Is event study a powerful tool to predict M&A performance based on accounting measure? : a comparison between stock market and accounting data using China data

(1)

MSc Economics

Track: Industrial Organisation, Regulation and Competition Policy

Master Thesis

Is event study a powerful tool to predict M&A

performance based on accounting measure? A

comparison between stock market and

accounting data using China data

by

Xiaoyin Hou

10881069

Supervisor

(2)

Abstract

This thesis presents an empirical test about the predictive power of event studies on the actual performance of mergers and acquisitions (M&A) based on accounting measures with evidence from China. I use a sample of 121 domestic mergers during 2008-2013 to estimate the Pearson correlation coefficient between a measure of performance based on stock market and one based on accounting data. Overall, for Chinese companies, the ability of event studies is very poor at capturing ex-post M&A effects derived from accounting data.

1. Introduction

The topic of how mergers and acquisitions (M&A) affects corporate companies’ performance has been a leading question since the 1960s (Meglio and Risberg, 2011). However, as per Zollo and Singh (2004), heterogeneity still exists in both the definition of M&A performance and in its measurement. This divergence is often argued as driven by a lack of consensus on how to measure performance. Scholars have applied various metrics to evaluate mergers and acquisitions’ impact, and there are mainly three criteria. (Papadakis and Thanos, 2010). The first one employs accounting rate of returns. Many studies assess the merger effect by comparing the performance of the merging firms with a benchmark group of firms and calculating the difference between pre- and post-merger rate of returns. The second one uses stock market’s reactions to unexpected M&A announcement as a measure to predict M&A performance. This approach is called event study. The third criteria relies on managers’ subjective assessments of mergers and acquisitions.

This wide range of alternative performance metrics and the lack of consistency from those metrics lead researchers face a dilemma of choosing an appropriate measure. In order to understand the discrepancies, some researchers employ

(3)

multiple criteria to measure the comparability of them. But comparing to the piles of work on measuring M&A performance, the work on the comparison is limited (Healy et al., 1992; Sirower and O’Byrne, 1998; Ghosh, 2001; Schoenberg, 2006; Zollo and Meier, 2008; Duso et al., 2010; Papadakis and Thanos, 2010). Hence I intend to add evidence on the comparison work by examining the relationship between measures based on financial and accounting data. Particularly, as the financial criteria is an ex-ante measure while the accounting-based criteria is an ex-post one, it makes sense to hypothesize that one is predictive for the other. Thus my research question is: Is event study a powerful tool to predict M&A performance based on accounting measure?

To answer this question, I employ a widely-used database in China called WIND and collect a sample of 121 domestic mergers and acquisitions from China to estimate 1) performance expectations of merging firms from stock market by means of event studies, 2) merger effects on accounting returns in merging firms up to 2 years post-merger, 3) the power of event study in predicting M&A performance based on accounting returns by Pearson correlation coefficient test. My work duplicates part of Duso, Gugler and Yurtoglu (2010) and Papadakis and Thanos (2010). In Duso et.al, they study the correlation between two methods on both merging firms and their competitors while my work only focus on merging firms. In Papadais and Thanos, they study domestic Greek mergers while mine is from China. I extend their work by specializing my research question to the type of merger (vertical, horizontal, conglomerate) and the type of industry (manufacturing and service).

The contributions of my paper over previous works are the following:

(a) I bring empirical evidence from China. Previous similar work focuses mainly in USA, UK and Europe. My result from a different continent can help to overcome the so called ‘overwhelming geographical bias’ in the field of management (Pettigrew, Thomas and Whittington, 2002). It may also provide

(4)

evidence for scholars to gain new lights about previous works regarding ‘culture’ or ‘political system’.

(b) For Chinese scholars, it is very important to find a proper way to understand the M&A performance now. China is witnessing a wave of mergers and acquisitions both domestic and international. As per WIND, one of the most used financial data source in China, till the end of 2014, M&A has exceeded $1000 billion and the number of completed one was around 1700. The number of M&A related work also increases, however most of them still applies accounting-based measurement. As this one dimension of approach usually leads to different result which will be discussed in detail in section 2.3.2, a complementary methodology should be considered. Due to the fact that financial data is easy to obtain, and that it is pervasive in M&A performance researches, event study is an ideal option. However many scholars hesitate to implement this method in their research because they all aware that Chinese financial market suffers several deficiencies. I think my work can help researchers to understand the similarities and differences of these two methodologies. It may serve as a hint to their choice of the methods, and hopefully bring some light on future work about “how to evaluate mergers and acquisitions performance”.

My thesis is organized as follows: Section 2 provides a literature review of the two methods and their comparison. Section 3 describes my data. Section 4 discusses my results. And Section 5 is the conclusion.

2. Literature review

This section summaries the previous literature from three strands: 1) discussions on event study methodology; 2) discussions on accounting based measurement; and 3) previous researches on the two methods and the comparison of these two.

(5)

2.1 Event study

Event study, a method widely used in financial researches, reflects the market expectation about an unexpected announcement on the value the related corporate company.

If financial markets are efficient and unbiased, any influence caused by any events (in my paper, the mergers and acquisition announcements) on a firm’s discounted profits will be reflected in the firm’s stock price (S. Davies and P. Ormosi, 2010). Thus, the magnitude of the price abnormal behavior can be considered as a measure of market expectation of the event impacts on the firm’s wealth. This mesure is called abnormal return, which is calculated by subtracting the “normal” return – the return generated by a particular market model if the merger had not happened - from the real return of the respective company.

Hence, the main issue for implementing an event study is to find a proper counterfactual if there were no such unexpected events. What is a proper construction of this counterfactual remains another heated debate. Since the introduction of event study, scholars have developed more and more sophisticated models to estimate the expected return. I choose the most commonly used model in previous event study works – the market model – to predict the expected returns. Under the assumption of efficient market hypothesis (EMH), the market model predicts that returns on firm i at day t ( 𝑅𝑖𝑡) is proportional to the market

portfolio returns (𝑅_𝑚𝑡) at the same day:

𝑅_𝑖𝑡 = 𝛼_𝑖 + 𝛽_𝑖𝑅_𝑚𝑡+ 𝜖_𝑖𝑡

𝛼𝑖 and 𝛽𝑖 are the market model’s parameters. The estimated value (𝑎𝑖, 𝑏𝑖)

of the two parameters are calculated using Scholes-Williams method.

Using the above formula, I can estimate the counterfactual of stock return had no mergers or acquisitions were announced. Then I can calculate the abnormal returns for firm i at day t.

(6)

between the predicted return 𝑅̂_𝑖𝑡 and the observed real return 𝑅_𝑖𝑡: 𝐴𝑅_𝑖𝑡 = 𝑅_𝑖𝑡− 𝑅̂_𝑖𝑡 = 𝑅_𝑖𝑡− 𝛼̂_𝑖 − 𝛽̂_𝑖𝑅_𝑚𝑡

As the return of firm i may be influenced by the existence of information leakages about the event announcement, event studies always define a window of time frame around the event announcement date and use the sum of the abnormal returns within this window. This sum is called cumulative abnormal returns:

𝐶𝐴𝑅_{𝑖,𝑚,𝑛} = ∑ 𝐴𝑅_𝑖𝜏

𝜏=𝑛

𝜏=−𝑚

Where the window is defined as [−m, +n] , i.e. m days before the announcement day (day 0) and n days after the announcement day.

