The disciplinary motive of the market for corporate control
Evidence from Operating and Stock Returns
Johnny van Teijen Faculty of Economics University of Groningen Student number: 1427857 E-mail: vanteijen@hotmail.com Telephone: (06)17138563
Final version: July 2005
The disciplinary motive of the market for corporate control
Evidence from Operating and Stock Returns
Abstract
Traditional finance theory predicts that poor managerial performance is being disciplined by the market of corporate control. This mechanism is called the inefficient management hypothesis or the disciplinary motive. This paper re-examines the inefficient management hypothesis with a sample of UK target firms. I investigate both operational and stock price performance of takeover targets. I find evidence regarding the stock price performance of takeover targets that supports the view that poor performance is being disciplined by the market for corporate control. I report a statistically significant CAAR of -46.65% from month -60 to the month of the acquisition announcement. However, the results concerning the pre- acquisition operational performance of target firms are dispersed and insignificant. The operational performance of takeover targets decreases as the firms reach the year of the acquisition announcement but there are no significant differences with the corresponding control firms. Overall, I conclude that the data shows signs that support the theoretical prediction of the inefficient management hypothesis. Nevertheless, further research is needed to elaborate on the relationship between the long-term pre-acquisition operational and stock price performance of takeover targets.
Contents
1 Introduction 4
2 Review of the Literature: 9
2.1 Theories explaining merger activity 9
2.2 Empirical evidence from event studies 11
2.3 Modelling and measuring pre-acquisition performance 14 2.4 Interrelation with the inefficient management hypothesis 16
3 Methodology 18
3.1 Measurement of stock price performance 18
3.2 Measurement of operational performance 22
4 Data 26
4.1 Sample description 26
4.2 Description of the main variables 29
5 Empirical findings 31
5.1 Results stock price performance 31
5.2 Results operational performance 33
6 Concluding comments 36
References 38
Appendixes 41
Introduction
The traditional finance literature and many financial economists state that takeovers are partly motivated by the desire to improve poorly performing firms. Manne (1965) and Alchian and Demsetz (1972) lay the foundations for this conceptual framework. They state that the threat of a takeover increases for poorly performing companies due to potential gains for acquiring firms. The acquiring firms can realize those potential gains by replacing the management or restructuring the acquired firm. This theory explains the inefficient management hypothesis or the disciplinary motive for takeovers. The inefficient-management hypothesis suggests that mergers and acquisitions provide mechanisms to remove inefficient management of the target enterprise. In addition, the hypothesis suggests that the threat of takeover improves the performance of actual and potential target firms. The theory predicts that managers are aware of the fact that their performance is being judged and they try to limit the threat of a corporate takeover. Therefore, the theory predicts that companies become targets when they deliver abnormal poor performance. In this way the market for corporate control functions as an external control mechanism. The need for this external control mechanism indicates that there is some degree of failure in internal control mechanisms. Examples of internal control mechanisms are the board of directors, ownership concentration and executive compensation.
These control mechanisms have the goal to balance the interests of the multiple stakeholders.
Other conceptual frameworks for explaining mergers and acquisitions are the transactions cost efficiency theory of Coase (1937), synergy framework of Bradley, Desai and Kim (1983), agency cost theory of Jensen and Meckling (1976) and the managerial hubris theory of Roll (1986). All these theories provide theoretical explanations for the occurrence of mergers and acquisitions. I will discuss these theories in detail in the literature section.
Agrawal and Jaffe (2003) made a review of the existing empirical evidence regarding the disciplinary motive for takeovers. They state that the existing empirical evidence in support of this inefficient management hypothesis is rather weak. Besides this they found that most studies do not use the most recent methodologies that are available. Therefore they made a re- examination of the disciplinary motive in a large scale empirical study. They found little evidence that US target firms were performing poorly before acquisition, using either operating or stock returns. Therefore, the discussion whether targets firms deliver poor performance prior to the acquisition remains open.
In this paper I re-examine the view that targets perform poorly prior to the acquisitions. The goal of this paper is to investigate whether the inefficient management hypothesis is supported by the data and to examine the relationship between stock price performance and operational performance. A re-examination is needed due to the fact that previous studies do not make use of recent methodologies for investigating long-term pre-acquisition performance of target firms and the evidence is rather dispersed. This paper tries to find evidence for the general view that target firms experience underperformance prior to the acquisition announcement. I examine the inefficient management hypothesis with an empirical study based on stock price performance and operational performance. In addition, a comparison between the several techniques and results of earlier work will be made. My research question follows,
¾ “Provides the inefficient management hypothesis any explanatory power regarding the existence of mergers and acquisitions?”
Besides the main research question I try to answer the following two sub-questions,
¾ “Does the pre-acquisition stock price performance of target firms reflect their pre- acquisition operational performance?”
¾ “What are the appropriate methods and variables for measuring long-term pre- acquisition stock price performance or operational performance?”
The research question is further developed in the main hypothesis of this paper. The null- hypothesis (H0) of this paper follows:
H0: Target firms experience no significant pre-acquisition underperformance when measuring their stock price performance or operational performance against comparable firms.
The alternative hypothesis (H1) states that the market for corporate control disciplines poor management by means of a takeover when certain companies deliver poor stock price performance or operational performance. The alternative hypothesis is in line with the theory of the inefficient management hypothesis or the disciplinary motive of the market for corporate control.
