Information Asymmetry in the Pre-event Period of an International Acquisition: The Effect of Rumors During the Pre-event on the Post-event Abnormal Returns.

(1)

1

Information Asymmetry in the Pre-event Period of an International

Acquisition: The Effect of Rumors During the Pre-event on the

Post-event Abnormal Returns.

Master Thesis Dorien de Vroome

MSC International Financial Management MSC Business and Economics Faculty of Economics and Business Faculty of Social Science

University of Groningen University of Uppsala

(2)

2

Introduction.

On average, foreign acquisitions announcements lead to a zero or negative abnormal return in the short term for bidding companies, compared to domestic acquisitions announcements (Moeller and Schlingemann, 2004, Jarrell and Poulsen, 1989). In the literature, several explanations which fall into two categories have been brought up to explain the zero or negative returns followed by an international acquisition announcement. Firstly, literature implies that foreign acquisitions may not be a wealth creating activity. On the other hand studies imply that the methodological approaches used in previous studies do not accurately measure the returns. Abnormal returns assigned to an acquisition announcement may be gained prior to the official announcement date due to information asymmetry. (Becher, 2009).

Since the short term effects of company news can be spread on an estimation window around the official announcement date, including a pre-announcement period, it is likely that actual returns differ from the expected return of a company prior to an announcement. These abnormal returns can be assigned to the news event in the future. Cumulatively, these abnormal returns prior to the events can cause more than half of the abnormal performance of company news announcements (Keown and Pinkerton, 1981, Jabbour et. al, 2000). Information asymmetry as a cause of abnormal returns in this period prior to the official announcement is considered in this research as an explanation of the official announcement returns of an international acquisition.

(3)

3 acquirers gain prior to the official announcement. Secondly, the market-anticipation hypothesis explains how the market anticipates rumors prior to the official announcement. In this way, investors learn from unofficial announcements and/or press releases indicating the probability of an international acquisition in the future. According to Keown and Pinkerton (1982), merger and acquisition announcements are poorly held secrets which make trading on this information plentiful. In this research I would like to dig deeper in the effect of rumors and information asymmetry prior to the news event on the announcement date and the anticipating ability of the market. More specifically, the effect of rumors and the timing of these rumors prior to the official announcement on the follow up period for bidding firms during an international acquisition announcement. The main research question of this paper will be:

Does the average abnormal return period prior to an official international acquisition announcement, caused by rumors and information asymmetry, have an effect on the post-event

abnormal return of the acquiring firm?

(4)

4 essential to find out whether (eventual) positive results of an international acquisition announcement will already be acquired prior to the announcement; the market already anticipated on the international acquisition announcement prior to the official announcement. Furthermore, could the (eventual) negative abnormal returns after the announcement be explained by the rumors prior to the event and the time period between the rumor date and the official announcement date? Considering an international acquisition, it could be important for management to anticipate differently when target companies are based in different countries with different law systems. Lastly, portfolio managers can use the articles’ results by using them in their decisions to create an overall higher return for their investors. Can they anticipate on rumors, and will these affect the post announcement returns.

(5)

5 companies (ranging from 120 to 5 days prior to the announcement date). However, in this research, every company’s own estimation window (from rumor date to announcement date) is taken to calculate the average abnormal return prior to the event.

Consistent with the literature, I find zero average abnormal returns after the official announcement. Furthermore, results suggest the average abnormal returns prior to the announcement are positive. However, the difference between abnormal returns after and the abnormal returns prior to the official announcement is not statistically significant. Furthermore, I find a (insignificant) negative relationship between the post average abnormal returns and the pre average abnormal returns of an international acquisition announcement. A (insignificant) negative relation between the amount of days prior to the official announcement and the post average abnormal returns is find as well. Both of these relations are controlled for relative size of the deal, market-to-book ratio of the acquirer, method of payment, the country law system of the target company, and the status of the target company.

(6)

6

Literature review

For the last decades, cross-border acquisitions have been the main activity in increasing foreign direct investments (Uddinand Boateng, 2011). Why do firms decide to involve in cross-border acquisitions as opposed to domestic acquisitions? According to Doukas and Travlos (1988) three options of value creation are considered for the multinational firm that differs from acquiring a domestic firm. Firstly, value is created by using the institutional restrictions wisely. In this case, tax codes and financial limitations in different countries could be used to create value for a multinational firm. Secondly, information and knowledge expands internationally. In this way, learning cost externalities are captured by the firm in the conduct of international business. Lastly, cost savings for the firm could be gained by the joint production of marketing and manufacturing in different countries. The complete explanation of foreign direct investments, according to Kogut (1983), is the development of a multinational network; “The primary advantage of the multinational firm, as differentiated from the national corporation, lies in the flexibility to transfer resources across borders through a globally maximizing network. The incremental value of internationalizing of the multinational corporation reflects the option of exercising it by the MNC itself, so the individual investors cannot trade or acquire this by themselves.”

(7)

7 acquisition should be reflected in the acquiring firm’s today’s market value. Fama (1997) explains that according to the positive-multinational network theory firms indeed acquire foreign corporations in an attempt to maximize shareholders’ wealth. The intention of this type of diversification is to create a higher return for the acquiring firm in the long run. Consequently, an international acquisition announcement will have a positive effect on the share price of the bidding firm in the short run (after announcing: post-event). Because a change in future cash-flows is suggested after announcing the acquisition, today’s share price will change due to the discounted future cash-flows that are reflected in today’s share price of the bidding firm, as is explained by the efficient market hypothesis. In theory, one could argue that announcing an international acquisition should lead to higher returns of the acquiring firm after the event occurred. However, in practice (empirical research), researchers find the wealth effect for the shareholders in bidding firms to be on average close to zero and even to be negative in some categories (Jarrell and Poulsen, 1989). For example, Mathur et. al (1994) find that shareholders of (foreign) bidding companies acquiring a US company earn a significant negative abnormal return after announcing the acquisition in the USA. Furthermore, these negative abnormal returns are significantly increasing over 15 days after the announcement. Moeller and Schlingemann (2004) find significantly lower announcement stock returns for US companies acquiring a cross-border target, compared to domestic targets.

(8)

(9)

9 observed in the post-announcement date. Lastly, Jabbour et. al (2000) find that 60 days prior to the announcement of a takeover Canadian target firm, abnormal returns gained. Which are related to insider trading and information leakage, consistent with the second hypothesis. Furthermore, they find in a short time prior to the event, runner ups and increased abnormal returns are related to the anticipation on these insider trading reports and rumors, consistent with the markets’ anticipation hypothesis. One can conclude that rumors before officially announcing an international acquisition could influence the abnormal returns after the announcement date for a target and or acquiring company, indicating the market anticipated already on the acquisition prior to the official announcement .