This methodology is pervasive among researchers. In leading journals, the number of published event studies already exceeded 500 in 2006 and continues to grow (S. P Kothari, J. Lewellen, and Jerold B. Warner, 2006). The widely adoption of event study is due to the fact that financial data is easy to obtain, and that stock prices are not as subject to manipulate as other indicators (McWilliams and Siegel, 1997).

However, despite the fact that most modern papers installed this method, skeptics on the approach remains. One major concern about the method is about the fundamental hypothesis of event study: is the financial market efficient? Numerous empirical evidence has shown that financial market is unlikely to be even “semi-efficient” all the time (Fama and French, 2007; Shleifer, 2000; Zajac and Westphal, 2004). The absence of market efficiency negatively influences the ability of financial market to predict future values, which can lead event study to incorrect and erroneous conclusions (Hendricks and Singhal, 2001). It may be necessary to supplement event study with other measures to accurately understand an event’s economic impact.

Second concern of event study is the appropriate design of an event. According to McWilliams and Siegel (1997), one very significant criteria of

(7)

designing the event is to make sure no compounding impacts on the firms’ value in an event window. However, there is no consensus among scholars on what is an appropriate window that contains all relevant information of the event but excludes other noise. Searchers usually select the event window based on previous works. With these concerns, it is of vital importance to associate event study with additional empirical evidence when measuring the M&A performance. One widely accepted method is the accounting –based measurement.

2.2 Accounting measurement

There are many metrics that have already been applied as indicators to measure M&A impact on acquiring firms’ profitability like return on sales (ROS), return on equity (ROE) and return on asset (ROA). As per Meeks and Meeks (1981), ROA is the most appropriate indicator for M&A financial performance measurement. The reason is that this ratio is the least sensitive to estimation bias induced by changes in leverage or bargaining power caused by mergers and acquisitions. Many existing literatures apply ROA in their work for corporation M&A analysis (Ramaswamy, 1997; Sudarsanam, 2003; Zollo and Singh, 2004). The basic idea behind the accounting measurement to evaluate M&A profit performance is quite similar to that of the event study, i.e. construct a counterfactual of return on asset had the merger or acquisition had not occurred, and compare it with the real ROA. The merging effect on acquiring forms profitability is the difference between the real and predicted ROA.

In my paper, I follow the steps implemented by Gugler et al. (2003) to predict the profits without mergers. The difference between my work and theirs is that they use profits directly while I decide to follow Meeks’ advice to use ROA. The first step is to calculate the projected change in ROA of the industry G from year t − 1 to t + n where acquiring firm I is categorized:

∆_{𝐼𝐺,𝑡−1,𝑡+𝑛}=Π𝐼𝐺,𝑡+𝑛 Κ𝐼𝐺,𝑡+𝑛

−Π𝐼𝐺,𝑡−1 Κ𝐼𝐺,𝑡−1

(8)

Where t is the year when the merger or acquisition is completed. Π_{𝐼𝐺,𝑡+𝑛}, Π_{𝐼𝐺,𝑡−1} are defined as the profit of industry G n years after and 1 year before the merger; K𝐼𝐺,𝑡+𝑛, Κ𝐼𝐺,𝑡−1 are the capital of industry G in the respective years.

This change calculates how ROA of industry I evolves during year t − 1 to t + n, and the assumption here is that firm G profitability will resemble the behavior of its industry. For example, if the industry earns a 0.10 return on asset in year t − 1, and 0.11 in year t + 2, I will assume the change of ROA of firm G is also +0.01 during the same period years.

Define ∆𝐼𝐷,𝑡−1,𝑡+𝑛 analogously for industry D where the target firm D is

categorized from year t − 1 to t + n. Now I can compute the predicted profit Π_{𝐶,𝑡+𝑛}𝑒𝑥𝑝𝑒𝑐𝑡 of the combined firm C in year t + n:

Π_{𝐶,𝑡+𝑛}𝑒𝑥𝑝𝑒𝑐𝑡 = Π𝐺,𝑡−1+ Κ_{𝐼𝐺,𝑡+𝑛} Κ𝐼𝐺,𝑡−1 Κ𝐺,𝑡−1Δ𝐼𝐺,𝑡−1,𝑡+𝑛+ Π𝐷,𝑡−1+ Κ_{𝐼𝐷,𝑡+𝑛} Κ𝐼𝐷,𝑡−1 Κ𝐷,𝑡−1Δ𝐼𝐷,𝑡−1,𝑡+𝑛

Π_{𝐺,𝑡−1}, K_{𝐺,𝑡−1} are profits and capital of the acquriing firm G in year t − 1, Π_{𝐷,𝑡−1}, Κ_{𝐷,𝑡−1} are profits and capital of the target firm D in year t − 1. This formula predicts the profit of combined firm C in year t + n to be the sum of profits of original bid firm G in year t − 1, plus the predicted growth in profit of firm G from year t − 1 to t + n, pulse the profits of the target firm D in year t − 1 ,plus the the predicted growth in profit of firm D from year t − 1 to t + n.

Hence I can calculate the predicted ROA of the combined firm C in year t + n as by deviding the expected profits of combined firm C over its capital in year t + n :

𝑅𝑂𝐴_{𝐶,𝑡+𝑛}𝑒𝑥𝑝𝑒𝑐𝑡𝑒𝑑 =Π𝐶,𝑡+𝑛

𝑒𝑥𝑝𝑒𝑐𝑡𝑒𝑑

Κ_{𝐶,𝑡+𝑛}

The merging effect in profits for the combined firm C is the change in the expected ROA and the actual ROA:

Δ𝑅𝑂𝐴_{𝐶,𝑡+𝑛}𝑒𝑓𝑓𝑒𝑐𝑡 = 𝑅𝑂𝐴_{𝐶,𝑡+𝑛}𝑎𝑐𝑡𝑢𝑎𝑙_{− 𝑅𝑂𝐴} 𝐶,𝑡+𝑛 𝑒𝑥𝑝𝑒𝑐𝑡𝑒𝑑

(9)

The merit of the measure based on returns is that this indicator evaluates the ex-post performance. Unlike event study, the change in ROA measures actual economic impact derived by the merger and acquisition. If the event really create value to the firm, the firm’s operating performance should eventually reflect this growth with a positive change in ROA.

However, the accounting measure suffers from the following problems. First, this methodology is criticized for their deficiencies in guiding the shareholders to maximize their wealth. Ignoring the opportunity cost of capital, a positive gain or a high return on asset does not necessarily maximize the shareholders wealth because the gain may fall short of the rate shareholders could have gained in other equities and assets of similar risk (Yook, 2004). Second, the data fails to examine the success of the particular mergers or acquisitions as it is an aggregated measurement for the profit performance (Lubatkin, 1983; Bruton, Oviatt and While, 1994; Chenhall and Langfield Smith, 2007). Third, accounting measures are running a risk of being manipulated. It is well-known that corporations are creative in applying accounting techniques to avoid true and fair reflection of their financial position in publication.

2.3 Result of previous work

Existing literature applying the two methods offers mix results, so does the researches comparing two methods. This section gives a review of the results from these three aspects.