H1: Target firms experience significant pre-acquisition underperformance when measuring their stock price performance or operational performance against comparable firms.
This paper makes use of the most current methodologies and examines the latest deals in the United Kingdom in order to provide a comprehensive empirical examination of the inefficient management hypothesis. It contributes to the existing empirical literature due to the use of modern methodologies, the use of recent mergers and acquisitions in a well-established capital market and the investigation of both pre-acquisition stock price performance and operational performance for the same firms.
I use several recommendations of Agrawal and Jaffe (2003) for investigating pre-acquisition performance in order to justify a valid comparison with their results. I took a sample from the UK stock market because of the fact that such a capital market is characterized by the same features as the US stock market. I took a sample of UK firms that were listed on the London Stock Exchange and were acquired between 1998 and 2004. I took a sample of quoted UK firms because of the wide availability of data and the fact that the London Stock Exchange is a well developed stock market. A well developed capital market creates the optimal conditions for the functioning of the disciplinary motive of takeovers since such markets are liquid, rational and prices adjust quickly to new information. I consider these facts as preliminary issues for a valid empirical investigation of the disciplinary motive for mergers and acquisitions.
In addition, I investigate both the long-run operating performance and stock price performance of takeover targets. This method is used because of the potential relationship between operational performance and stock price performance. A common belief is that bad operational performance should be reflected in the stock price performance. Traditional financial-economic theory assumes that capital markets are rational, efficient, complete and perfect. In such an idealized theoretical concept, stock prices reflect all available information and prices are set in a rational manner. In a rational and efficient capital market, poor operational performance should, at least partly, be reflected in the stock prices. In addition, investors judge the available information on its quality and set prices in a rational manner.
This implies that there should be a positive relationship between stock prices and earnings.
The theory of efficient capital markets predicts that I should find the same results when using operational data or stock prices for investigating the evidence for the inefficient management hypothesis.
Some theories predict that the relationship between stock prices and operational data is not as strong as advocated by the traditional finance literature. For instance, Behavioral finance theory argues that there are other factors besides fundamental aspects that determine stock prices1. The theory advocates that stock prices are not fully explained by their fundamental values. In this theory, investors are not fully rational and trade against arbitrageurs. The resources of arbitrageurs are limited due to risk aversion, short horizons and agency problems.
This brings along that there is a disparity between actual prices and fundamental prices which contradicts the efficient markets hypothesis. Shleifer (2000) reports several models in which he demonstrates that factors like the role of investor sentiment, limits to arbitrage and cognitive biases influence stock prices. This theory provides plausible arguments for a potential disparity between operational and stock price performance of target firms.
In addition, there are several papers which report evidence that demonstrate that the relationship between stock price performance and operational performance is not as strong as predicted by the efficient market hypothesis. The findings of these papers are in line with the Behavioral Finance Theory and report serious anomalies regarding the stock price reaction of earnings announcements. For instance, Sloan (1996) concludes that stock prices do not fully reflect all publicly available information. He investigates whether stock prices reflect information about future earnings through the accrual and cash flow components of current earnings. Penman and Zhang (2002) investigate the quality of earnings and the corresponding stock market reaction. They examine whether accounting methods affect the perceived quality of earnings. Their results indicate that accounting methods influence the perceived quality of earnings announcements and this is reflected in the stock prices. Other evidence by Dechow and Schrand (2004) indicates that the correlation between stock returns and earnings has dropped over time. They investigated the correlation between stock returns and earnings from 1952 through 1994 and found that it has dropped from 45 percent to 15 percent. They argue that the drop in correlation suggests a drop in the quality of earnings. The interpretation regarding earnings quality is driven by the motivations and arguments of investors, analysts, and corporate managers. The different results regarding the pre-acquisition performance of target firms when using stock prices or operational data can be caused by the misinterpretation or perceived quality of information by investors. In this way, the reported operating performance of target firms might deviate from their stock price performance.
1 See A. Shleifer (2000). “Inefficient Markets, An Introduction to Behavioral Finance” Oxford University Press
Investigating both the operational and stock price performance allows me to examine the relationship between operational performance and stock price performance. In this way, I can investigate whether the data supports the theoretical prediction of the efficient market hypothesis or the view of alternative theories like the Behavioral Finance Theory.
I have the following structure in this paper. The next section describes the empirical evidence and theories regarding the inefficient management hypothesis. In this section I make the distinction between theories that explain merger activity, the empirical evidence from event studies and studies that develop models or methods for measuring long-term pre-acquisition performance. Section III and IV explain the methodology and the data which is used for measuring long-term pre-acquisition performance of target firms. Section V reports the empirical findings when using stock prices or operational data for measuring the long-term pre-acquisition performance of the target firms. Finally, section VI presents the conclusions which arise from this paper.
2 Review of the Literature
I start this section by analysing the main theories that try to explain the occurrence of mergers and acquisitions. I distinguish four theoretical concepts that provide the basis for the theories that investigate the wealth effects of merger announcements. I mention these theories to highlight the position of the inefficient management hypothesis in the existing finance literature. In addition, this overview provides useful insights in the main theories that try to explain the occurrence of mergers and acquisitions.