Market anticipation

(10)

10 period influences the information gathering in the first period. The final period is described as the public consuming. Kim and Verecchia (1991) focus on “how anticipating a public announcement, through information gathering, affects the market reaction to the announcement.” Findings include a weaker price reaction at the time of announcement due to more private information gathering in an anticipating announcement. The price sensitivity, at the official announcement, declines when an announcement is anticipated on. Information asymmetry prior to the announcement affects the price (and volume) sensitivity at the time of the announcement. The quality of this information, the level of noise, and the cost of gathering this information affects this relationship.

(11)

11 expectation of the acquisition occurring in the future and the market already anticipating on this. The first hypothesis will test whether the average pre-abnormal return between the rumor date and announcement date of an international acquisition announcement influences the post-event returns. I argue, in line with the market anticipation hypothesis, that prior to the international acquisition announcement investors already anticipated on the future event by rumors. This is consistent with the findings of Keown and Pinkerton (1981), hence they find the market efficiently impounds rumor information. In this way, the market correctly assesses the average probability of a future takeover. Consistent with this, I suggest negative relationships between pre-runner ups and post-event abnormal returns during an international acquisition announcement.

Hypothesis I: The pre-runner ups from bidding companies before an international acquisition announcement have a negative effect on the post-event returns after the international acquisition announcement .

(12)

12 date, the rumor date and the announcement date are the same. One could argue that the surprising international acquisitions announcements are not vulnerable to market anticipation, hence, nobody could have anticipated prior to this announcement (excluding illegal insider trading).

(13)

13

Hypothesis II: The abnormal return after an international acquisition announcement will be lower when the rumor date occurs more days prior to the announcement date.

Data

The sample of the research contains international acquisitions of UK listed companies. The time period used for the sample is between 2004 and 2012. This time period is after the implementation of the requirements in compliance with the market abuse directive. The UK legal regulation involved in market abuse is aligned with the European Union directive of market abuse (2003/6/EC) as of the year 2003. This directive obliges firms to release all information that is considered to have an effect on stock price. Furthermore, penalties consider illegal insider trading increased (Prevoo and ter Weel, 2010). The 2003 market abuse directive gave all European Union countries a measurement of discretion. Since the UK regulation already had a broad compliance with this directive, but still differed in some material aspects, the UK government implemented the requirements of directive but certain provisions remained (Slaughter and May, 2011). Choosing the time period after this implementation suggests there is less of a chance of insider trading and leakage, as opposed to market anticipation. This assumption is consistent with the findings of Durney and Amrita (2007). They find, on average, insider trading restrictions reduce the amount of private information trading. Results from this research therefore could be assigned to market anticipation instead of (illegal) insider trading.

(14)

14 Completed acquisitions used in this research include acquiring at least 50% of the target shares, or increases its stake to at least 50% of the target shares. As explained above, to consider market anticipation as a possible explanation of the average abnormal return after the announcement, the rumor date is used. The amount of days between the rumor date and official announcement date of the international acquisition is considered as a variable to measure to measure the market anticipation effect. The rumor date will be the date reported as “rumor date” in Zephyr data set. “This is the date on which the deal was first mentioned, as far as Zephyr researchers can ascertain.”

Not all of the 1591 deals have a difference in rumor and announcement date. It is important in this research, however, to only include the deals that have difference in rumor and announcement date. This leaves the sample to a number of 281 deals. It is important to consider that even though Zephyr did not have a rumor date for the majority of the deals, this does not mean that there has not been a pre-bid run-up, and eventually insider trading or rumors prior to the official

Table 1: description of deals.

Number of Deals 158 Method of Payment Cash 69 Other 89 Total 158 Legal System Common-law 84 Civil-law 74 Total 158

Target Company listing

Private 129

Public 29

(15)

15 announcement occurred. However, for this research in line with testing the market anticipation hypothesis, it is important to consider the difference between rumor and announcement date. Acquisitions with an unknown deal value, unknown ISIN number and insufficient or incomplete data are excluded in the sample. This leaves the sample to a number of 158 deals, in the table below a summary of these statistics is shown (see for complete list appendix 1).

(16)

16

Variable Description Calculation Source

Deal Value The consideration paid for the actual stock required to gain at least 50% of the target’s stock. In pounds.

Zephyr

Rumor date The date the deal is first mentioned “as far as Zephyr

researchers can ascertain”. (date=rd) Zephyr

Announced date The date a formal offer has been made or when of the companies involved confirmed that the deal is to go ahead.(date=0)

Zephyr

Days Number of days between the rumor date (rd) and the official announced date (0).

Days = Date 0 – Date rd Zephyr

Daily Return Index The theoretical daily growth in value of a company Datastream Daily Return Market index The theoretical daily growth in the value of the local

market: Financial Times Stock Exchange All Share Index

Datastream

Abnormal Return The percentage abnormal return gained daily for the acquirer.

Abnormal return = Actual return – Market Return

Datastream

Post Average Abnormal Return The average daily abnormal return after the official

announcement of the international acquisition.

Datastream

Pre Average Abnormal Return The average daily abnormal return prior to the

official announcement of the international acquisition

_Datastream

Method of Payment The method used to pay for the acquired stock. Cash, Share, or a combination of both is the most used method of payment. Other methods include, debt, loan notes and earn out.

Zephyr

Market to Book ratio Market value of the ordinary equity divided by the balance sheet value of the ordinary equity in the company (book value).

Datastream

Relative Size Deal value in pounds (the stock required from the target company; the sum of all consideration paid) divided by total assets (the sum of total current assets, long term receivables, of the Acquirer.

(17)

17

Dummy Method of payment The method the deal value is paid with. For example cash (payment is done by cheque or transfer of funds) or shares (the acquirer gives its own shares, or issues new shares for the deal). Dummy is set 1 when the deal is paid with cash.

Zephyr

Dummy Country law system of target company

The law system used in the country of the target company, either common-law or civil-law. Dummy is set 1 when the country of the target company uses a common-law system.

Zephyr

Dummy Status target company The status of the target company; either privately owned or listed. Dummy is set 1 when the company is privately owned.

Zephyr

(18)

18

Methodology

In calculating the average abnormal return assigned to an event, like an international acquisition announcement, first the abnormal return is calculated for every day on the estimation window. These will be calculated using the actual return for company i at time t ( minus the market return on time t ( .

