendix 7.
25
the average cumulative abnormal return and calculating the daily average abnormal return. This does however give us a smaller sample for each separate day.
From all the daily abnormal returns and cumulative abnormal returns in appendix 6 and 7, the average abnormal return and the cumulative average abnormal return can be calculated. By calculating the daily average abnormal return for period of interest (-1, 1), with the formulas in section 3.3 and 3.4 we gained the following table:
Table 2: Daily Average Abnormal Return (-1, 1)
Daily Average Abnormal Return %
Day Acquirer Target Combined
-1 -0.827% -0.045% -0.436% t-value -1.136 -0.048 -0.73 0 0.273% 2.170%** 1.222%** t-value 0.375 2.278 2.04 1 -0.483% 0.359% 0.006% t-value -0.663 0.377 0.01 Total return -1.037% 2.484% 0.792% *, **, *** Significance at the 10%, 5% and 1% level, respectively
The results in table 2 have been arrived at by taking the average of all 18 bidding/target firms' abnormal returns for day -1, 0 and 1. Also the average abnormal return of all 36 companies together is calculated in the third column ‘combined’. Next a t-test is done by using the formulas explained in section 3.4, which gave us the t-value. Of all nine Average Abnormal Returns calculated in this analysis only two are statistically significant.
By using the information obtained in appendix 6 and 7 we can also calculate the Cumulative Average Abnormal Return and if they are statistically significant for the period of interest (-1, 1):
Table 3: Cumulative Average Abnormal Return (-1, 1)
Cumulative Average Abnormal Return (%)
Period of interest Acquirer Target Combined
CAAR(-1,+1) -1.037% 2.483% 0.723%
t-value -0.823 1.51 0.70
*, **, *** Significance at the 10%, 5% and 1% level, respectively
We can do the same for the period of interest (-10, 1), which results in the following tables for the Daily Average Abnormal Return and the cumulative average abnormal return. Because not all stock data for all 12 days could be found, we took the average of the mergers that did have results, for that specific day, to calculate the daily average abnormal return.
Table 4: Daily Average Abnormal return (-10, 1)
Daily Average Abnormal Return (%)
Day Target Acquirer Combined
1 0.332% -0.479% -0.074% t-value 0.351 -0.657 -0.12 0 2.146%** 0.277% 1.211%** t-value 2.269 0.380 2.03 -1 -0.050% -0.824% -0.437% t-value -0.053 -1.129 -0.73 -2 1.804% 0.547% 1.175% t-value 1.243 0.519 1.31 -3 -0.553% -0.061% -0.307% t-value -0.307 -0.038 -0.25 -4 0.704% -0.862% -0.079% t-value 0.641 -0.942 -0.11 -5 -0.080% -0.007% -0.043% t-value -0.073 -0.009 -0.07 -6 0.174% -1.184% -0.505% t-value 0.179 -1.562 -0.82 -7 -0.080% -1.007% -0.543% t-value -0.084 -1.379 -0.91 -8 2.499%** -0.339% 1.080% t-value 2.112 -0.376 1.45 -9 1.371% 1.638% 1.505%* t-value 0.945 1.554 1.68 -10 0.345% 0.425% 0.385% t-value 0.344 0.478 0.57 Total return 8,662% -1.876% 3.368% *, **, *** Significance at the 10%, 5% and 1% level, respectively
Table 5: Cumulative Average Abnormal Return (-10, 1)
Cumulative Average Abnormal Return (%)
Period of interest Acquirer Target Combined
CAAR(-10,+1) -2.307% 6.463%** 2.078%
t-value -0.913 1.97 1.00
*, **, *** Significance at the 10%, 5% and 1% level, respectively
Bidding firms
From Table 2, it can be seen that the acquiring firms experienced an insignificant negative abnormal return of -0.827% one day before the merger. On the announcement date itself, target firms earned a positive abnormal return of 0.273% which was also not statistically significant. On the day after the announcement there also was a statistically insignificant negative abnormal return of -0.483%. In Table 3 we find an insignificant negative cumulative abnormal return of -0.421% during the period of interest.
When analyzing the period of interest (-10, 1), there is not one statistically significant result for the daily average abnormal return in Table 4. In Table 5 we also find an insignificant negative cumulative abnormal return of -2.307% for the shareholders of the bidding firms. This means that there is no statistical evidence that shareholders of the acquiring firm got a positive return during the merger announcement.