2.3.1 Results of event study

Starting with event study, scholars suggest that shareholders from the target firms obtain significant positive gains (Hitt et al., 1998; Delong 2001; Martinez-Jerez 2002; Sudarsanam and Mahate, 2003). But the findings of M&A impact on acquiring firms, as per Bruner (2002), is inconsistent and distributes rather

(10)

evenly. He studies 44 papers to conclude that almost one third of the paper reports value destruction (13), one third shows value conservation (14), and the rest one third reports value creation (17). For example, Ben-Amar and Andre (2006) use a sample of 327 Canadian transactions over 1998-2002 with a window of [-1, 1] and show significant positive CAR. Houston et al. (2000) study 64 bank mergers during 1985-1996 by a window of [-4, 1], and show a significant negative abnormal return for the acquiring banks.

2.3.2 Results of accounting-based measure

With respect to accounting-based measure, Mueller (1980) collects an unusual rich global dataset and he conclude two patterns of merging impact on the

profitability of bid firms: no consistent conclusion whether M&A will increase or decrease firms profitability, some firms show improvement while others show destruction; all the impacts on profitability are modest.

It worth mentioning that for researches regarding M&A researches in China, the majority of publications applies accounting related measurements due to the fact that before the reform in 2005, China stock market shows a sever deviation from efficient market hypothesis. For the works regarding Chinese mergers and acquisitions, the conclusions of improved post-merger performance, in line with what Mueller states, do not converge. Wang Meng (2010) uses 6 accounting measurements to find out that the performance of acquiring firms is improving in the beginning, then declining, and increasing again. Li Hanjun, Zhang li and Ai Jie (2013) studies Chinese listed domestic mergers with four accounting measurements and reports both improvement and decline of the post-merger performance in his sample. Zhang Yi, Qiao Yuanbo and He Xiaofeng (2015), on the other hand, finds no value creation for bid firms.

(11)

2.3.3 Results of comparison

In addition, not only the empirical works testing the M&A performance are discordant, researches focusing on comparing the differences between those measures also lack a consensus. Schoenberg (2006) work shows no clear relationship between the measures. Zollo and Meier (2008) hold the view that long-term performance are linked between accounting and financial measures, but the short-term event study is independent of all other metrics. This view is enhanced by Papadakis and Thanos (2010). In Papadakis and Thanos’ report, they investigate the comparability of the two methods using a sample of 50 domestic Greek mergers. They conclude that these two criteria are not correlated. On the contrary, Duso et al. (2010) show significant correlation between

accounting and financial measures. They collect 482 involving firms from 1990-2002 to apply the two criteria, and used pairwise correlation to test for

dependence. The coefficient is positive and highly significant

3. Data Description

I collect the data from the WIND (China’s leading financial data and solutions provider) by the following criteria.

1) The sample is chosen for those announced mergers or acquisitions in 2008 to 2013 and are completed by the end of 2013. WIND has M&A related data only since 2008. However, I evaluate this limit of the database would not influence my analysis. Chinese government published new China Accounting Standards which was implemented in 2007 in order to be in line with the international standards. But as the changes of standards were continuous, some business corporates probably had delayed their standard change implementation. Hence year 2008 is a proper starting point to avoid potential data structure inconsistency. The ending year is 2013 for the purpose of having a comparable sample which I am able to predict ex-post profits 2 years after the merger. One

(12)

might argue that 2 years were not sufficient to fully complete the integration process, thus leading my analysis negatively biased as some resources need more time to provide positive impact on the new firm. However, Papadakis and Thanos (2010) point out that many executives in their research regard 2-year period adequate to evaluate a merger or acquisition is successful or not. Choosing 2 years also makes sure I have as many observations as possible in WIND.

2) All events are domestic. This choice come again because of the database limit. Many companies involved in cross-border mergers and acquisitions lack the accounting data in the system.

3) All acquiring firm do not have other announcement two months before or after the merger announcement. The purpose is to exclude other event impact on stock prices.

4) The acquiring firm should have taken at least half of the shares of the target firm. This is to make sure that the merging firms have the dominant position in target firms or new ventures.

With these principles, I am able to identify 276 different firms either as acquiring or target firm from the system. As some firms are involved in several mergers and acquisition activities, the actual number of events announced is 147. Then I match this list with the balance sheet in the database. Due to missing relevant accounting measures for 26 observations, I end up with 121 events in my sample which met all the above criteria.

3.1 Advantage and disadvantage of my data

Before analyzing, I first discuss the advantages and disadvantages of my data. There are already several empirical works comparing these two measures using different regional data. However most of the data come from the U.S and Europe. To my best knowledge, there is no such empirical work that focuses only on the

(13)

Chinese listed companies. My work is quite new considering the country being discussed. One thing to mention is that my focus is not to find out if one method is better than the other. Rather I’m only interested in whether results from the U.S. and Europe carry over to China.

My data may also have various disadvantages. One relevant question is whether the Chinese stock market is efficient. The work of Groenewold, Tang, and Wu (2003) shows deviation in Chinese stock market from the market efficiency assumption. Burton G. Malkiel (2007) comes to the same conclusion. The work of Chen, Kim, Yao, and Yu (2010) argues that the predictability of Chinese stock is weaker than that of U.S.A. Nonetheless, Chong, Lam, and Yan (2012) claim Chinese financial market is more efficient since the structural inform in 2005-2006 by studying the trading strategies profitability. Latest evidence does not come to a conclusion. If the market efficiency assumption is not valid, as discussed in the second section, the results from event study are much less reliable.

Second, my sample only focus on domestic mergers and acquisitions. Scholars generally believe that cross-border M&A activities have some different attributes from domestic ones. Different cultural, economic and regulatory structure make international mergers and acquisitions unique from domestic ones (Hofstede, 1980; House et al., 2002). Those challenges make it much more difficult for cross-border acquiring firms to reach their customers, adjust and learn from the local market and realize their goals than the domestic ones. Hence cross-border firms faces in particular risks in ex-post stage (Shimizu et al., 2004). On one hand, considering the characteristics divergence of the two types, I find it necessary to focus on domestic and international M&A separately for the purpose of evaluating these two methods’ ability to capture other M&A prospects excluding cross-border impact. On the other one hand, in order to provide a profound comparison between two methodologies, I should also assess the ability of event study to predict and the accounting approach to capture these

(14)

multinational impacts. I agree with this argument, and agree that due to this sample limit, my work calls for further investigation. It is possible that one method performances better than the other one in certain circumstances while the other way around in other situation. While focusing on only domestic mergers enhances my conclusion about the two methods for domestic mergers, it weakens my conclusion for international mergers.

4. Results and Discussions

In this section, I present and discuss the results from my data analysis. Section 4.1 provides general descriptive statistics about my full sample, and defines the kinds of mergers by type and by industry for later sections. Section 4.2 demonstrates the results from Pearson correlation coefficient on the full sample. Section 4.3 divides the full sample as per their merger types and takes a deeper look into the predictive power of event study. Analogous to section 4.3, section 4.4 discusses correlation between the two methods by industry.