Section 2.2 and 2.3 discuss the existing empirical evidence regarding the inefficient management hypothesis. The academic field has showed its interest for the inefficient management hypothesis in a large number of studies. It is possible to make a general distinction between the existing (empirical) studies. The first category is filled with event studies who examine stock return performance, operating performance or q-ratios prior to acquisition. Section 2.2 provides an overview of the main empirical papers that investigated the inefficient management hypothesis. It demonstrates that the existing evidence regarding the inefficient management hypothesis is rather dispersed. The second category as described in section 2.3 contains studies which try to develop models which have the goal to predict the probability of a takeover from past performance. The models use variables which are expected to have predictive power regarding the characteristics of potential takeover targets. I added this part because it delivers insights in the factors that drive mergers and acquisitions.
Finally, I discuss the interrelation of the inefficient management hypothesis with the main theories and empirical evidence in section 2.4. In addition, I highlight the implications for the methodology of this paper which arise from the results of earlier studies.
2.1 Theories explaining merger activity
Traditional finance theory identifies four major concepts regarding the existence of mergers and acquisitions. The theoretical concepts regarding the existence of mergers are efficiency improvements or synergies, agency costs or entrenchment motives, the hubris theory and disciplinary mechanisms by means of the market for corporate control.
Efficiency improvements or synergies after mergers or acquisitions are explained by economies of scale, economies of scope, synergies and transaction costs. Bradley, Desai and Kim (1983) describe a theory based on synergy effects for explaining the existence of mergers
and acquisitions. They argue that economies of scale are achieved when fixed costs of operations can be spread over a larger number of units. This will enable a firm to produce more efficient and will reduce the production costs per product. Economies of scope enable a firm to produce related additional products at lower cost because of experience with existing products. Firms might be able to produce more efficiently than existing companies and therefore enable gains when acquiring those firms. Other synergies are achieved when the combination of existing companies delivers other benefits than economies of scope or scale.
Efficiency improvements explained by the transactions cost theory of Coase (1937) state that decisions about firm size and therefore possible mergers should be determined by the relative transaction costs within and outside the firm. The theory explains that a company should weight transactions costs of separate versus merged entities. Overall, this theoretical concept predicts an increase in the combined firm value, positive gains to the target and neutral gains to the bidder.
The agency cost theory as described in Jensen and Meckling (1976) argues that free cash flow causes managers to undertake value-reducing mergers. Managers use the internal funds that are in excess of the investments required to fund positive net present value projects. Managers are inclined to take up negative net present value projects even when there are no positive net present value projects available. The theory of managerial entrenchment as described by Shleifer and Vishny (1989) states that managers make investments that increase their value to shareholders. Managers make themselves more important and difficult to replace by making acquisitions. Managers are hesitant to pay out cash to shareholders as in the agency cost approach. These two theories predict that mergers have a negative effect on the combined firm value due to the fact that bidder firms pay too much and undertake negative net present projects.
The third theory is based on work of Roll (1986). His work is often described as the managerial hubris theory. He made a model in which markets are strong-form efficient and where managers are prone to excessive self-confidence. In this model, the manager who is the most optimistic about a target firm is confronted with the winner’s curse in a bidding competition. Managers make too many acquisitions due to the over-confidence bias. The winner’s curse states that the highest bid probably is above the actual value of the asset or firm. This theory explains that targets only sell when the bid exceeds the target’s value. There is a wealth distribution from the shareholders of the bidding firm to those of the target firm.
The theoretical prediction of the hubris theory suggests that the gain to the combined firm value is zero. The positive gains that are experienced by the target shareholders are merely offset by losses that are borne by the shareholders of the bidding firm.
Another interesting theory is the theoretical view that the market for corporate control acts as a disciplinary motive for managers. This theory is described by researchers like Manne (1965) and Alchian and Demsetz (1972). This theoretical view predicts an increase in the value of the combined firm. This theory is often described as the inefficient management hypothesis or the disciplinary motive of the market for corporate control. If the management of a particular firm are viewed as merely responsible for the poor performance of the company, another firm or management can use an acquisition to remove the existing management. They may be convinced that there are opportunities to realize large gains by managing the company in a different way. Necessary conditions for the working of the disciplinary mechanism by means of the market of corporate control are a liquid and rational capital market, low degree of anti- takeover defences, relatively small stock holdings of the management and good legal protection of shareholders. This article examines whether the data supports this theoretical view. I took a sample of UK target firms that are listed on the London Stock exchange, because the UK stock market is characterized by the conditions that improve the working of the disciplinary mechanism.
2.2 Empirical evidence from event studies
This section reports the main event studies which examined the inefficient management hypothesis. I try to give an overview of the empirical results and the methods that are used in these studies. The existing evidence is presented in a chronological way. I do not mean to be exhaustive, instead I mention the main papers which I think that are strongly interrelated with the subject of this paper and have made a contribution to the existing literature.
The existing evidence regarding the inefficient management hypothesis is mixed and certainly not convincing. There are a few papers who find statistically significant underperformance for takeover targets. In addition, the methods that are used in the various papers are widely dispersed. The methods for measuring abnormal returns around a corporate event vary from 2 or 3-factor market models, size/beta adjustment, size/industry control firm and estimates with the empirical security market line.
