1
The Effect of Failed Takeover Attempts on Targets:
the Difference Between Europe and North America
Name Maurice Bernards
Student Number 10243801
Programme Economics & Business Programme code BSc ECB
Track Finance & Organisation Name Supervisor Tolga Caskurlu
Date July 13th, 2015
Abstract
The goal of this thesis is to analyze the effects on target firm revaluation in a case of a failed takeover attempt. These effects are compared to the effects found in the North American market for M&A in related literature. Moreover, Europe is split up in the UK on the on hand and the rest of Europe on the other hand to determine the effect within Europe.
The sample in this thesis consists of 295 failed takeover attempts in the period 2000-2015. To determine individual effects, the sample is split up in cash and share bids first. Consequently, the sample is split up in UK and the rest of Europe. The effects on the target revaluation are studied using various event windows. Lastly, the cumulative abnormal return of the targets is regressed in two regressions against cash bids and location, separately.
The results suggest that there is strong evidence for a higher revaluation when cash bids are used, compared to share bids. Evidence for the location effect is weak, suggesting that the location has a weak effect on the target revaluation. The overall results suggest that the magnitudes of the effects in Europe are different from the magnitudes found in North America. However, the relationships between the variables do show resemblance to the relationships found in North America.
2
Statement of Originality
This document is written by Student Maurice Bernards who declares to take full responsibility for the contents of this document.
I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it.
The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.
3
Contents
1 Introduction ... 5
2 Related Literature ... 7
2.1 The abnormal return of the target after a takeover announcement ... 7
2.2 The effect of a failed takeover attempt ... 8
2.3 The difference between cash and share bids ... 9
2.5 The difference between the Europe and the North America, and within Europe, in the market for M&A ... 11
3 Hypotheses ... 13
3.1 Hypothesis 1 ... 13
3.2 Hypothesis 2 ... 13
4 Data & Model ... 14
4.1 Data ... 14
4.2 Model ... 15
4.2.1 Cumulative Abnormal Return (CAR) 15 4.2.2 Regression analysis 15
5 Results ... 17
5.1 General results ... 17
5.2 Event windows ... 18
5.2.1 Overall sample 19 5.2.2 Cash versus Share Bids 19 5.2.3 UK versus the rest of Europe 21 5.3 Regression analysis ... 22
6 Discussion & Conclusion ... 26
References... 28
4
Appendix 2 ... 31
Appendix 3 ... 33
5
1 Introduction
This thesis analyzes the European market for mergers and acquisitions (M&A), more specifically, target firm revaluation after a failed takeover attempts. However, much of the existing research into this subject has focused on the North American market for M&A. As the two markets are fundamentally different from each other (laws, institutions, integration, etc…), the results and conclusions drawn from the existing literature based on the North American market cannot be automatically applied to the European market. Even within the European market, there are many differences. The United Kingdom (UK) for example, shows more resemblance to the market in North America, than the rest of Europe. Previous literature has found large effects from M&A (failed) takeover attempts on target firms, and effects on the bidder firm are small, statistically insignificant or even non-existent. Consequently, this thesis will focus on the effect of the targets in the European market.
This brings me to the research question, “Are the consequences of failed takeover attempts on target firms the same in Europe as they are in North America?” This research question will be answered using two hypotheses. The first one, stating that, compared to share bids, cash bids make for a higher and positive revaluation of the target. In other words, the cumulative abnormal return at the end of the event window is positive and is higher for cash bids than for share bids. The second hypothesis states that if the target is located in the UK, the revaluation is higher than in other European countries. These two hypotheses will be answered by comparing the existing research to a dataset of European failed takeover attempts.