4.1 General description

Table 1 summarizes characteristics of domestic mergers completed between 2008 and 2013. I categorize the sample into three merging types: vertical, horizontal and conglomerate. To be defined as vertical mergers, all involving firms should be in the same supply chain, i.e. bidding firms purchase/sell items from the targets. To be defined as horizontal mergers, purchases or sales between companies in this merger must be in the same industry. Mergers that are neither vertical nor horizontal are categorized as conglomerate mergers [1]. Among these, vertical mergers take up 13.2%, horizontal mergers 50.4% and conglomerate mergers 36.4%.

(15)

Also, I classify the sample into manufacturing and service sectors as per the criteria Gugler et al. (2003) implemented. They apply Standard Industrial Classification (SIC) to their sample, then separate industries as manufacturing sector with SIC codes smaller than 4000 and larger than 4000 as service sector. Although the classification criteria are different between SIC and CSRC, fortunately in my sample all industries can find counterfactual in SIC. Hence I do not have problems to follow their rules. 71.9% of completed mergers has taken place in manufacturing sector and service sector make up of 28.1%.

Table 1 characteristics of M&A sample

All 2008 2009 2010 2011 2012 2013 Total 121 19 33 7 27 28 7 Vertical 14.0% 5.3% 24.2% 0.0% 14.8% 14.3% 0.0% Horizontal 51.2% 42.1% 45.5% 71.4% 51.9% 53.6% 71.4% Conglomerate 34.7% 52.6% 30.3% 28.6% 33.3% 32.1% 28.6% Manufacturing 71.9% 78.9% 60.6% 71.4% 74.1% 75.0% 85.7% Service 28.1% 21.1% 39.4% 28.6% 25.9% 25.0% 14.3%

4.2 Correlation of full sample

I choose the event window as per Duso et al. (2010), i.e. 2 days before and after the announcement date; five days before and after; 25 days before and 5 days after; Finally, 49 days before and 5 days after. Table 2 reports the mean, median, standard deviation of my full sample’s CARs (cumulative abnormal returns) based on the 4 choices of event windows.

(16)

A successful merger is defined when its CAR is positive. From the table, for all different window, means of CARs for the merging firm are slightly positive. And more than half of CARs are positive (56.2% - 62.0%). However, only the CAR mean from [-2, 2] window differs significantly from zero at the 10% level. Thus, in my sample, the overall market expectation of M&A on the merging firm is positive and partially significant. This result is in conformity with some previous studies which also reported insignificant expectation of improved performance about acquiring firms (Bhagat and Black, 2001; Choi and Russell, 2004), but is inconsistent with other works that reported either negative CARs (Zhang Xin, 2003) or significantly positive CARs of acquiring firms (Chi, Sun & Young, 2011).

Table 2 Preliminary statistics—abnormal returns all mergers

CAR[-2, 2] CAR[-5, 5] CAR[-25, 5] CAR[-49, 5]

Mean 0.0221 0.0070** 0.0178 0.1366 0.0202 0.3005 0.0202 0.4466 Median 0.0091 0.0187 0.0308 0.0250 St. Dev. 0.0884 0.1298 0.2124 0.2904 Positive 68 75 72 69 Obs. 121 121 121 121 Success rate 56.2% 62.0% 59.5% 57.0%

Notes: In the Mean row, I report means of cumulative abnormal returns from different event windows (first row) as well as their p-values (second row).

* Significance at the 10% level. ** Significance at the 5% level.

Table 3 summarizes the properties of ROAs derived from accounting data. The merging effect is calculated by subtracting expected ROA, which is introduces in Section 2.2, from the real ROA. From the table, the mean of merging effect is

(17)

insignificantly positive. As per Papadakis and Thanos (2010), if the merging effect is positive, we define this merger as successful, and vice versa. With both mean and median of the merging effect are positive, more than half of the mergers are successful. To be precise, my sample ends up with 78 successful cases, with a success rate of 64.5%. The conclusion is supported by the previous work shortly described in the literature review section (Li Hanjun, 2013; Wang Meng, 2010).

Table 3 Preliminary statistics—return on asset and merging effect all mergers

Expected ROA Real ROA Merging effect

Mean 0.0164 0.0464 0.0299

0.6209 0.0000** 0.3602

Median 0.0005 0.0363 0.0289

St. Dev. 0.3633 0.0414 0.3568

Notes:Merging effect is calculated by subtracting expected ROA form the real ROA. In the Mean row, I report means of expected ROA, real ROA and merging effect (first row) as well as their p-values (second row).

Here some similarities and difference of the two methods are already observable by the results obtained. First, all means and medians of CARs (with different event windows) have the same positive sign as the accounting based measure – merging effect, although only one window reports significant result. And more than half of the events are evaluated as successful by both methods. However, the successful rate from CARs is relatively lower than that calculated from accounting data. It seems that event study underestimates the profit effect comparing to accounting measure. Finally, the variability of the accounting measurement is much higher than the one based on stock prices.

(18)

To test the predictive power of event study on accounting based measure, I conduct a pairwise correlation analysis at the merger level. Table 4 summarizes the outcome.

Table 4 Pairwise correlations: all mergers

CAR[-2, 2] CAR[-5, 5] CAR[-25, 5] CAR[-49, 5] Merging Effect 0.1784 0.0503 0.0948 0.3011 -0.0469 0.6093 -0.1177 0.1983

Notes: I report pairwise correlation coefficients (first row) as well as their p-values (second row).

It is clear that although CARs and the differences between two ROAs give the same general results for mergers, their Pearson coefficients show no statistically significant correlation. The correlation coefficients for CAR [-2, 2] & merging effect and CAR [-5, 5] & merging effect are positive, but negative for the other two windows. The coefficients’ absolutes are at most 0.2, and none of them are significant at 10% level. In other words, event study, as an ex-ante measurement, does not capture the ex-post financial performance successfully indicated by accounting measurement, at least for my sample from China.

My finding is contrary to Duso et al. (2010) as they show positive and significant correlation between the abnormal return and ex-post merging effects. But my work is supported by Papadakis and Thanos (2010).

The different findings may be attributable to the fact that the research designs in Duso et al (2010), Papadakis and Thanos (2010) and mine are not identical to each other (national context, merging type, industry effects).

(19)

assumption varies between the stock markets in regions that are discussed in the above three researches. Almost 25% of the events in Duso et al.’s sample involves listed companies in the U.S. and Canada, 60% are mergers between European corporations, and 15% has companies either as target or bidder from the rest of the world. In a world, their sample represents a global view for mergers and acquisitions with a bias to European companies. On the contrary, Papadakis and Thanos’ sample focuses only on Greece domestic mergers. My sample, as emphasized, contains merging activities only from China. Samitas (2004) shows the efficient market hypothesis (EMH) is rejected in Athens Stock Exchange. Following his work, Borges (2008) tests the EMH in six European countries, namely France, Germany, UK, Greece, Portugal and Spain. Greece and Portugal reject the hypothesis while the rest four countries hold. But she also points out that the behavior of daily data in Greece and Portugal is approaching a random walk after 2003. With respect to China, as mentioned in the previous sector, whether Chinese financial market is mature is questionable. So many variables can influence the stock price that prices are usually not able to reflect companies’ performance. Markets also show improvement in efficiency after the reform, but as long as there is dispute, whether to accept EMH remains questionable. However, stock markets in the U.S. and Canada are usually considered as mature and efficient. The difference between these markets’ efficiency deteriorates the ability of event study to correctly predict the discounted future value in the corresponding country.