One of the first event studies regarding the inefficient management hypothesis is the paper of Mandelker (1974). He examines the inefficient management hypothesis using the empirical security market line which is estimated by Fama and MacBeth (1973). Mandelker examines stock return performance relative to the empirical security market line. His research delivers an insignificant cumulative average abnormal return (CAAR) of -3% for 252 takeover targets that are quoted on the New York Stock Exchange (NYSE). This CAAR is measured over months (-40 to -9) relative to the month of acquisition completion. He takes the month of acquisition completion instead of the month of acquisition announcement. He makes the assumption that the stock prices of the eight months prior to the merger completion are influenced by the acquisition announcement effect. The announcement effect typically occurs 2-3 months before the acquisition announcement and causes a price run-up.
Another example of the earlier event studies regarding the inefficient management hypothesis is the study of Smiley (1976). The author develops a state preference model of a tender offer and incorporates the risk-adjusted transaction costs concerning the offer. He incorporates the theory of a strong form efficient market which asserts that no superior trading rules or profits can be made. He links this to his hypothesis that the market price of a poorly run corporation will fall no further than the highest potential price less the (ex-ante) transaction costs of a tender offer. He examines his hypothesis using a 3-factor market model (beta, zero-beta asset and industry index). His sample consist 95 tender offers. The data comes from the Center for Research in Security Prices (CRSP) files which consists stocks listed on the NYSE. He finds a statistically significant cumulative average abnormal return of -55.6% over months (-120, -1) for the target firms. These results are clearly not in line with the null hypothesis of zero abnormal returns for target firms.
Asquith (1983) finds a statistically significant CAAR of -14.8% for 211 target firms over days (-480, -60) prior to the announcement date. He adjusts for beta when measuring daily excess returns. His findings are consistent with those of Smiley (1976) and are therefore not in line with the null-hypothesis of zero abnormal returns. The findings of Malatesta (1983) are in contradiction with the prediction of the inefficient management hypothesis. He finds a statistically significant CAAR of 12.6% for the target firms over months (-60, -25) prior to the month of acquisition announcement. He examines 85 US target firms with deal values of at least $10 million. The CAAR is calculated using the market model.
Other studies which use the market model and result in positive pre-acquisition stock price performance of takeover targets are the studies of Martin and McConnel (1991) and Kini, Kracaw and Mian (1995). They arrive at positive CAARs of respectively 4.3% and 2.4%.
Both studies examine US tender offer targets. However, the CAARs that are reported in these studies are insignificant and lower of magnitude than those reported by Malatesta (1983). In addition, both studies examine the relation between pre-acquisition performance and CEO turnover during the post-takeover period. Both studies come to the conclusion that the pre- acquisition performance is inversely related to the CEO turnover during post-takeover period.
Their findings reveal that targets which experience high CEO turnover during the post- takeover period perform significantly worse during the pre-takeover period than targets with a low CEO turnover. This supports the hypothesis that the takeover market disciplines management for poor market-related performance.
Franks and Mayer (1996) examined hostile takeovers in the UK using a market model. They examine the pre-acquisition performance and the board turnover of UK takeover targets. They find an insignificant CAAR of -0.3% over month (-60,-1). They examine 33 hostile takeovers that were announced during 1985 and 1986. The authors find high board turnover and significant levels of post-takeover restructuring in their sample as compared with a control group. They argue that these findings are not caused by poor pre-acquisition performance due to the fact that the targets of hostile takeovers do not perform worse prior to bids than other acquisitions or non-merging firms. Therefore they conclude that the market for corporate control does not function as a disciplinary device for poorly performing companies.
More recent evidence comes from Agrawal and Jaffe (2003) who examined the pre- acquisition operating and stock price performance of US takeover targets. They investigated a large sample of 2083 acquisitions over the period from 1926 to 1996. They use the recommendations of recent studies like Lyon et al. (1999) who investigated the methods for measuring long-term abnormal stock price performance. Agrawal and Jaffe find little evidence regarding underperformance of takeover targets using either operating or stock returns. They conclude that the disciplinary motive regarding takeover targets is not supported by their data. The work of Agrawal and Jaffe (2003) is consistent with a lot of earlier empirical results. However, this section demonstrates that the existing evidence is rather dispersed and therefore I made a re-examination of the inefficient management hypothesis using a different market. Most evidence comes from an investigation of US target firms.
Therefore I took a different, but still a comparable, active and liquid capital market to re- examine the evidence of Agrawal and Jaffe.
2.3 Modelling and measuring pre-acquisition performance
This section discusses the studies and developments regarding the methods for measuring long-term abnormal performance. I give an overview of the influential papers that investigate long-term performance measures and discuss their major recommendations and findings. The choice of a certain model for measuring pre-acquisition performance becomes more important when measuring over a longer period. The use of inadequate models often delivers overstated test statistics. The recommendations of the papers as described in the following sub-sections are therefore extremely relevant for the methodology of this paper. In addition, this section presents the main studies which try to model the predictability of takeovers from past performance or firm characteristics. It delivers insights in the factors that might explain the occurrence of mergers and acquisitions. The outcomes of these studies can be used when choosing variables for investigating the inefficient management hypothesis.