This is not in line with Knapp (1990) who concluded that acquirers earned an excess return between 6% and 12%. According to Knapp (1990) these significant positive abnormal returns for acquirers demonstrate that the airline merger market was less than competitive. Knapp (1990) investigates American airline mergers in 1986, when the first deregulation package in Europe still had to be implemented. Between 1986 and 2014, a lot of laws and deregulation packages have been implemented to enforce competition in the airline industry13. The differences in results could be because the airline market has become more competitive in the intervening period.
Singal (1996) analyzed airline mergers in the United States between 1985 and 1988. Singal (1996) concluded that bidding firms earn statistically significantly positive cumulative returns ranging from 1.32% to 1.88%. Because this is around the same time Knapp (1990) did his research, it is likely that the difference between our results and that of Singal (1996) is also because of increased competition in the airline market.
13For example the first deregulation package in Europe in 1987 and two more in 1990 and 1993.
28
Target firms
It can be seen in Table 2 above that the target firms experienced an insignificant positive abnormal return of 0.337% the day before the merger. On the announcement date itself, target firms earned a positive abnormal return of 2.220%, which is significantly different from zero at a 5% significance level. One day after the announcement there was a statistically insignificant negative abnormal return of -0.763%. In Table 3 it can be seen that during the period of interest there was an insignificant positive cumulative abnormal return of 2.483% during the period of interest.
When analyzing the period of interest (-10, 1), we can see 2 statistically significant results in Table 4. Just like the period of interest (-1, 1), target firms earned a positive abnormal return of 2.146% on day 0. Further a statistically significant positive abnormal return of 2.499% was earned by the shareholders on day -8. In Table 5 we find a significant positive cumulative abnormal return of 6.463% for the shareholders of the target firms. That the results of the cumulative abnormal return in the period of interest (-1, 1) was not significant, but is significant in the period of interest (-10,1), is in line with Knapp (1990). Knapp (1990) concluded that most of the gains for the target firms were made in the 20 days before the merger announcement. This means that there is statistical evidence that shareholders of the acquired firm got a positive return during the merger announcement, for a period of interest between 10 days before and 1 day after the merger announcement. These results do also agree with Andrade, Mitchell & Stafford (2001) who revealed that when mergers create shareholder return, most of the gains are accrued by the target company (Andrade, Mitchell & Stafford, 2001).
According to Singal (1996) the target firms earn statistically significant positive abnormal returns ranging from 18.43% to 22.00% depending on the event period. The total return in our study for the period of interest (-10,1) is 8.662%, which is a lower than the results from Singal (1996), even though Singal (1996) had smaller periods of interest, ranging between 2 and 7 days. This could again be explained by a more competitive airline market.
Combined firms
The combined firms experienced an insignificant negative abnormal return of -0.289% one day before the merger, as can be seen in Table 2. On the day of the announcement the combined firms experienced a statically significant positive abnormal return at a 5% significance level of 1.531%. The day after the merger announcement there was an insignificant negative abnormal return of -0.555%. In Table 3 we find an insignificant positive cumulative abnormal return of 1.031% during the period of interest.
When analyzing the period of interest (-10, 1), we can see two statistically significant results in Table 4. Just like the period of interest (-1, 1), combined firms earned a positive abnormal return of 1.211% on day 0. Shareholders of the combined firms also earned a significant positive abnormal return of 1.505% on day -9. In Table 5 we find an insignificant positive cumulative abnormal return of 2.078% for the shareholders of the combined firms.
Singal (1996) concluded that the combined firms earn statistically significantly positive cumulative returns ranging from 3.157% to 4.193%, depending on the event period. In the period of interest (-10,1) we calculated a positive abnormal return of 3.368%, which completely fits Singal's (1996) conclusion. Our results are however over 12 days where the period of interest of Singal (1996) ranged between 2 and 7 days. When we compare the results of the period of interest (-1, 1) with each other, our study measured a positive abnormal return of 0,792% and Singal (1996) calculated a positive abnormal return of 4.193%. This difference in results could be explained by a more competitive airline market since 1985-1988, which were the years Singal (1996) investigated.