Second, with respect to the merging type, both Duso et al and Papadakis & Thanos study only horizontal mergers, while I have both horizontal, vertical and conglomerate mergers. To find if the merging type will affect the correlation between two methodologies, I calculate separately the CARs and merging effects for each merging type, and have three Pearson correlation coefficient analysis. It is notable that many scholars argue industry effect plays a role on the performance

(20)

of mergers. Hence, I also run the comparison for two large sectors I separated in section 4.1.

4.3 Correlation of different merger type

4.3.1 Horizontal mergers

Table 5 reports the mean, median, standard deviation of horizontal mergers’ CARs based on different event windows. Again, for all different window, means and medians of CARs are slightly positive. More than half of CARs are positive (53.2% - 61.3%). Only the CAR mean from [-2, 2] window is significant from zero at 10% level. It is not surprising to find the results of horizontal mergers resemble those of the full sample as they make a substantial part of all M&A activities.

Table 5 Preliminary statistics—abnormal returns horizontal mergers

CAR[-2, 2] CAR[-5, 5] CAR[-25, 5] CAR[-49, 5]

Mean 0.0174 0.0681* 0.0068 0.7214 0.0166 0.5798 0.0013 0.9736 Median 0.0098 0.0188 0.0362 0.0103 St. Dev. 0.0732 0.1481 0.2329 0.3003 Positive 37 38 38 33 Obs. 62 62 62 62 Success rate 59.7% 61.3% 61.3% 53.2%

I summarize the properties of accounting based measures for horizontal mergers in Table 6. The mean and median of the merging effect are slightly

(21)

positive. There are 37 successful horizontal mergers out of 62, with a successful rate of 59.7%.

Table 6 Preliminary statistics—return on asset and merging effect horizontal mergers

Mean 0.0243 0.0493 0.0250

0.5326 0.0000** 0.5065

Median -0.0014 0.0363 0.0287

St. Dev. 0.3026 0.0395 0.2921

Hence for horizontal mergers, CARs and the difference between two ROAs lead to very similar conclusion. More than half of the mergers can be considered as successful, but the positive effect of mergers on firm value is rather small.

Table 7 summarizes the outcome of pairwise analysis for the two methods with only horizontal mergers. Again, although event study and accounting measurement lead to similar results about companies’ performance with horizontal mergers, their Pearson coefficient does not show correlation. The correlation coefficients between window CAR [-49, 5], CAR [-5, 5], CAR [-25, 5] and merging effect are positive, but negative regarding CAR [-49, 5] and the merging effect. The coefficient absolutes are at most 0.14, and all are insignificant from zero at 10% significant level. Because this subsample only contains horizontal mergers, my data structure is the same as Papadakis and Thanos (2010). And my conclusion is exactly the same as their work: for domestic horizontal mergers,

(22)

CARs and accounting measures does not related to each other.

Table 7 Pairwise correlations: horizontal mergers.

CAR[-2, 2] CAR[-5, 5] CAR[-25, 5] CAR[-49, 5] Merging Effect 0.1397 0.2788 0.0763 0.5554 0.0017 0.9898 -0.1024 0.4285

4.3.2 Vertical mergers

Table 8 reports the mean, median, standard deviation of vertical mergers’ CARs based on different event windows. This time, all means from the four event windows are slightly negative but insignificant. Despite the median of window [-49, 5], all medians are also negative. Windows [-2, 2], [-5, 5] and [-25, 5] report more than half of the mergers as failures (successful rate 35.3% - 41.2%). Only the event window of [-49, 5] shows a 58.8% successful rate. Financial market does not show their faith in the acquiring firms involved in vertical mergers

Table 8 Preliminary statistics—abnormal returns vertical mergers.

CAR[-2, 2] CAR[-5, 5] CAR[-25, 5] CAR[-49, 5]

Mean -0.0124 0.9203 -0.0128 0.5136 -0.0265 0.9111 -0.0178 0.4594 Median -0.0290 -0.0474 -0.0253 0.0130 St. Dev. 0.0658 0.0809 0.1798 0.2555 Positive 6 7 7 10 Obs. 17 17 17 17

(23)

Success rate 35.3% 41.2% 41.2% 58.8%

Table 9 summarizes the properties of accounting based measures for the same type of mergers.Both mean and median of the merging effect are positive, and the mean is significant. There are 11 successful vertical mergers out of 17, leading to a successful rate as high as 64.7%.

Table 9 Preliminary statistics—return on asset and merging effect vertical mergers

Mean -0.0674 0.0447 0.1121

0.2515 0.0023** 0.0430*

Median 0.0213 0.0276 0.0200

St. Dev. 0.2264 0.0494 0.2040

It is quite interesting to notice that for vertical mergers, event study and accounting measure lead to opposite conclusions. On average, financial market is pessimistic about vertical mergers, while the accounting data shows many vertical mergers do provide additional value to the acquiring firms. Still, I compare the two methods with a pairwise analysis in table 10.

(24)

Table 10 Pairwise correlations: vertical mergers.

CAR[-2, 2] CAR[-5, 5] CAR[-25, 5] CAR[-49, 5] Merging Effect -0.4019 0.1098 -0.3456 0.1742 -0.1218 0.6414 -0.0590 0.8220

It is not surprising to see negative correlation coefficients for all windows. This results reveal that the financial market in China is very poor at predicting vertical mergers. To my best knowledge, only Yu Chunhui (2002) applies event study to analyse vertical mergers in China. In his work, the CAR for vertical merger is significantly positive. However, the data is before 2005, the year when Chinese stock market reformed, and his sample only contains 10 events. Those restrictions lead the conclusion of his work less convincing now. Other existing researches of Chinese vertical mergers all implement accounting data to investigate the performance. The results from these works are similar to my conclusion from merging effect. For example, Li Lei and Song Zhiguo (2009) study mergers in 2005 and conclude that vertical mergers create values. Hence for vertical merger in China, I conclude that it is more reliable to apply accounting data than the financial data.

4.3.3 Conglomerate merger

Finally Table 11 reports the mean, median, standard deviation of conglomerate mergers’ CARs based on different event windows. All means of different windows are positive. However, they vary from the significance. The mean of event window

(25)

[-2, 2] is significantly positive from zero at 5% level, the means of event window [-5, 5] & [-25, 5] are significant and positive at 10% level, while the mean of [-49, 5] is positive but insignificant. The range of successful mergers is 59.5% - 71.4%. In all, financial market expect conglomerate mergers to create value for the acquiring firms.

Table 11 Preliminary statistics—abnormal returns conglomerate mergers.