There are several recent developments in measuring abnormal performance. Studies like Lyon et al. (1999) and Kothari and Warner (1997) give some recommendations regarding the measurement of long-term pre-acquisition stock price performance. Kothari and Warner focus mainly on the problems that arise when measuring long-term abnormal stock price performance using an asset pricing model. Both studies investigated the empirical power of test statistics in event studies which are designed to measure long-term abnormal stock returns. Other studies like Brown and Warner (1980,1985) and Campbell and Wasley (1995) investigate the empirical power of test statistics that are designed to measure abnormal stock returns in the short run (< 3-4 months). Lyon et al. (1999) argue that many of the common methods used to calculate long-run abnormal stock returns are conceptually flawed and lead to biased test statistics. Lyon et al. conclude that matching sample firms to control firms of similar sizes and book-to-market ratios yields test statistics that are well specified in virtually all sampling situations. They favour the control firm approach because it yields well-specified test statistics and alleviates the new listing, rebalancing, and skewness biases.
Barber and Lyon (1996) investigate methods for measuring long-term operating performance.
The authors found that there is a lot of variation in the measures and test statistics for measuring abnormal operating performance. They evaluate accounting-based performance
measures, statistical tests and models of expected operating performance in their article. In addition, they analyse the impact of the various measures on the power of certain test statistics. They conclude that non-parametric tests statistics like the Wilcoxon signed ranks test perform uniformly better than non-parametric test statistics. The Wilcoxon signed ranks test statistics are particularly more powerful when the sample follows a non-normal distribution. They further conclude that models which try to measure expected operating performance by measuring the changes in operating performance are more powerful than models which analyse median or mean levels of operating performance. The authors recommend matching sample firms on size and pre-event performance when measuring long- term abnormal operating performance. In addition, they recommend analysing the distinctive features of cash-based and accrual-based operating performance measures when conducting event studies. Accrual-based performance measures can be biased through corporate events like issuing new securities, which can lead to large increases in book values in the short run without proportionally increases in operating income. Cash-based measures like the operating return on sales are more useful in these situations.
Palepu (1986) develops a model to predict acquisition targets using public data. His findings suggest that the likelihood of a takeover is negatively related to a firm’s abnormal return over the previous four years. These findings are consistent with the inefficient management hypothesis. A limitation of this model is the relative small explanatory power. The explanatory power of the model is by no mean superior to the ability of the stock market to predict takeover targets.
Morck, Shleifer, and Vishny (1988) develop a probit model of the probability of hostile and friendly takeovers. Their findings indicate that the probability of a hostile takeover is significantly negatively related to the q-ratio of the firm’s industry. The view that the probability of a hostile takeover is negatively related to the q-ratio is consistent with the inefficient management hypothesis.
Song and Walkling (1993) made a study in which they performed logistic regression relating the probability of being a target to the firm’s return on equity and its market-to-book value ratio. They analyse the joint impact of managerial ownership, size, institutional ownership, and financial characteristics on a) the probability of an acquisition attempt, and b) the probability of contested and uncontested acquisition attempts. Their results indicate that target
firms are associated with significantly smaller levels of managerial ownership than non- targets. In addition, they find that the shareholder returns of target firms are positively and significantly related to managerial ownership in contested but successful acquisitions.
2.4 Interrelation with the inefficient management hypothesis
Previous sections described theories, empirical evidence and methods which are interrelated with an empirical investigation of the inefficient management hypothesis. The fundamental theories that try to explain merger activity all provide plausible theoretical explanations.
Nevertheless, there are some overlapping features between these fundamental theories. The behaviour and performance of managers plays a major role in most theories. For instance, the behaviour or performance of managers is non-optimal in the agency cost theory, the managerial hubris theory and with the inefficient management hypothesis. The inefficient management hypothesis links poor stock price or operational performance to bad management. Management is responsible for poor performance and should be disciplined or corrected. The market for corporate control acts as the disciplining or correcting mechanism. I have to find evidence of poor performance of takeover targets in order to justify the theoretical view of the inefficient management hypothesis and to reject the null hypothesis of this paper.
Overall, it is clear that the existing empirical evidence regarding pre-acquisition stock price performance is rather dispersed. Some papers find evidence which contradicts the inefficient management hypothesis and others find evidence that supports this view or the alternative hypothesis (H1) of this paper. Table 1 presents an overview of the results of the main papers which are mentioned in section 2.2. Smiley (1976) and Asquith (1983) find evidence that rejects the null hypothesis of no significant pre-acquisition underperformance of target firms.
Table 1: Results previous event studies regarding pre-acquisition stock price performance
Study CAAR (%) Months around ann. date Method
Mandelker (1974) -3.0% (-40,-9) Security Market Line
Smiley (1976) -55.6% (-120,-1) 3-factor market model
Asquith (1983) -14.8% (-480,-60) Beta adjustment
Malatesta (1983) 12.6% (-60,-25) Market model
Martin & McConnel (1991) 4.3% (-48,-3) Market model
Kini, Kracaw & Mian (1995) 2.4% (-48,-3) Market model
Franks and Mayer (1996) -0.3% (-60,-1) Size-industry adjusted
Agrawal and Jaffe (2003) 3.76% (-60,-3) Size, B/M, Momentum adjusted
In addition, the methods vary a lot between the several empirical studies. Previous work of authors like Barber and Lyon (1996, 1999), and Kothari and Warner (1997) as described in section 2.3 solve some of the discussions regarding the optimal method for investigating long- term pre-acquisition performance. They prove that certain methods are superior to others when measuring long-term stock price or operating performance. Their recommendations are particularly relevant for investigating the inefficient management in a framework as described in section 3.1. The other evidence as presented in section 2.3 provides useful insights when choosing variables for investigating the inefficient management hypothesis. For instance, I use a cash based measure in the form of the return on sales for measuring operating performance which is advocated by Barber and Lyon (1996). Other evidence demonstrates that it is necessary to adjust for the market-to-book equity ratio when measuring abnormal stock price performance. These findings are incorporated in the methodology of this paper as described in the following section.