Daily stock return

= - Abnormal return

The average abnormal returns assigned to the event of an international acquisition in the future will be calculated using the abnormal returns starting from the rumor date, t=rd to the official announcement date t=0, divided by the number of dates between rumor and announcement date.

Average abnormal return between rumor and announcement date: pre average abnormal return

In which 0 = announcement date, rd = rumor date, N=number f days between rumor and announcement dater: announcement date – rumor date

For the short term abnormal return after the announcement, the average of a 60-day period is chosen. In previous research, the short term estimation window ranged from +1 ( Dutta et. al., 2013) to + 30 or even 120 days (Cumming and Li, 2011). The 60 day estimation widow used in this research, is an average of these previous research. Furthermore, the pre period is, on average 76 days, so I decide to choose a post period of somewhat equal to the pre period.

(19)

19 A t-test will be performed to test whether the post average and the pre average abnormal returns are significant. Furthermore, the difference between the post and pre average abnormal return is calculated for the entire sample, and also test for statistical significance.

In the next step, regression analysis is being used to see if market anticipation can account for the difference between post and pre average abnormal return. This tests whether the prior abnormal returns and the amount of days between the rumor and official announcement date can explain for the post abnormal return of a bidding company in the UK after an official international acquisition. In performing the regression for this dependent variable, the independent, control, and dummy variables, are added step by step. In the first step the pre average abnormal return and the amount of days (independent variables) are used to explain the post average abnormal return after the official announcement. The equation is as follows:

In which:

(20)

20

Relative size of acquisition to the acquiring company.

Firstly, previous literature uses regressions to explain bidder return in acquisitions gains, usually control for the relative size of the target company compared to the size of the acquirer. If the abnormal return gained by the acquirer is not dependent on the size, the abnormal return should increase with an increase of the relative size of the acquisition (Moeller et. al. 2004). However, a bigger acquisition results in more shares outstanding for the acquiring company, which could lead to a decrease in demand of the acquirers share. Consequently the abnormal return will fall if the relative size increases; a negative relationship. In practice however, the sign for relative size variable varies among previous literature, but in most cases the variable is significant. The relative size is added as a control variable, suggesting a negative relationship, which is explained by theory.

Market to Book ratio.

(21)

21 information is an increasing function of firm size. Therefore, it is riskier to trade in smaller companies; consequently, investors demand a higher expected return (Haugen and Lakonishok, 1988). Hence, rumors are more common and picked up in large companies, as they are closely watched. Rumors in small firms, on the other hand, are more uncertain and are less predictable. Furthermore, Jensen and Ruback (1983) argue that bidders tend to be larger firms and therefore also more watched upon resulting in less surprises. As Fama and French (1993) describe, how variation of stock returns is caused by the difference of the size in the company. Furthermore they find “book-to-market equity captures the variation in stock returns associated with size”. Therefore, the market-to-book ratio from the acquiring firm is used as a control measure.

In the second step of the regression, the control variables size and market to book ratio are added creating the following equation:

In which:

Other control variables used that are identified in the literature are used as a dummy in the final regression and added in the final and third step.

Difference in Legal system

(22)

22 vulnerable for rumors. Another factor that should be considered in explaining abnormal returns in the process of an international acquisition announcement is the traditional legal system of both countries. Particularly, abnormal returns can be influenced when a target company operates under in a country with a different legal system (Barpolous et al., 2012). The common law system, which originated in England, is the current traditional legal system in the United Kingdom and former United Kingdom colonies like the US, Australia, New Zealand and many others. Usually, differences in legal systems between countries make the integration of a foreign target more complex and costly, consequently affecting the abnormal returns. On the other hand, Barpolous et al. (2012) find UK acquirers gain more acquiring targets in civil-law countries in both the announcement period and in the long run. As opposed to a civil law system, the common law system is based on judges’ decisions, resulting in a higher investor protection in countries practicing the common law system. This results in higher takeover premiums, leaving fewer benefits for the shareholders of the UK acquirer. Consequently, consistent with the findings of Barpolous et al. (2012) acquiring a company in a civil-law country, with lower investor protection, will result in a lower takeover bid and higher abnormal returns in the announcement period.

Method of payment

(23)

23 acquisitions. For example, the target company’s management will change during a cash offer which could influence the post-acquisition abnormal returns negatively. Furthermore, in stock offers, management of the target companies is more likely to retain a level of ownership. This could be important as local managers will monitor the local organization which could result in “better synergy realization”. Further, local investors could have access to specific information, which could be valuable in stock valuation (Uysal et al., 2008). Dutta et al. (2013) used a sample of Canadian acquiring firms and find evidence that during the announcement period stock offers lead to higher abnormal returns in cross-border acquisitions. However, when considering the same sample over a long term period, Dutt et al. (2013) find that stock financed acquisitions significantly underperform. This means that in the long-run, the market makes adjustments for the abnormal returns gained in the short announcement period. “Cross-border stock financed deals do not live up to the expectations in the long-run.” This does means that for the short time effect of the abnormal returns after the announcement examined in this study, the results should be controlled by the method of payment, suggesting a positive relationship.

Private vs public targets.

(24)

24 Furthermore, Conn et al. (2005) find similar results for both domestic and cross-border acquisition for UK bidders. However, this is only true for the announcement returns. In contrast, they find significantly negative returns for the three-year post-event returns for both publicly or privately held targets in domestic as well as cross-border acquisitions. Several arguments could address these results. Firstly, because private targets are less visible to the public, negotiating can be done internally without “losing face”. This results in avoiding overpayment (Conn et al., 2005). Furthermore, private targets are more difficult to sell due to their liquidity, therefore they could sell at discount (Fuller et al., 2002). In this way, acquiring firms could benefit from acquiring private targets as compared to public targets, hence creating higher short term abnormal returns from acquiring a private target.

In the last step of the regression, the previously explained dummies, to control, will be added in the regression creating the complete equation used to test the hypotheses. This equation will be:

In which:

(25)

25 considered. When pre-bids do have a negative influence on the post-event returns, the should be negative. Hence, the first hypothesis is accepted if this is a negative number.

Hypothesis I:

The next hypothesis, the abnormal return after an international acquisition announcement will be lower when the rumor date occurs more days prior to the announcement date, suggests there is also a negative relationship between the amount of days between the rumor and the announcement date and the post-event abnormal returns. In order to accept this hypothesis, should be negative as well.