CAR[-2, 2] CAR[-5, 5] CAR[-25, 5] CAR[-49, 5]

Mean 0.0431 0.0048** 0.0463 0.0747* 0.0443 0.0685* 0.0636 0.2085 Median 0.0156 0.0306 0.0170 0.0604 St. Dev. 0.1091 0.1100 0.1880 0.2836 Positive 25 30 27 26 Obs. 42 42 42 42 Success rate 59.5% 71.4% 64.3% 61.9%

Table 12 summarized the properties of accounting based measurement for conglomerate mergers. Both mean and median of the merging effect are positive, but very close to zero. There are 30 successful mergers, and the successful rate reaches 71.4%. The rate is identical to the highest successful rate derived from event study but the mean is much smaller than any of the CARs. It seems to indicate that financial market has more extreme expectations of creating and losing value in conglomerate mergers than the actual performance of those companies. However, a relatively larger deviation of merging effect indicates that although in

(26)

the big picture the actual performance of acquiring firms is moderate around zero, there is some super breakdown that financial market never predicts.

Table 12 Preliminary statistics—return on asset and merging effect conglomerate mergers

Mean 0.0387 0.0427 0.0039

0.6004 0.0000** 0.9577

Median 0.0015 0.0365 0.0553

St. Dev. 0.4700 0.0401 0.4695

In general, like horizontal mergers, CARs and merging effect lead to similar conclusion for the conglomerate mergers. Both measures see value creation overall for the acquiring firm.

For the pairwise analysis, the following table 13 summarizes the results. Window of [-2, 2] and the merging effect a significantly positive related at 5% level. However, for the rest, the correlation coefficients are insignificant. One noticeable finding is that the Pearson’s coefficient for merging effect and window [-25, 5], [-49, 5] are negative. From table 12, it can be seen that as the event window is longer, the means of CAR are increasing as well as the successful rates. It seems with a longer event window, the financial market is more optimistic about certain mergers. But the negative correlation between abnormal returns and merging effect implies that market overestimates the value created by those certain mergers. This results might indicate that extra information from longer window

(27)

can be “too much”.

Table 13 Pairwise correlations: vertical mergers.

CAR[-2, 2] CAR[-5, 5] CAR[-25, 5] CAR[-49, 5] Merging Effect 0.3055 0.0492** 0.2275 0.1474 -0.0730 0.6458 -0.1381 0.3830

4.4 Correlation of different industry

Industry effect is another variable that will impact the merger performance of acquiring firms. I’m also interested in investigating how financial indicator, CAR, and the accounting indicator, ROA, are related if I exclude the industry effect. As shown above, I divided the sample into manufacturing and service sectors following the same way as Gugler et al. (2003).

4.4.1 Manufacturing sector

Table 14 shows the CAR result for manufacturing sector. In general all means of CARs are positive, but only the result from event window [-2, 2] is significant from zero. The successful rate in the manufacturing mergers varies from 54.0% to 62.1%. The rate declines when the event window contains more days. It indicates that with more information, the financial market predicts mergers in manufacturing sector to be less profitable than they first believe

Table 15 shows the result derived from accounting data. With both mean and median of the merging effect are positive, although the mean of merging effect is insignificant. There are 55 successful cases, and the successful rate is 63.2%,

(28)

which is much similar to the successful rate of CAR with event window [-2, 2].

Table 14 Preliminary statistics—abnormal returns manufacturing mergers

CAR[-2, 2] CAR[-5, 5] CAR[-25, 5] CAR[-49, 5]

Mean 0.0324 0.0022** 0.0178 0.2659 0.0181 0.4812 0.0096 0.7689 Median 0.0118 0.0190 0.0354 0.0130 St. Dev. 0.0950 0.1473 0.2379 0.3012 Positive 54 52 50 47 Obs. 87 87 87 87 Success rate 62.1% 59.8% 57.5% 54.0%

Table 15 Preliminary statistics—return on asset and merging effect manufacturing mergers

Mean 0.0179 0.0460 0.0281

0.6640 0.0000** 0.4837

Median 0.0005 0.0285 0.0378

St. Dev. 0.3815 0.0426 0.3707

(29)

methodologies in the manufacturing sector. Again, only window of [-2, 2] has a significant coefficient with merging effect at 10% significant level. Altogether, for mergers in manufacturing sector, event study with a very short event window outperforms other window choice. It yields closer results to the accounting –based measure. However, since the coefficient of these two methodologies is at best 0.2, event study does not explain much about ex-post merging effects.

Table 16 Pairwise correlations: manufacturing mergers.

CAR[-2, 2] CAR[-5, 5] CAR[-25, 5] CAR[-49, 5] Merging Effect 0.2096 0.0513* 0.1164 0.2830 -0.0037 0.9727 -0.0830 0.4447

4.4.2 Service sector

Table 17 shows the CAR result for service sector. Event window [-2, 2] reports negative abnormal returns, while the other windows report positive CARs. However, no abnormal returns is significant from zero for all event windows. The successful rate of mergers in service sector ranges from 41.2% to 67.6%. The [-2, 2] window is more pessimistic about merging in service sector than the other windows. Very interestingly, the pattern of CARs in service sector is the opposite of those in the manufacturing sector. With longer window, financial market gains confidence in the acquiring firm in the service sector, turning its belief from losing to gaining profits.

(30)

Table 17 Preliminary statistics—abnormal returns service mergers

CAR[-2, 2] CAR[-5, 5] CAR[-25, 5] CAR[-49, 5]

Mean -0.0042 0.2075 0.0176 0.6115 0.0253 0.4558 0.0476 0.7227 Median -0.0053 0.0132 0.0146 0.0662 St. Dev. 0.0614 0.0662 0.1254 0.2588 Positive 14 23 22 22 Obs. 34 34 34 34 Success rate 41.2% 67.6% 64.7% 64.7%

The next table 18 shows the result derived from accounting data for mergers with acquiring firms in the service sector. Both mean and median of the merging effect are positive, but the mean is still insignificant. There are 23 successful cases, and the successful rate is 67.6%. It seems that accounting data reinforces the results from longer windows than the shortest one. Comparing to the manufacturing sector, mergers in service sector provides more profits, with a larger positive merging effect.

Table 18 statistics—return on asset and merging effect service mergers

Mean 0.0126 0.0472 0.0345

0.8173 0.0000** 0.5378

Median 0.0058 0.0435 0.0180

(31)

As usually, I conduct the pairwise comparison between the two measures. Table 19 illustrates the results. The coefficients for CARs and merging effect is, however, negative for windows [-5, 5], [-25, 5] and [-49, 5], and is significant with [-25, 5]. The only positive coefficient is with [-2, 2], but is insignificant. This table shows that although in general, ex-ante expectations from the financial market and the ex-post outcome all believe in a profit growth in the service merger, the financial market expects in a different reason for the growth from what accounting data indicates. It is much more like a coincident in my sample that two methods come to a same conclusion. Pearson correlation coefficient shows the two methodologies is not related.

This conclusion is very important because many literature in evaluating M&A performance uses accounting data as a robust test for the event study without investigating into the results relation. However, my sample shows that by simply implement another methodology to test the same problem may run into risk of getting a similar conclusion by accident. It is necessary to also run a pairwise analysis between those results to see if they explain reasons in the similar direction.

Table 19 Pairwise correlations: service mergers.

CAR[-2, 2] CAR[-5, 5] CAR[-25, 5] CAR[-49, 5] Merging Effect 0.0668 0.7074 -0.0308 0.8628 -0.2968 0.0883* -0.2412 0.1693

(32)

(second row).