3 Methodology
The following two sub-sections explain the methodology I use to examine the long-run stock price performance and the operational performance of takeover targets.
3.1 Measurement of stock price performance
I follow several recommendations of Agrawal and Jaffe (2003) for investigating the long-run stock price performance of takeover targets. They adjust for three factors when calculating abnormal stock returns. They adjust for firm size and market to book because several papers like Fama and French (1992), (1993), Lakonishok, Shleifer, and Vishny (1994) and others show that stock returns are negatively related to firm size and positively related to the market to book ratio. Size and the market to book equity ratio can be used to adjust for the risk factor.
Thirdly, they calculate abnormal returns after adjusting the stock prices with the prior 6- month return. This argument comes from evidence of Jegadeesh and Titman (1993). They show that a firm’s current stock return is positively related to its returns over the previous six months.
They use equally-weighted portfolios after adjusting for size, market-to-book and the prior six month return when measuring the monthly abnormal stock returns of target firms. The equally-weighted portfolios are formed after matching the target firms with NYSE listed firms that fall within the same size quintile, the same market-to-book quintile and the same prior six-month return. Sorting the firms in this way delivers 125 portfolios. All the firms in a particular portfolio are equally weighted when calculating monthly returns for this portfolio.
The monthly return of the portfolio is then used to calculate the monthly abnormal return.
This abnormal return is the difference between the monthly return of a particular target firm and the monthly return of the corresponding portfolio.
Their adjustment regarding the use of equally weighted portfolios is based on evidence from Lyon, Barber and Tsai (LBT) (1999) who examine the efficacy of a number of procedures for calculating long-run abnormal stock returns. These authors examine the statistical power of test-statistics after using several procedures for calculating long-run abnormal stock returns.
They prove that the formation of equally-weighted portfolios yields well-specified test- statistics in non-random samples. Their results indicate that the use of equally-weighted portfolios in non-random samples delivers test-statistics that outperform the ones that are constructed when using value-weighted portfolios of monthly abnormal returns or a Fama and French-type regression framework.
In this paper I investigate the abnormal (monthly) stock return of target firms. The monthly abnormal return is defined as the difference between the return of the target firm and the return of the corresponding control portfolio in month i. For each target firm there is a control portfolio in month i matched on size and the market-to-book ratio. The following sub-section briefly explains the methodology which is used in forming control portfolios.
In year -5 prior to the year of the announcement date, I formed five quintiles of FTSE firms based on firm size (market capitalization). These size quintiles are made with data of the firms that are listed on the FTSE All share Index. The target firms are allocated in one of the size quintiles based on their market capitalization. Within each size quintile, five quintiles are formed based on the market-to book ratio (M/B). In this way, I formed a control portfolio for every target firm. This results in 25 portfolios. I use annual data for both the quintiles representing size and the quintiles based on the book-to-market ratio. The firms in the control portfolio are therefore in the same size quintile and the same book-to-market quintile as the target firm. In addition, the control portfolio of each target firm is revised after 12 months and the firms in the control portfolios are equally weighted when calculating the return of the portfolio. The monthly abnormal return for target i is the difference between the return of the target firm and the return of the corresponding portfolio. The monthly return of a particular portfolio is the average return of the firms in that portfolio. The monthly stock returns of both the target firms and the matching firms are based on the Total Return Index from DataStream.
The Total Return Index from DataStream adjusts for dividends. The index shows a theoretical growth in value of a share holding over a specified period, assuming that dividends are re- invested to purchase additional units of an equity or unit trust at the closing price applicable on the ex-dividend date. The availability of detailed dividend payment data enables a more realistic method to be used in which the discrete quantity of dividend paid is added to the price on the ex-date of the payment. The formula for the return index (RI) follows:
1 1 *
−
= −
t t t
t P
RI P
RI (1)
Where, p = price on ex-date, t pt−1 = price on previous day. The formula slightly changes when t = ex-date of the dividend payment D , then it follows: t
1 1 *
−
− +
=
t t t t
t P
D RI P
RI (2)
ct it
it r r
AR = −
rct
rit
∑
= 2
1 T
T
AAR t
CAAR
t i
it
t N
AR AAR
= ∑
The abnormal return for each target firm is the difference between the firm’s stock return and the return on the appropriate control portfolio. The (monthly) abnormal return (ARit ) of stock i in month t is computed as:
(3) Where is the return on firm i's stock in month t, is the return on the equal-weighted portfolio. This portfolio, as described in the previous sections, consist firms that are in the same size (market capitalization) quintile as firm i and the same B/M quintile as firm i. I calculate for every target firm the monthly abnormal return (ARit ) starting in month -60 and proceeding until month 0. The next step is then to calculate the average monthly abnormal return of month t for the target firms. The average abnormal return (AAR t) across the target firms (N) in month t is calculated with the following formula:
(4)
The final step is to cumulate the average abnormal return of each month over the particular event periods. The cumulative average abnormal return (CAAR) of the target firms over period (t1,t2) is defined with the following formula:
(5) This CAAR represents the cumulative result of the average monthly abnormal returns over period (t1, t2) prior to the month of the acquisition announcement. Table 6 and 7 in section 5.1 reports the CAARs over particular intervals. For every period I test the following hypothesis,
¾ The null-hypothesis: H0: CAAR (t1,t2) = 0
¾ The alternative hypothesis: H1: CAAR (t1,t2) ≠ 0
The cumulative average abnormal returns of the takeover targets have to be negative and statistically significant over pre-acquisition intervals in order to generate evidence for the argument that the inefficient management hypothesis drives mergers and acquisitions because of poor performance of the takeover targets. When finding statistically significant proof for the inefficient management hypothesis I can reject the null-hypothesis of zero abnormal returns for takeover targets.