When looking at the results of North America, there is a big difference in whether there is a cash bid or a share bid. This also holds for Europe, as we see a big difference between the two. There are various reasons for the difference between cash and stock bids. First one is the fact that with cash bids, investors are subject to direct capital gain taxation. Therefore, they will demand a higher direct return than with a share bid, where the tax obligations are deferred. Second, for financially constrained bidders, the cash takeover attempts for targets with a high (re-)valuation lead to a takeover failure, and ultimately to an over proportional failure of cash bids in the sample. The last reason has to do with two effects combined, namely that cash bids are often used in smaller deals, and that smaller targets convey more new information to the market, compared to larger targets. These two combined can explain that cash bids inherently yield a higher return. The magnitudes of the difference between cash and share bids in Europe, however, do differ from the magnitudes seen in North America. The effects discovered in the research in the North American markets do partially hold for the
6 European market, however the magnitudes are different, as are the underlying reasons for the effects. For cash bids, for example, the magnitude is comparable to that in North America. Another important finding is that the relative firm size has a positive impact on the cumulative abnormal return, and certain periods within the sample as well.
The second hypothesis states that the location, within Europe, also has an impact on the revaluation of the target. The reason for this distinction is that the UK market shows more resemblance to the North American market, instead of the European market. The results show that a target located in the UK is revaluated higher compared to a target located in the rest of Europe, even when controlling for various other effects. However, the evidence for this effect is weak.
The two hypotheses will attempt to explain the difference between the European and the North American market, and even the differences within the European market. This thesis will begin with discussing the related literature on the subject (Part 2). First, related literature concerning the effect of takeovers on target revaluation and the effect of failed takeover attempts on target firms will be discussed. Consequently, the literature is reviewed for each hypothesis, followed by an overview of the two hypotheses (Part 3). This section will be followed by the description of the data and the models that are used to analyze the data (Part 4). In part 5, the results are discussed, followed by part 6, which ends with a discussion and a conclusion. The references to this thesis are found at the end.
7
2 Related Literature
2.1 The abnormal return of the target after a takeover announcement
The effect of a takeover announcement has been discussed in various papers, dating back to 1968. In his paper, Block (1968, pp. 890) describes the effect of a takeover announcement for both bidders and targets. In this paper, he compares 35 takeovers, and compares the performance of the firms involved in the merger to a control group of similar firms. For the bidders he found no evidence that the price increase was significantly different from the firms in the control group (the relative price increase for the stock of the bidder firms is 20,94%, compared to 17,28% in the control group). On the contrary, the share prices of the target firms show a different pattern. The difference, however, does not show until three months before the announcement. The largest increase in share price is in the period between one month before the announcement and the announcement. In this period, the share prices of the targets show an increase of 17,01%, compared to 3,08% in the control group. Over the whole time period (nine months before the announcement until the consummation of the target), the share price increase is 35,26% for the target firms and 10,50% for the control group. One explanation that Block gives for what he calls the “merger effect” for targets is the premium paid, which should explain the upward share price movement.
In a paper by Jensen and Ruback (1983, pp. 10), amongst others, the abnormal returns for target firms after a takeover announcement are also discussed. In a summary of thirteen studies they found out that targets (of successful takeover attempts) realize a large and significant increase in their share prices. For announcement of mergers that turn out to be successful mergers, they found that the weighted average abnormal return is 29,1% and that half of the abnormal return associated with the announcement occurs before the announcement. For the mergers that turn out to be unsuccessful (failed takeover attempt) the average abnormal return to shareholders of the target firm is 35,2%. This suggests that the market expects approximately the same abnormal return for both successful and unsuccessful takeover attempts at the time of the announcement.
Huang & Walkling (1987, pp. 331) based their paper on the effect of takeover announcements and focused on the target-firm abnormal returns. However, they distinguish between tender offers and normal mergers. The sample period is t-50 until t+50, where t is the announcement day. They found that from day t-50 to day t-2, the cumulative abnormal return (CAR) of the target firm is 9,1%. The last day before the announcement (t-1), the cumulative abnormal
8 return is 23,4%. In their research, they found that the announcement effect is greater for tender offers (CAR is 27,5%) than for mergers (CAR is 22,6%). They also compared the effect of the announcement between cash and share bids. Here they found that, at the announcement, the CAR of cash bids is higher than the CAR of share bids (29,3% versus 14,4%, respectively).