Up till now, I analyse the subsamples by dividing the original one by merger type and industry. Results show that even excluding merger type and industry effect, the correlation between the two methodologies is very weak. It is reasonable to argue that, with these results, the two methods capture different aspects of M&A performance. Meglio and Risberg (2011) conduct a narrative literature review to support this view. They promote to regard the performance of mergers and acquisitions as a complex umbrella construct. Umbrella construct is an ambiguous concept to envelop and account for multiple sets of phenomena (Hirsch and Levin, 1999). Because an umbrella construct is ambiguous, there is no consensus about how to measure it. Hence the problem of lacking correlation between methodologies evaluating the M&A performance does not lie in the different construct measurement or the research design. Rather, it is due to the lack of clarifying which aspect of the construct indicators are measuring. The lack of consistency and significance of my work support this idea. By simply making pairwise analysis between two measurements, it is time to hold one step back and define clearly what these indicators measure of the concept.

5. Conclusion

In this thesis, I try to answer the question whether event study is a powerful tool to predict M&A performance based on accounting based measure by examining the correlation between these two methods. My work origins from previous work of Duso, Gugler & Yurtoglu (2010) and Papadakis & Thanos (2010). Duso, Gugler & Yurtoglu examine whether it is useful to apply event study for M&A performance analysis by comparing financial and accounting data. They use a

(33)

sample of 114 horizontal mergers during 1990 – 2002. Papadakis & Thanos use 50 domestic horizontal mergers in Greece during 1997 – 2003. The main difference between my work and theirs lies in two folds. First, my sample is domestic mergers in China. Second, my sample contains not only horizontal mergers, but also vertical and conglomerate ones. So, I extend their works by separating my sample as per merger types and industries and implementing the same analysis for the subsamples.

My findings of the full sample are quite similar to Papadakis & Thanos. The successful rate of M&A ranges around 60%, a bit higher than what Papadakis & Thanos claim in Greece (40% - 50%). But we all find no significant correlation between CARs and change in ROAs. My subsample of horizontal mergers reinforces my full sample findings and the conclusion of Papadakis & Thanos. All indicators of my horizontal mergers sample resemble the behavior of full sample. Findings from subsamples become fruitful when it comes to vertical mergers. Event study and accounting-based measure lead to opposite results in vertical mergers. It indicates that when analyzing vertical mergers in China, event study is not a very useful alternative tool to accounting data. Additionally, the correlation coefficients show that Chinese financial market cannot correctly predict the future success or failure of vertical mergers. I would advise to avoid applying event study for this kind of research, at least for the M&A in previous years considering Chinese stock market is not widely accepted as efficient. Conglomerate mergers, unlike horizontal or vertical mergers, shows strong correlation between stock market and accounting data with very short window. For long windows, event study tends to overestimate M&A performance, lowering the reliability of financial data. However, all my findings are different from Duso, Gugler & Yurtoglu (2010). Their evidence shows positive and highly significant correlation between the two methods, and they show that a long event window enhances the predictability of event study to capture the ex-post merging effect.

(34)

I extend the work by another dimension: dividing the sample into manufacturing and service sectors as per the acquiring firms’ industries. Event studies for these two sectors produce opposite results. For manufacturing sector, longer window predicts lower profits. Short window capture the accounting data information better than the long windows. Their correlation coefficient is significantly positive with short window, but the relatively low value reflects the poor ability of event study to capture post-merger performance evaluated by the accounting measure. For service sector, the CARs increase with longer windows. But I do not find significant correlation between CARs and change in ROAs. It is important to notice that although in service sector mergers, CARs and change in ROAs has similar results, the correlation coefficients are negative. This evidence indicates the ante expectation of event study does not capture the causes of ex-post merging effect. Their resembling conclusion can be considered as a coincidence. Considering many previous papers have implemented accounting based measure as a complementary test for the event study (or the other way around), I strongly suggest in the future scholars also test the results correlations of the two to avoid wrong conclusions.

In a word, the two methodologies show no correlation with each other. And scholars should work with caution when studying performance of M&As with multiple criteria. It is better to consider the performance of M&A as an umbrella concept, and define clearly what aspect of the performance the indicator intends to measure.

It is necessary to bear in mind that my work has several limitations. First, I focus only on domestic Chinese mergers, hence the results cannot be applied to other countries, nor cross-border M&A activities. Second, although managers suggest two years are sufficient for mergers integration process of the acquiring firm, it is better to extend my work with more than two years ex-post mergers and acquisition evidence. Third, I only compare two methods for evaluating M&A

(35)

performance without considering other criteria like managers’ personal assessments (Schoenberg, 2004; Brock, 2005; Homburg and Bucerius, 2006) and subsequent divestments (Bergh, 1997; Capron, Mitchell and Swaminathan, 2001; Schoenberg, 2006).

Thus, I would give the following suggestions for future work. First, I believe international mergers and acquisitions have different attributes from domestic ones. A comparison for cross-border M&As will provide important evidence on the correlation between the two methods. Second, comparison between other measuring criteria should be encouraged. It has been shown that similar results from different methodology may be caused by accidents. A better understanding of the correlations between these indicators can contribute to the reliability of works applying multiple criteria. Finally, as I shortly discussed, a lack of correlation does not necessarily suggest one method outperform another. Hence future works are strongly suggested to define clearly what aspect of the M&As performance their indicator measures.

6. References

Ben‐Amar, W., & André, P. (2006). Separation of ownership from control and acquiring firm performance: The case of family ownership in Canada. Journal of Business Finance & Accounting, 33(3‐4), 517-543.

Bergh, D. D. (1997). Predicting divestiture of unrelated acquisitions: An integrative model of ex ante conditions. Strategic Management Journal, 715-731.

Bhagat, S., & Black, B. (2001). The non-correlation between board independence and long-term firm performance. J. CorP. l., 27, 231.

Borges, M. R. (2010). Efficient market hypothesis in European stock markets. The

European Journal of Finance, 16(7), 711-726.

Brock, D. M. (2005). Multinational acquisition integration: the role of national culture in creating synergies. International Business Review, 14(3), 269-288.

(36)

Bruner, R. F. (2002). Does M&A pay? A survey of evidence for the decision-maker. Journal of Applied Finance, 12(1), 48-68.

Bruton, G. D., Oviatt, B. M., & White, M. A. (1994). Performance of acquisitions of distressed firms. Academy of Management Journal, 37(4), 972-989.

Capron, L., Mitchell, W., & Swaminathan, A. (2001). Asset divestiture following horizontal acquisitions: A dynamic view. Strategic management journal, 22(9), 817-844. Chen, X., Kim, K. A., Yao, T., & Yu, T. (2010). On the predictability of Chinese stock returns. Pacific-Basin Finance Journal, 18(4), 403-425.

Chenhall, R. H., & Langfield-Smith, K. (2007). Multiple perspectives of performance measures. European Management Journal, 25(4), 266-282.

Chi, J., Sun, Q., & Young, M. (2011). Performance and characteristics of acquiring firms in the Chinese stock markets. Emerging markets review, 12(2), 152-170.

Chong, T. T. L., Lam, T. H., & Yan, I. K. M. (2012). Is the Chinese stock market really inefficient?. China Economic Review, 23(1), 122-137.

Choi, J., & Russell, J. S. (2004). Economic gains around mergers and acquisitions in the construction industry of the United States of America. Canadian Journal of Civil Engineering, 31(3), 513-525.