∑
∑
2 / 1
2 2
1
* 1
t t
AAR t
σ σ
Statistical significance is assessed using a variant of the calendar portfolio method of Jaffe (1974) and Mandelker (1974) The calendar portfolio method of Jaffe (1974) and Mandelker (1974) is recently also referred as the calendar time abnormal return (CTAR) approach by Lyon, Barber and Tsai (1999) and Mitchell and Stafford (2000). Agrawal and Jaffe (2003) made a variant of the calendar portfolio method of Jaffe (1974) and Mandelker (1974). Fama (1998) and Mitchel and Stafford (2000) strongly advocate the calendar portfolio method for several reasons. First, they argue that any methodology that ignores cross-sectional dependence of event-firm abnormal returns produces overstated test statistics. The calendar portfolio method deals with this problem by forming monthly calendar time portfolios in which cross-correlations between abnormal returns are automatically accounted for in the portfolio variance. In addition, the distribution of the portfolio variance using a calendar time methodology approaches a normal distribution which allows for classical statistical inference.
I use the recommendations of Agrawal and Jaffe (2003) for assessing statistical significance.
The t-statistic as explained in equation 7 accounts for cross-correlation between abnormal returns by standardizing the average monthly abnormal return. I start with calculating an average monthly abnormal return (AAR t) for month -60 through month 0 by taking an average of the returns of the target firms in month t. Second, for every monthly abnormal return, I calculate an estimated variance (σt2). This estimate variance in month t is calculated over the abnormal returns of the target firms in month t. Then, I weight the monthly average abnormal returns (AAR t) by dividing it by the estimated variance of that average return, (σt2). The standardized or weighted average monthly abnormal return (AAR t) across the target firms in month t is calculated with the following formula,
2 t t t
AAR AAR
= σ (6)
I defineAAR t to be the average (equally-weighted) abnormal return across all target firms in month t. This method delivers sixty-one weighted monthly average abnormal returns from month -60 through month 0. I then calculate the appropriate t-values for the several intervals which are reported in table 6 and 7 of section 5.1 as explained in equation 7.
T-statistic (t1,t2) = (7)
When calculating the t-statistic for certain intervals, I first define the cumulative results of the numerator of equation 7 for all the standardized (monthly) abnormal returns that fall within a particular interval. Then I calculate the cumulative results of the denominator of equation 7 for the same standardized abnormal returns, take the square root of this cumulative result and divide the cumulative results of the numerator by those of the denominator.
I also calculate an alternative t-statistic for every interval with a method advocated by Barber and Lyon (1999) in order to investigate any possible differences between the two methods.
They use a parametric test statistic to test the null hypothesis of zero mean cumulative abnormal returns for a sample of n firms. I use the following formula to calculate the t- statistics of the mean cumulative abnormal returns over interval (t1,t2):
) / ) (
/( 1, 2
2 , ) 1
2 , 1
( CAAR CAAR n
TCAAR t t = t t σ t t (8)
where CAAR t1 t, 2 represents the cumulative average abnormal return over period (t1,t2) and )
(CAAR t1 t, 2
σ is the cross-sectional standard deviation of the abnormal returns in period (t1,t2) for the sample of n firms. Barber and Lyon (1999) advocate that when the CAARs are clearly non-normal distributed, the Central Limit Theorem guarantees that the distribution of the mean abnormal return measure converges to normality as the size of the sample increases.
In addition, this theorem assumes that the measures of abnormal returns in the cross-section of firms are independent and identically distributed drawings from finite variance distributions.
Barber and Lyon (1999) also investigate the use of cross-sectional standard deviations vs.
time-series standard deviations. They argue that the new listing bias is more severe when using a time-series standard deviation. Therefore, they favour the use of cross-sectional standard deviations. I use their recommendations in calculating the alternative t-statistics for the several intervals that are reported. Finally, for both methods I use the student’s t distribution with v degrees of freedom to determine the critical values of the t-statistic.
3.2 Measurement of operational performance
I use the recommendations of Barber and Lyon (1996) for investigating the pre-acquisition operational performance of takeover targets. I measure operating performance in terms of the operating return on sales (OPS). The OPS is calculated as operating income before depreciation divided by net sales. I adjust for industry, firm size, and past performance when measuring operational performance. The following sub-section explains the methodology which is used when adjusting for industry, firm size, and past performance.