Hypothesis II:

(26)

26

Robustness test

Since insider trading is hard to detect and/or measure, some may argue the results of the regression can be a sign of insider trading rather than the rational expectation of an international acquisition occurring in the future. To test the robustness of the research in providing support for importance of timing and information asymmetry in line with the market anticipation hypothesis, I substitute the time window of the pre-event (rumor date until official announcement date) to a 60-day pre-event time period. Time windows between rumor and announcement dates of international acquisition in the UK differ from a 2-year to a 5 day time range. As suggested earlier, the market impounds the information immediately after the rumor date. If, as I suggest, the market impounds information starting from the rumor date, the results should differ with using a 60-days estimation window. Using a 60-days estimation window is consistent with the research of Jabbour et. al (2000), in that abnormal returns prior to the announcement start up to 60 days before the official announcement. If results do not differ using this different estimation window, it could be suggested that the difference in the rumor date and official announcement date do not have an influence on the short term abnormal returns of the post-event, therefore not supporting the market anticipation hypothesis.

Average abnormal return prior to the event

The regression equation will be:

(27)

27 The second beta in this equation is irrelevant, because the amount of days between the rumor date and the official announcement date of the acquisition will be the same for every deal. Consistent with the market anticipation hypothesis, I suggest that after regressing this robustness equation, the beta coefficient of the pre average abnormal return will differ from the beta coefficient used in the original regression:

Results

After collecting the data, the descriptive statistics of the different variables are presented. Table 3 presents these descriptive statistics. When comparing the maximum and the minimum variables to the means of the variables, one can conclude some of the variables include high outliers. The outliers could have a serious effect on the estimates; the ordinary least square regression could have a “penalty” by example, of the form of an increased residual sum of squares. Therefore, the extreme values are limited using the process of winsorizing. Winsorizing transforms the data set by limiting the extreme values, and reducing the possible effects of outliers. Table 4 presents the descriptive statistics of the data using a winsorization percentile of 5% (the percentile is not used for the dummies). This means that the data above (below) the 97,5th (2,5th) percentile are set to the 97,5th (2,5th) percentile (descriptive statistics with different winsorization percentiles can be found in appendix 2). Comparing table 4 to table 3, one can conclude that the minimum values of the variables increased and the maximum values decreased. Furthermore, the Kurtosis and the Jarque-Bera both decreased after the winsorizing process, indicating the data tend to move to a normal distribution.

(28)

28

Post AAR Pre AAR Days Size MB ratio Law Meth Pay Status Target Mean 0.0000 0.0007 136.6815 0.9254 4.6497 0.5350 0.4395 0.8217 Median 0.0005 0.0003 76.0000 0.0668 1.9400 1.0000 0.0000 1.0000 Maximum 0.0076 0.0370 728.0000 44.9640 249.3600 1.0000 1.0000 1.0000 Minimum -0.0310 -0.0206 3.0000 0.0000 0.1300 0.0000 0.0000 0.0000 Std. Dev. 0.0038 0.0055 150.4500 5.1536 20.3678 0.5003 0.4979 0.3840 Skewness -3.9099 2.6181 1.7092 8.0273 11.3836 -0.1405 0.2438 -1.6805 Kurtosis 29.799 20.285 5.5866 67.8310 136.3779 1.0197 1.0594 3.8242 Jarque-Bera 5098.241 2133.935 120.2069 29181.46 118239.3 26.1692 26.1897 78.3435 Probability 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 Observations 157 157 157 157 155 157 157 157 Table 3 Descriptive statistics variables without winsorizing.

Post AAR Pre AAR Days Size MB ratio Law Meth Pay Status Target Mean 0.0000 0.0005 134.5911 0.3373 2.8402 0.5350 0.4394 0.8217 Median 0.0005 0.0003 76.0000 0.0668 1.9400 1.0000 0.0000 1.0000 Maximum 0.0051 0.0122 565.4500 4.0034 15.1788 1.0000 1.0000 1.0000 Minimum -0.0092 -0.0090 4.0000 0.0006 0.3350 0.0000 0.0000 0.0000 Std. Dev. 0.0027 0.0036 143.1737 0.7648 2.7883 0.5003 0.4979 0.3840 Skewness -1.1944 0.5395 1.4765 3.6937 2.7769 -0.1405 0.2438 -1.6805 Kurtosis 5.3240 6.0452 4.2736 16.6140 11.8842 1.0197 1.0594 3.8241 Jarque-Bera 72.6642 68.2793 67.6598 1569.4380 708.9763 26.1692 26.1898 78.3436 Probability 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 Observations 157 157 157 157 155 157 157 157 Table 4 Descriptive statistics variables using winsorization percentile of 5%.

(29)

29

Hypothesis testing

Test of Hypothesis: Mean: 0.0000 Included observations:157 Winsorization percentile: 5%

Post Average Abnormal Return

Pre Average Abnormal Return

Difference Post and Pre Average Abnormal Return

Sample Mean 0.0000 0.0006 -0.0005

Sample Std. Dev. 0.0028 0.0036 0.0048

T-statistic 0.2600 1.8652 -1.2609

Probability 0.7952 0.0640 0.2092

Table 5 Hypothesis testing, simple t-test.

The next step is to test whether multicollinearity exists between the (winsorized) variables used later in the regression. Mutlicollinearity occurs when the assumption of non-correlated

explanatory variables in an OLS regression does not hold. In this case, adding or removing a variable from the regression cause other variables to change. Table 6 presents the correlation between the variables used in the OLS regression. The highest (negative) correlation number is between the size and the method of payment. This correlation number is not of concern in the next steps of this research.

Post AAR Pre AAR Days Size MB ratio Law Meth Pay Status Target Post AAR 1.0000 -0.0969 -0.0179 -0.2569 -0.0903 -0.1178 0.1248 0.0656 Pre AAR -0.0969 1.0000 -0.0100 0.0256 -0.0091 -0.0392 -0.1352 -0.0119 Days -0.0179 -0.0100 1.0000 0.0036 0.0947 -0.1934 -0.0352 -0.0853 Size -0.2569 0.0256 0.0035 1.0000 0.2045 0.0732 -0.2983 0.0962 MB ratio -0.0903 -0.0091 0.0947 0.2045 1.0000 0.0551 -0.0272 0.0524 Law -0.1177 -0.0391 -0.1934 0.0731 0.0551 1.0000 -0.1296 -0.0063 Meth Pay 0.1248 -0.1352 -0.0352 -0.2982 -0.0272 -0.1296 1.0000 0.0434 Status Target 0.0656 -0.0118 -0.0853 0.0962 0.0524 -0.0063 0.0434 1.0000 Table 6 Correlation Matrix variables using winsorization percentile of 5%

(30)

30 abnormal return and days) do not make major changes when the control variables and dummies are added. This is consistent with the correlations found between the (in)dependent variables and the assumption they do not influence each other. Recalling the control variables and dummies in the literature section, I suggest the size variable has a negative effect on the post average abnormal return. As can be seen in table 7, size indeed has a negative sign, and is the only significant variable in the regression. The market to book ratio also has a negative (insignificant) sign. The other controlling coefficients in the regression give the prospected sign prior to the coefficient. Firstly the dummy for the target country law system; the dummy is 1 when the target company is in a common law system, suggesting less average abnormal return after the official announcement due to the high investor protection in these law systems. In the regression, this coefficient is indeed negative but insignificant, with a t-value of -1.376. The literature states that mergers and acquisitions paid with cash create higher post average abnormal returns. The dummy (set 1 when the acquisition is paid in cash) shows a positive beta for the method of payment dummy, however insignificant. Lastly, literature suggests that privately owned targets create higher average abnormal returns, this is consistent with our findings, hence is positive, however insignificant as well.