Davies, S. W., & Ormosi, P. L. (2010). Assessing competition policy: Methodologies, gaps and agenda for future research.

DeLong, G. L. (2001). Stockholder gains from focusing versus diversifying bank mergers. journal of Financial Economics, 59(2), 221-252.

Duso, T., Gugler, K., & Yurtoglu, B. (2010). Is the event study methodology useful for merger analysis? A comparison of stock market and accounting data. International Review of Law and Economics, 30(2), 186-192.

Fama, E. F., & French, K. R. (2007). Disagreement, tastes, and asset prices. Journal of financial economics, 83(3), 667-689.

Ghosh, A. (2001). Does operating performance really improve following corporate acquisitions?. Journal of corporate finance, 7(2), 151-178.

(37)

Gugler, K., Mueller, D. C., Yurtoglu, B. B., & Zulehner, C. (2003). The effects of mergers: an international comparison. International journal of industrial organization, 21(5), 625-653.

Groenewold, N., Tang, S. H. K., & Wu, Y. (2003). The efficiency of the Chinese stock market and the role of the banks. Journal of Asian Economics, 14(4), 593-609.

Healy, P. M., Palepu, K. G., & Ruback, R. S. (1992). Does corporate performance improve after mergers?. Journal of financial economics, 31(2), 135-175.

Hendricks, K. B., & Singhal, V. R. (2001). The long-run stock price performance of firms with effective TQM programs. Management Science, 47(3), 359-368.

Hitt, M., Harrison, J., Ireland, R. D., & Best, A. (1998). Attributes of successful and unsuccessful acquisitions of US firms. British Journal of Management, 9(2), 91-114. Hirsch, P. M., & Levin, D. Z. (1999). Umbrella advocates versus validity police: A life-cycle model. Organization Science, 10(2), 199-212.

Hofstede, G. (1980). Motivation, leadership, and organization: do American theories apply abroad?. Organizational dynamics, 9(1), 42-63.

Homburg, C., & Bucerius, M. (2006). Is speed of integration really a success factor of mergers and acquisitions? An analysis of the role of internal and external relatedness. Strategic management journal, 27(4), 347-367.

Houston, J. F., James, C. M., & Ryngaert, M. D. (2001). Where do merger gains come from? Bank mergers from the perspective of insiders and outsiders. Journal of financial economics, 60(2), 285-331.

House, R., Javidan, M., Hanges, P., & Dorfman, P. (2002). Understanding cultures and implicit leadership theories across the globe: an introduction to project GLOBE. Journal of world business, 37(1), 3-10.

Kothari, S. P., Lewellen, J., & Warner, J. B. (2006). Stock returns, aggregate earnings surprises, and behavioral finance. Journal of Financial Economics, 79(3), 537-568. 李汉君 (Li Hanjun), 张丽 (Zhang Li), & 艾杰 (Ai Jie). (2013). 上市公司并购绩效实证研究. ( An empirical analysis of Chinese listed companies’ post-merger performance)

(38)

统计与决策, (12), 166-168.

李蕾 (Li Lei), & 宋志国 (Song Zhiguo). (2009). 基于因子分析法的我国上市公司并购绩效实证研究 (Empirical Analysis of M&A of Chinese Listed Companies Based on Factor Analysis Method). 技术经济与管理研究, 6, 14-20.

Lubatkin, M. (1983). Mergers and the Performance of the Acquiring Firm. Academy of Management review, 8(2), 218-225.

Malkiel, B. G. (2007). A random walk down Wall Street: the time-tested strategy for successful investing. WW Norton & Company.

Martínez-Jerez, A. (2002). Interaction between accounting and corporate governance: evidence from business combinations. Harvard Business School, mimeo.

McWilliams, A., & Siegel, D. (1997). Event studies in management research: Theoretical and empirical issues. Academy of management journal, 40(3), 626-657.

Meeks, G., & Meeks, J. G. (1981). Profitability measures as indicators of post-merger efficiency. The Journal of Industrial Economics, 335-344.

Meglio, O., & Risberg, A. (2011). The (mis) measurement of M&A performance—A systematic narrative literature review. Scandinavian journal of management, 27(4), 418-433.

Mueller, D. C. (1980). The determinants and effects of mergers: An international comparison (Vol. 24). Oelgeschlager, Gunn & Hain; Königstein/Ts.: Verlag A. Hain. Papadakis, V. M., & Thanos, I. C. (2010). Measuring the performance of acquisitions: An empirical investigation using multiple criteria. British Journal of Management, 21(4), 859-873.

Pettigrew, A., Thomas, H., & Whittington, R. (2002). Strategic management: the strengths and limitations of a field. Handbook of strategy and management, 3.

Ramaswamy, K. (1997). The performance impact of strategic similarity in horizontal mergers: evidence from the US banking industry. Academy of Management Journal, 40(3), 697-715.

(39)

capital market. Zagreb International Review of Economics and Business, 7(1), 23-38. Schoenberg, R. (2006). Measuring the performance of corporate acquisitions: An empirical comparison of alternative metrics. British Journal of Management, 17(4), 361-370.

Shimizu, K., Hitt, M. A., Vaidyanath, D., & Pisano, V. (2004). Theoretical foundations of cross-border mergers and acquisitions: A review of current research and recommendations for the future. Journal of International Management, 10(3), 307-353.

Shleifer, A. (2000). Inefficient Markets: An introduction to behavioural finance. OUP Oxford.

Sirower, M. L., & O'Byrne, S. F. (1998). The Measurement of Post ‐ Acquisition Performance: Toward a Value‐Based Benchmarking Methodology. Journal of Applied Corporate Finance, 11(2), 107-121.

Sudarsanam, S. (2003). Creating value from mergers and acquisitions: The challenges: An integrated and international perspective. Pearson Education.

Sudarsanam, S., & Mahate, A. A. (2003). Glamour acquirers, method of payment and post ‐ acquisition performance: the UK evidence. Journal of Business Finance & Accounting, 30(1‐2), 299-342.

王萌 (Wang Meng). (2011). 上市公司不同支付方式并购绩效分析. (Analysis of post-merger performance with different methods of payment) 财会通讯: 综合 (中), (2), 9-10. Xin, Z., & China Securities Regulatory Commission. (2003). Do Mergers and Acquisitions Create Value: Evidence from Chinese Listed Companies [J]. Economic Research Journal, 6, 1-10. (in Chinese)

Yook, K. C. (2004). The measurement of post-acquisition performance using EVA. Quarterly Journal of Business and Economics, 67-83.

Zajac, E. J., & Westphal, J. D. (2004). The social construction of market value: Institutionalization and learning perspectives on stock market reactions. American Sociological Review, 69(3), 433-457.

(40)

市公司并购绩效的经验与实证分析. (Lesson from Chinese mergers: an empirical analysis) 财经问题研究, 1, 60-66.

Zollo, M., & Meier, D. (2008). What is M&A performance?. The Academy of Management Perspectives, 22(3), 55-77.

Zollo, M., & Singh, H. (2004). Deliberate learning in corporate acquisitions: post‐ acquisition strategies and integration capability in US bank mergers. Strategic Management Journal, 25(13), 1233-1256.