For each target firm I select a control group that operates in the same industry and has an amount of net sales that is between 50% and 200% of the net sales of the target firm in year -5 prior to the year of acquisition announcement. In this way I adjust for industry and size when selecting an appropriate control group for every target firm. Finally, from this control group, I select the firm which OPS is near the OPS of the target firm in year -5. In this last step I adjust for past performance before measuring the abnormal operating performance over the reported intervals. I do not select control portfolios but control firms when adjusting for past performance due to the fact that these control portfolios do not optimally adjust for past performance. The median or average OPS of such a control portfolio deviates more from the OPS of the target firm than the chosen control firm. Nevertheless, when computing the abnormal operating performance of the target firms as a group, I take the median operating performance across target firms and compare it with the median operating performance across the control firms. In this way, I form a control portfolio in the final stage of my investigation.
Every target firm has five yearly operating returns on sales, ranging from year -5 to year -1 prior to the year of the acquisition announcement. However, only the operating return of year -5 is used to select the control firm. The returns of the target firms in the following years are used to calculate the performance relative to that of the control firm which has previously been selected. Thus, for each target, there is a control firm based on OPS. When there is no control firm that matches the criteria mentioned above, we consider all firms, regardless of industry, with the amount of net sales within 50% and 200% of the net sales of the target firm in year -5. Table 8 of section 5.2 reports the median values of the operating return on sales for both the target and the control firms in year -5 to -1 prior to the year of the acquisition announcement. For instance, the median operating return on sales (OPS) across the target firms in year -5 prior to the year of takeover is defined by ranking all the operating returns of the target firms in year -5 and determining the median value. This procedure is repeated for the control firms. In this way, I compare the median operational performance of target firms with those of the control firms for year -5 through year -1.
The Wilcoxon signed ranks test is then used to determine whether any differences in the median values are significant. This test is being explained in the final paragraph of this section. Table 8 of section 5.2 reports for every year the median values concerning the OPS of the target and the control firm. I also calculate the median changes between the years that are reported in table 9 of section 5.2. Equation 9 explains the way I calculated the relative
% 100
*
%
1 ,
1 , 5
, 5
1 ,
−
=
−
=
−
=
−
−
= −
t i
t i t
i vs
i OPS
OPS ChangeOPS OPS
changes between the several years that are reported in table 9 of section 5.2. For every target firm i, I defined the relative change between the OPS of certain years. Equation 9 gives an example of the calculation of the relative change between the OPS of year -1 vs. -5 for target firm i.
(9) In this way I defined the relative change for every target firm. I then ranked the relative changes of all the target firms and defined the median change. This procedure is repeated for the control firms. Finally, the Wilcoxon signed ranks test is used for assessing the statistical significance between the median values of the relative changes between the years that are reported in table 9 of section 5.2.
In measuring the operational performance for takeover targets I first took a sample of targets from the Zephyr database. This is the same sample as used in measuring stock price performance of takeover targets. Therefore, the same firms can be examined on both operational and stock price performance. With an efficient capital market and rational investors there should be a link between operational and stock price performance. Poor operational performance should be reflected in the stock price performance of individual firms. I organized this sample by industry and the announcement date. After this I gathered the operating return on sales (WC08316) and net sales (WC01001) for these targets from DataStream. I then made a selection of all FTSE firms by industry and gathered the individual firm data concerning the operating return on sales and net sales of these firms in order to select an appropriate control firm. After I determined the appropriate control firm I start gathering the operating return on sales for year -5 to -1 relative to the year that consists the announcement date for both the target as the corresponding control firm. The year of announcement is left out of this analysis because of a relatively high level of unavailable data for the target firms.
The statistical significance between the operational performance of the targets and their control firms is examined with the matched pairs Wilcoxon signed ranks test. This test can be computed when there are two subgroups. I use the findings of Barber and Lyon (1999) in choosing the appropriate statistical test. They find that nonparametric test statistics like the Wilcoxon signed ranks test are uniformly more powerful in non-random samples than parametric t-statistics, regardless of the operating performance measure employed. I need a
non-parametric test statistic for assessing statistical significance due to the fact that the data is not normally distributed. Parametric test statistics are only appropriate when the data is normally distributed. I use the statistical package EViews to compute the matched pairs Wilcoxon signed ranks test. This test examines the null hypothesis that the subgroups have the same general distribution, against the alternative that at least one subgroup has a different distribution. The alternative hypothesis states that the values of (the first group) tend to differ from the values (of the second group).
When calculating the Wilcoxon signed rank test manually, compute the difference between the OPS of the target firm i and that of the control firm c. The difference ( di ) between the two populations is then calculated as di = Ti - Ci where Ti and Ci are the target and the control firm, respectively. Rank the di without regard to sign. After ranking, restore the sign (plus or minus) to the ranks. Then compute W+ and W- which are the sums of the positive and negative ranks respectively. If the two population means are in fact equal, then the sums of the ranks should also be nearly equal. The null hypothesis is rejected when the difference between the sums of the ranks is in the rejection region (See Sheskin (1997), pp. 82-94 for a calculation of the rejection region). Table 7 and 8 in section 5.2 reports the p-values regarding the differences between the median values and median changes of the operational performance of target and control firms. There are signs of significant abnormal operating performance when the p-values are below 0.05.