(31)

31 an international acquisition announcement will be lower when the rumor date occurs more days prior to the announcement date; more days in between the rumor and announcement days allow more information in the market, therefore the market already anticipated on the international acquisition before the official announcement. The coefficient of in the output of the regression has a tendency to be negative, hence really small, but insignificant as well. So no evidence is found to support both of the hypotheses.

Ordinary Least Squares Dependent Variable:

Post Average Abnormal Return Method: Least Squares

Step 1 Step 2 Step 3

Coefficient Coefficient Coefficient

C 0.0001 (0.0003) 0.0006 (0.0004) 0.0003 (0.0007)

Pre Average Abnormal return -0.0743 (0.0619) -0.0713 (0.0615) -0.0716 (0.0621)

Days -0.0000 (0.0000) -0.0000 (0.0000) -0.0000 (0.0000)

Size -0.0009 (0.0003)* -0.0009 (0.0003)*

Market to Book Ratio 0.0000 (0.0000) 0.0000 (0.0000)

Law System Dummy -0.0006 (0.0004)

Method of Payment Dummy 0.0001 (0.0004)

Target Company Status Dummy 0.0006 (0.0006)

Observations 157 157 157

R-squared 0.0096 0.0760 0.0951

Table 7: Coefficients(std error) of ordinary least square regression performed. * = 1% significance level, ** = 5% significance level, *** = 10% significance level.

Split sample

(32)

32 suggested previously, the post average abnormal return is lower in the group with more days between rumor and official announcement day of the international acquisition. However, the ANOVA test shows this is not statistically significant.

Test for Equality of Means of Post Average Abnormal Return Categorized by values of Dummy of Days

Winsorization percentile: 5% Included observations: 157

Method df Value Probability

t-test 155 -0.6470 0.5186

Satterthwaite-Welch t-test* 146.0874 -0.6481 0.5179

Anova F-test (1, 155) 0.4187 0.5186

Welch F-test* (1, 146.087) 0.4200 0.5179

*Test allows for unequal cell variances

Dummy of Days Count Mean Std. Dev. of Mean

0 79 -0.0002 0.0040 0.0005

1 78 0.0001 0.0031 0.0003

All 157 0.0000 0.0036 0.0003

Table 8 Test of Equality of Means for the “Post Average Abnormal Return” variable. Categorized by the groups split, based on the amount of days between rumor and official announcement day of the international acquisition.

Next, the two groups are regressed separately, using the same steps as previously used. Table 9 and table 10 show the regression results for the high group and the low group respectively. The pre average abnormal return show, for both groups, a negative beta coefficient in all steps. This is consistent with the results of the regression with the original (entire) data sample (table 7). While the group with the higher amount of days show consistent results with the original regression (table 7), a tendency to be negative, the other group shows a positive beta for the amount of days instead of negative one. While this difference is very small, since the numbers are small, this results suggest that time indeed is of importance in explaining the post abnormal returns of an international acquisition. Since, the results are not significant no conclusions can be drawn from this.

(33)

33 the target company status is inconsistent with the original regression. Furthermore, none of the coefficients are statistically significant in this regression.

C 0.0000 (0.0006) 0.0003 (0.0007) -0.0009 (0.0009)

Days -0.0000 (0.0000) 0.0000 (0.0000) 0.0000 (0.0000)

Size -0.0002 (0.0000)** -0.0002 (0.0000)**

Market to Book Ratio -0.0000 (0.0000) -0.0000 (0.0000)

Law System Dummy -0.0007 (0.0006)

Target Company Status Dummy 0.0012 (0.0007)***

R-squared 0.0007 0.0988 0.2072

Table 9: Coefficients(std error) of ordinary least square regression performed of split sample. Sample; above mean pre announcement days. * = 1% significance level, ** = 5% significance level, *** = 10% significance level.

C 0.0002 (0.0006) 0.0008 (0.0008) 0.0019 (0.0013)

Days 0.0000 (0.0000) 0.0000 (0.0000) 0.0000 (0.0000)

Size -0.0006 (0.0005) -0.0007 (0.0005)

Market to Book Ratio -0.0002 (0.0002) -0.0002 (0.0002)

Law System Dummy -0.0002 (0.0007)

Method of Payment Dummy -0.0006 (0.0008)

Target Company Status Dummy -0.0009 (0.0010)

R-squared 0.0180 0.0461 0.0662

(34)

34

Robustness test

Since one could argue that the results presented above are due to insider trading as opposed to market anticipation, a robustness test is performed. Therefore, the same regression is run however, the pre average abnormal return is substituted by a new pre average abnormal return, over 60 pre days for every company. Using this calculation for the pre average abnormal returns will delete the amount of days variable in the regression. In appendix 5, the new descriptive statistics for the pre average abnormal returns is presented. The regression results are presented in table 11 below. The coefficients show the same sign for all the betas compared to table 7, except for the pre average abnormal return. Instead of a negative sign of in table 4, table 7

Table 11: Coefficients(std error) of ordinary least square regression performed the robustness test wit winsorize percentile of 1 percent. * = 1% significance level, ** = 5% significance level, *** = 10% significance level.

shows a positive . As I argued above, the beta coefficient for the pre average abnormal return in the robustness test is indeed different than the beta coefficient in the original regression. In this case the beta of the robustness test is higher than the original beta and positive as opposed to negative. This means that when the amount of days prior to the announcement is the same for all

Post Average Abnormal Return

C 0.0000 (0.0002) 0.0004 (0.0003) 0.0001 (0.0006)

Pre Average Abnormal return 0.0278 (0.0917) 0.0571 (0.0903) 0.0485 (0.0911)

Size -0.0009 (0.0003)* -0.0008 (-2.8626)*

Market to Book Ratio 0.0000 (0.0000) 0.0000 (0.0000)

Law System Dummy -0.0005 (0.0004)

Target Company Status Dummy 0.0006 (0.0006)

(35)

35 deals, the post acquisition returns are higher. This suggests we find more support for the importance of the amount of days between rumor and official announcement date hence the difference in days between rumor and official announcement date has a different (negative as opposed to positive) output of the coefficient. This result is insignificant, however. Therefore we cannot reject the insider trading hypothesis; nor is support found to accept the market anticipation hypothesis.