5th merger wave
By calculating the cumulative average abnormal return for both periods of interest for mergers between 1998 and 2001 we arrived at the following table:
Table 6: Cumulative Average Abnormal Return (-1, 1) and (-10,1) 5th merger wave 5th merger wave Cumulative Average Abnormal Return (%) Period of interest Acquirer Target Combined
CAAR(-10,+1) -5.563% 5.775% 0.106%
t-value -1.690 1.10 0.03
CAAR(-1,+1) -1.033% 3.593%** 1.280%
t-value -0.388 2.20 0.82
*, **, *** Significance at the 10%, 5% and 1% level, respectively
When comparing Table 6 with Table 3, the cumulative abnormal return for the shareholders of the bidding firm is almost the same and shareholders of the target and combined firms earned an even higher cumulative abnormal return. However, a contradiction arises when comparing table 6 with table 5. The target, bidding and combined shareholders have a lower cumulative abnormal return during the period of interest (-10, 1) in comparison to the period (-1, 1).This combination of results makes it impossible to say anything conclusive about the 5th merger wave.
This conclusion is however exactly the reason why we used the mergers of the 5th merger wave in combination with the 6th merger wave. Because only 2% of the merger observations during the 5th merger wave was responsible for the loss of $240 billion in shareholder return (Moeller, Schlingemann & Stulz, 2005), the stock data during the 5th merger wave can still be combined with the stock data of the 6th merger wave, without having consequences that influence the analysis.
6. Conclusion
To decide whether merger announcements in the North American and European airline industry have an effect on the stock price of the target and the acquiring firm we did an event study. We used the Market Model to calculate the Average Abnormal return and the Cumulative Average Abnormal Return over the periods of interests (-10, 1) and (-1, 1). We did this for a total of 18 mergers.
We calculated that in the period of interest (-1, 1) no statistically significant cumulative abnormal returns have been obtained. The combined firms had an insignificant positive cumulative abnormal return of 1,031%, the target firms had an insignificant positive cumulative abnormal return of 2,483% and the acquiring firms had an insignificant negative cumulative abnormal return of -0,421%. However, in the period of interest (-10, 1) the shareholders of the target firms earned a statistically significant cumulative abnormal return of 6.463%. This corresponds with Knapp (1990) who concluded that most of the gains for the target firms were gained in the 20 days before the merger announcement. The cumulative abnormal returns for the shareholders of the combined and bidding firms during the period of interest (-10, 1) were just like during the period (–1, 1) insignificant. We therefore cannot say that there was statistically significant positive return gained by the shareholders of the combined and bidding firms during the merger announcement.
We can however say that on the announcement day itself the shareholders of the target airline companies have a statistically significant benefit from the merger announcement. This is the case because on the announcement date itself, in the periods of interest (-10, 1) and (-1, 1), target firms earned a positive abnormal return of 2,146% and 2,220%, respectively. This is significantly different from zero at a 5% significance level. The combined return gained by the shareholders at the day of the merger announcement in the periods of interest (-10, 1) and (-1, 1) were also positive. There was a statistically significant positive abnormal return at a 5% significance level of 1,211% and 1,531%, respectively. We may speak of statistically significant higher shareholder return on the announcement date of the merger.
However, this study has some limitations. There was no official list of airline mergers because neither the U.S. Department of Transportation nor A4A maintains official records of airline M&A activity. This is why we used an unofficial compilation of completed mergers and acquisitions for this study, but there is always the possibility we missed mergers that did occur between 1998 and 2014. Further, the market model assumes that the abnormal return is jointly normally distributed and independent through time, which is not specifically checked for. We also assume the error term εit of the Market
Model to be zero, because the relation between the security and market return keeps unchanged. This study assumes the market to be rational, which means that effects of an event will be shown in the market prices in a relative short time period.
When comparing the results of this thesis with those of Knapp (1990) and Singal (1996), the most important difference is that our study measures lower abnormal returns. According to Knapp (1990) these significant positive abnormal returns for acquirers demonstrate that the airline merger market was less than competitive. Between 1986 and 2014, a lot of laws and deregulation packages, like the first deregulation package in Europe in 1987, have been implemented to enforce competition in the airline industry. These laws and deregulation packages, together with our findings of lower abnormal returns could implicate a more competitive airline market in comparison to 20 years ago.