Discussion and implications

(36)

36 of Jabbour et. al. (2002) arguing market anticipation is a follow up on the illegal insider trading. This research tries to exclude illegal insider trading in its measurements by selecting the time frame of UK firms between 2004 and 2012, where the alignment with European regulation, which make an attempt to decrease illegal insider trading, was fully implemented.

Furthermore, the rumor date, the first date the international acquisition is mentioned as far as the database Zephyr can ascertain, is an important variable in this research. However, the rumor date can be seen as an arbitrary date. The definition of the rumor date used in this research is given by Zephyr as “The date on which the deal was first mentioned, as far as Zephyr researchers can ascertain. The report may be in the press, in a company press release or elsewhere. The rumor is an unconfirmed report for example, ABC Company is looking at acquisitions in the United States, or XYZ Ltd is planning to list on the Frankfurt Stock Exchange in the next 3 months. If the first mention of the deal is when it is officially announced, then that date is entered as Announced with the same date for both the Rumor date and Announcement date.” Besides this definition being arbitrary itself, it also argues that no rumors occurred prior to the official announcement when the announcement and rumor data are equal. Which would be an false assumption, hence rumors and information asymmetry could be present without notice.

(37)

37 However, when different (smaller) estimation windows (5, 10 or 20 post days) are used, the results do not change significantly.

Lastly, this research does not include the rational expectation of a potential international acquisition occurring in the future. This means investors could expect an acquisition based on the historical pattern of a company. In this way, a wave of acquisitions could increase the likelihood of a new acquisition in the future.

Conclusion

This research considers timing and information asymmetry, in line with market anticipation theory, as a proxy to explain post announcement returns in international acquisition. The pre average abnormal returns and rumors prior to the official announcement days are used as variables to explain the average abnormal returns after the official announcement of the international acquisition. To measure the timing, the days between the rumor date and the official announcement date of an international acquisition is used. The research question answered is:

Does the average abnormal return period prior to an official international acquisition announcement, caused by rumors and information asymmetry, have an effect on the post-event

abnormal return of the acquiring firm?

(38)

38 however. When the sample is split based on the amount of days, a difference is shown, suggesting a higher influence on the post abnormal return when more days occur between the rumor date and announcement date. A robustness test is performed with the sample, using the same amount of days (60 prior days) as a pre announcement period. Regressing this sample, the pre average abnormal returns have a positive influence n the post average abnormal returns, as opposed to using difference timing range for each deal individually. These results are also insignificant. Even though the results are mainly insignificant, it suggests the time prior to the announcement is of importance in explain the post average abnormal return. Since the timing that is considered cannot be used as the main proxy for market anticipation, these results cannot be used to reject or accept the market anticipation hypothesis. Further research is needed to find if market anticipation is the main explanation for the short term announcement return for an international acquisition. Furthermore, to distinguish between insider trading and market anticipation is a difficult task. Therefore, further research is needed as well.

References

 Aw, M.S.B., Chatteriee, R.A. (2004) The Performance of UK Firms Acquiring Large Cross-Border and Domestic Takeover Targets. Applied Financial Economics. Vol. 14, Iss. 5, p337-349  Barbopoulos, L., Paudyal, K., Pescetto, G. (2012) Legal Systems and Gains From Cross-Border

Acquisitions. Journal of Banking and Finance. Vol.29, p533-564

 Becher, D.A. (2009) Bidder Returns and Merger Anticipation: Evidence from Banking Deregulation. Journal of Corporate Finance. Vol.15, Iss. 1, p85-98

(39)

39  Conn, R. L., Cosh, A., Guest, P. M., Hughes, A. (2003) The Impact on UK Acquirers of

Domestic, Cross-Border, Public and Private Acquisitions. Journal of Business Finance & Accounting. Vol. 32, p815–870.

 Cumming, D., Li, D. (2011) Run-up of Acquirer’s Stock in Public and Private Acquisitions. Corporate Governance: An International Review. Vol. 19, Iss. 3, p210-239

 Cutler, D.M., Poterba, J.M., Summers L.H. (1993) What Moves Stock Prices? Moves in Stock Prices Reflect Something Other Than News about Fundamental Values. Advances in Behavioral Finance. p133-151

 Dutta, S., Saadi, S., Zhu, P. (2013) Does Payment Method Matter in Cross-Border Acquisitions? International Review of Economics & Finance. Vol. 25, p91-107

 Durney, A.A., Amrita, S.N. (2007) Does Insider Trading Regulation Deter Private Information Trading? International Evidence. Pacific-Basin Finance Journal, Vol.15, Iss. 5, p409-433  Fama, E.F., French, K.R. (1992) The Cross-Section in Expected Stock Returns. Journal of

Finance. Vol. 47, Iss. 2, p427-465.

 Fama, E.F., French, K.R. (1993) Common Risk Factors in the Returns on Stocks and Bonds. Journal of Financial Economics. Vol. 33, Iss. 1, p3-56

 Fama, E.F. (1998) Market Efficiency, Long-Term Returns, and Behavioral Finance. Journal of Financial Economics. Vol.49, Iss. 3, p283-306

 Fuller, K., Netter, J., Stegemoller, M. (2002) What Do Returns to Acquiring Firms Tell Us? Evidence From Firms That Make Many Acquisitions. Journal of Finance. Vol. 57, p1763-1793  Haugen, R.A., Lakoniskhok, J. (1988) The Incredible January Effect, the Stock Markets’

Unsolved Mystery. Dow Jones-Irwin, Homewood, Illinois 60430

(40)

40  Jensen, M., Ruback, R. (1983) The Market for Corporate Control. Journal of Financial

Economics. Vol. 11, p5-50

 Keown, A.J., Pinkerton, J.M. (1981)Merger Announcements and Insider Trading Activity: An Emirical Investigation. Journal of Finance. Vol. 36, Iss. 4, p855-869

 Kim, O., Verrecchia, R.E. (1991) Market Reaction to Anticipated Announcements. Journal of Financial Economics. Vol. 30, Iss. 2, p273-309

 Kogut, B. (1983) Foreign Direct Investment as a Sequential Process. In C. Kindleberger and D. Andretsch (eds.), The Multinatioanl Corporation in the 1980s. Cambridge, MA: The MIT Press, p38-56

 Lloyd-Davies, P., Canes, M. (1978) Stock Prices and Publication of Second-hand Information. Journal of Business. Vol. 51, p43056

 Loughran, T., Ritter, J.R. (2000) Uniformly Least Powerful Tests of Market Efficiency. Journal of Financial Economics. Vol. 55, Iss. 3, p361-89

 Mathur, I., Rangan, N., Chhachhi, I., Sundaram, S. (1994) International Acquisition in the United States: Evendence from Return to Foreign Bidders. Managerial & Decision Economics. Vol. 15, Iss. 2, p107-118

 Moeller, S.B., Schlingemann, F.P., Stulz, R.M. (2004) Firm Size and the Gains from Acquisitions. Journal of Financial Economics. Vol. 73, Iss. 2, p201-228

 Moeller, S.B., Schlingemann, F.P. (2005) Global Diversification and Bidder Gains: A Comparison Between Cross-Border and Domestic Acquisitions. Journal of Banking and Finance. Vol. 29, p533-564

(41)

41  Myers, S., Majluf, N. (1984) Corporate Financing and Investment Decisions when Firms Have

Information Inverstors do not Have. Journal of Financial Economics. Vol. 87, p355-374

 Pound, J., Zeckhauser, R. (1990) Clearly Heard on the Street: The Effect of Takeover Rumors on Stock Prices. Journal of Business. Vol. 63, Iss. 3, p291-309

 Prevoo, T., Ter Weel, B. (2010) The Effects of a Change in Market Abuse Regulations on Abnormal Returns and Volumes: Evidence from the Amsterdam Stock Market. De Economist, Vol. 158, Iss. 3, p237-293

 Sanders Jr, R.W., Zdanowicz, J.S. (1992) Target Firm Abnormal Returns and Trading Volume around the Initiation of Change in Control Transactions. Journal of Financial and Quantitative Analysis. Vol. 27, Iss. 1, p109-129

 Schipper, K., Thompson, R. (1983) Evidence on the Capitalized Value of Merger Activity for Acquiring Firms. Journal of Financial Economics. Vol. 11, Iss. 1-4, p85-119

 Slaughter and May, report, The EU/UK Market Abuse Regime – Overview, 2011

 Uddin, M., Boateng, A. (2011) Explaining the Trends in the UK Cross-Border Mergers & Acquisitions: An Analysis of Macro-Economic Factors. International Business Review. Vol. 20, Iss. 5, p547-556

 Uysal, V.B., Kedia, S., Panchapagescan, V. (2008) Geography and Acquirer Returns. Journal of Financial Intermediation. Vol. 17, p256-275

(42)

42

Appendix 1

Country Code

Target Country Number of deals Civil-law Common-law Other1 AT Austria 2 2 AU Australia 12 12 BE Belgium 2 2 BG Bulgaria 1 1 BM Bermuda 2 2 BR Brazil 1 1 CA Canada 10 10 CH Switzerland 2 2 CL Chile 3 3 CN China 1 1 CO Colombia 2 2 CR Costa Rica 1 1 DE Germany 5 5 DK Denmark 3 3 DO Dominican Republic 1 1 ES Spain 6 6 ET Ethiopia 1 1 FR France 7 7 HK Hong Kong 2 2 IE Ireland 5 5 IN India 3 3 IT Italy 4 4 JP Japan 1 1 KE Kenya 1 1 KR Korea 2 2 KZ Kazakhstan 1 1 LU Luxemburg 1 1 NA Namibia 1 1 NL Netherlands 4 4 NO Norway 3 3 PE Peru 2 2 PH Philippines 1 1 PL Poland 1 1 PT Portugal 2 2 RU Russian Federation 2 2 SE Sweden 2 2 SK Slovakia 1 1 US United States 51 51 VG Virgin Islands 1 1 ZA South Africa 3 3 ZM Zambia 1 1 Total 158 64 84 10 1

(43)

43

Appendix 2

Descriptive statistics without winsorizing.

POSTAAR PREAAR DAYS SIZE MBRATIO DLAW DMETH DPUBLIC

Mean -9.21E-05 0.000772 136.6815 0.925451 4.649677 0.535032 0.439490 0.821656 Median 0.000482 0.000326 76.00000 0.066844 1.940000 1.000000 0.000000 1.000000 Maximum 0.007579 0.037025 728.0000 44.96419 249.3600 1.000000 1.000000 1.000000 Minimum -0.031055 -0.020589 3.000000 3.67E-05 0.130000 0.000000 0.000000 0.000000 Std. Dev. 0.003843 0.005515 150.4500 5.153676 20.36778 0.500367 0.497913 0.384027 Skewness -3.909947 2.618179 1.709190 8.027354 11.38361 -0.140473 0.243830 -1.680535 Kurtosis 29.79922 20.28546 5.586551 67.83137 136.3779 1.019733 1.059453 3.824197 Jarque-Bera 5098.241 2133.935 120.2069 29181.46 118239.3 26.16921 26.18979 78.34358 Probability 0.000000 0.000000 0.000000 0.000000 0.000000 0.000002 0.000002 0.000000 Sum -0.014460 0.121241 21459.00 145.2958 720.7000 84.00000 69.00000 129.0000 Sum Sq. Dev. 0.002304 0.004744 3531090. 4143.419 63886.39 39.05732 38.67516 23.00637 Observations 157 157 157 157 155 157 157 157

Descriptive statistics winsorizing 1% => 0,5%

POSTAAR_U PREAAR_U DAYS_U SIZE_U MBRATIO_

U

DLAW DMETH DPUBLIC

(44)

44

Descriptive statistics winsorizing 5% => 2,5% POSTAAR_

T

PREAAR_T DAYS_T SIZE_T MBRATIO_ T

DLAW DMETH DPUBLIC

(45)

45

Appendix 3 Hypothesis testing

Hypothesis testing Post Average Abnormal Return Test of Hypothesis: Mean = 0.00000

Included Observations: 157 Without winsorizing Winsorizing 1% percentile Winsorizing 2% percentile Winsorizing 5% percentile Winsorizing 6% percentile Winsorizing 10% percentile

Sample Mean -9.21e-05 -6.37e-05 1.06e-05 5.78e-05 6.92e-05 0.000156

Sample Std. Dev.

0.003843 0.003610 0.003128 0.002785 0.002699 0.002306

T-statistic -0.300298 -0.221008 0.042380 0.260041 0.321511 0.850145

Probability 0.7644 0.8254 0.9662 0.7952 0.7483 0.3965

Hypothesis testing Pre Average Abnormal Return Test of Hypothesis: Mean = 0.00000

Included Observations: 157 Without winsorizing Winsorizing 1% percentile Winsorizing 2% percentile Winsorizing 5% percentile Winsorizing 6% percentile Winsorizing 10% percentile Sample Mean 0.000772 0.000770 0.000764 0.000537 0.000523 0.000504 Sample Std. Dev. 0.005515 0.005347 0.004958 0.003611 0.003476 0.002874 T-statistic 1.754553 1.804278 1.929772 1.865224 1.886981 2.197754 Probability 0.0813 0.0731 0.0554 0.0640 0.0610 0.0294

Hypothesis testing Difference Post Average Abnormal Return And Pre Average Abnormal Return Test of Hypothesis: Mean = 0.00000

(46)

46

Appendix 4: OLS Regression Winsorizing 1%

Step 1

Dependent Variable: POSTAAR_U Method: Least Squares

Date: 02/06/13 Time: 14:30 Sample: 1 157

Included observations: 157

Variable Coefficient Std. Error t-Statistic Prob. PREAAR_U -0.033181 0.054295 -0.611131 0.5420 DAYS_U -1.67E-06 1.93E-06 -0.862213 0.3899 C 0.000189 0.000395 0.479476 0.6323 R-squared 0.006869 Mean dependent var -6.37E-05 Adjusted R-squared -0.006029 S.D. dependent var 0.003610 S.E. of regression 0.003621 Akaike info criterion -8.385016 Sum squared resid 0.002020 Schwarz criterion -8.326616 Log likelihood 661.2237 Hannan-Quinn criter. -8.361298 F-statistic 0.532558 Durbin-Watson stat 1.785913 Prob(F-statistic) 0.588179

Step 2

Date: 02/06/13 Time: 14:30 Sample: 1 157

Variable Coefficient Std. Error t-Statistic Prob. PREAAR_U -0.034524 0.053867 -0.640913 0.5226 DAYS_U -1.15E-06 1.92E-06 -0.596330 0.5519 SIZE_U -0.000164 5.65E-05 -2.906860 0.0042 MBRATIO_U 7.75E-06 1.81E-05 0.429174 0.6684 C 0.000242 0.000403 0.600828 0.5489 R-squared 0.059915 Mean dependent var -6.13E-05 Adjusted R-squared 0.034846 S.D. dependent var 0.003634 S.E. of regression 0.003570 Akaike info criterion -8.400836 Sum squared resid 0.001912 Schwarz criterion -8.302661 Log likelihood 656.0648 Hannan-Quinn criter. -8.360959 F-statistic 2.390015 Durbin-Watson stat 1.775745 Prob(F-statistic) 0.053381

Step 3

(47)

47

Variable Coefficient Std. Error t-Statistic Prob. PREAAR_U -0.033908 0.054685 -0.620058 0.5362 DAYS_U -1.61E-06 1.96E-06 -0.822015 0.4124 SIZE_U -0.000157 5.69E-05 -2.758608 0.0065 MBRATIO_U 1.02E-05 1.80E-05 0.565256 0.5728 DLAW -0.001005 0.000591 -1.699396 0.0914 DMETH 0.000538 0.000599 0.897791 0.3708 DPUBLIC 0.000361 0.000749 0.482718 0.6300 C 0.000288 0.000893 0.322660 0.7474 R-squared 0.088340 Mean dependent var -6.13E-05 Adjusted R-squared 0.044927 S.D. dependent var 0.003634 S.E. of regression 0.003551 Akaike info criterion -8.392829 Sum squared resid 0.001854 Schwarz criterion -8.235749 Log likelihood 658.4442 Hannan-Quinn criter. -8.329026 F-statistic 2.034895 Durbin-Watson stat 1.799063 Prob(F-statistic) 0.054402

Appendix 5: robustness test

Robustness test descriptive statistics for Pre Average Abnormal Return Without winsorizing Winsorizing; 1%

percentile

Winsorizing; 5% percentile

PREAAR PREAAR_S PREAAR_T

Mean 0.000764 0.000760 0.000707 Median 0.000687 0.000687 0.000687 Maximum 0.016310 0.015380 0.008255 Minimum -0.008599 -0.008360 -0.004569 Std. Dev. 0.003044 0.003010 0.002437 Skewness 1.291811 1.205763 0.457312 Kurtosis 9.820026 9.220184 4.341007 Jarque-Bera 347.9372 291.1444 17.23623 Probability 0.000000 0.000000 0.000181 Sum 0.119986 0.119295 0.110964 Sum Sq. Dev. 0.001446 0.001413 0.000927 Observations 157 157 157

Robustness test OLS winsorizing 1%

Step 1

Dependent Variable: POSTAAR_S Method: Least Squares

Date: 02/06/13 Time: 15:58 Sample: 1 157

(48)

48

PREAAR_S -0.009165 0.096351 -0.095119 0.9243 C -5.67E-05 0.000298 -0.190208 0.8494 R-squared 0.000058 Mean dependent var -6.37E-05 Adjusted R-squared -0.006393 S.D. dependent var 0.003610 S.E. of regression 0.003622 Akaike info criterion -8.390920 Sum squared resid 0.002033 Schwarz criterion -8.351987 Log likelihood 660.6873 Hannan-Quinn criter. -8.375108 F-statistic 0.009048 Durbin-Watson stat 1.774045 Prob(F-statistic) 0.924343

Step 2

Date: 02/06/13 Time: 15:58 Sample: 1 157

Variable Coefficient Std. Error t-Statistic Prob. PREAAR_S -0.006528 0.095245 -0.068537 0.9454 DSIZE_S -0.000167 5.62E-05 -2.972711 0.0034 MBRATIO_S 8.34E-06 1.80E-05 0.461935 0.6448 C 6.37E-05 0.000307 0.207631 0.8358 R-squared 0.055348 Mean dependent var -6.13E-05 Adjusted R-squared 0.036580 S.D. dependent var 0.003634 S.E. of regression 0.003567 Akaike info criterion -8.408892 Sum squared resid 0.001921 Schwarz criterion -8.330352 Log likelihood 655.6892 Hannan-Quinn criter. -8.376991 F-statistic 2.949065 Durbin-Watson stat 1.763504 Prob(F-statistic) 0.034700

Step 3

Date: 02/06/13 Time: 16:05 Sample: 1 157

(49)

49