The above papers are all based on data from North America. A paper by Goergen & Renneboog (2004, pp. 9) investigates the effects on the European market, and makes a clear distinction between the UK and the rest of continental Europe. In their overall sample and the overall time period, the CAR of the bidding firms is 21,66%. They do, however, split the time period up into various smaller time periods, such as the merger announcement day itself (day -1 and 0), a slightly larger announcement period (-2 until +2), and the period leading up to the announcement (-40 until 0). The resulting CARs are 9,01%, 12,96%, and 23,10% respectively. The difference between 23,10% (-40, 0) and 21,66% (-60, +60) can be explained by the fact that, either the takeover attempt failed, or the longer a takeover attempt lasts, the more doubt is raised about its ultimate success.
2.2 The effect of a failed takeover attempt
When it comes to examining the effect of failed takeover attempts, among the most cited papers is the paper by Dodd (1980, pp. 109). The results indicate that, as described in the papers discussed above, that on the day of the announcement and the day before, target shareholders earn a large positive abnormal return, and this also counts for failed takeover events. However, the announcement of the failure of the takeover yields the target firm a negative abnormal return of -4,52% on the announcement day (day 0) and -4,16% the day after (day 1). As the stock exchange might process the news of the failed takeover attempt before the actual announcement day, the two-day abnormal return is also relevant. In tender offers, the effect of a failed takeover attempt is harder to isolate to one date, as the shareholder response is released gradually to the market.
Davidson et al. (1989, pp. 1077) re-examine the market reaction to failed takeover attempts, and base their paper on the paper by Dodd (1980, pp. 109). They found that the revaluation as stated by Dodd (1980, pp. 109) disappears when targets are not involved in subsequent takeover activities, but it does persist for targets that do continue with takeover activities. There is also a distinction made between who cancels the takeover attempt. When the target cancels, and remains active in takeovers, there is a positive CAR. Even on the cancellation
9 date there is a positive abnormal return. When targets cancel but do not take part in future takeover activities, the cancellation causes a negative abnormal return. The firm value then returns to pre-merger announcement levels. The same conclusions are reached when the bidder cancels.
The paper by Goergen & Renneboog (2004, pp. 30) also discusses the effect of failed takeover attempts, and comparing them to successful ones. As mentioned before, they split up their time period into various smaller time periods. In the overall time period (-60, +60) the target CAR of failed takeover attempts is much higher than target CAR of successful takeovers, 25,02% versus 20,58%. On the announcement day (-1,0), however, the CAR of unsuccessful bids is much lower than the CAR of the successful bids, 5,51% versus 10,30%. Overall, the target CAR of the failed takeover attempts does not revert to its original level from before the announcement. A reason for this is that the share price of the target still contains a merger premium, indicating a possible follow-up bid in the near future.
In their results, Malmendier, Opp, and Saidi (2012, pp. 11) find that small targets have a higher revaluation (higher CAR). The reason they give for this is that a bid may contain useful and new information about a target. When the target is small, there might be more new information compared to bigger targets.
2.3 The difference between cash and share bids
The recent paper by Malmendier, Opp, and Saidi (2012, pp. 11) examines the difference between cash and share bids on the revaluation of the target after a failed takeover attempt. When comparing both types of bids over the whole period (25 days before the takeover announcement until 25 days after the cancellation announcement), they see a strong cumulative announcement return for both cash and share bids, with 25% and 15% on average, respectively. This is a similar result as in earlier studies. At the time of the cancellation, however, the average share bid CAR has returned to its original level (CAR=0), but the average cash bid CAR remains at 15% above its original level. The remaining days both types of bids experience a slight upward trend, but share bids remain, on average, more than 15% below the cash bids. Next, they turn to a controlled regression framework. Without any controls, the cash coefficient on the target CAR is 22,1% in pure-deals (cash-only deals or share-only deals). When controlling with various variables for deal- and entity-specific characteristics, the cash coefficient in the pure-deals remains roughly the same, at 22,6%. They do not, however, find an explanation for why the revaluation of cash bids is so much
10 higher than share bids. An explanation they offer is that the choice of cash versus share payment is correlated with the failure of a takeover attempt. However, when comparing successful deals with unsuccessful ones, they discover that there is no significant difference between cash and share bids among the two. Another explanation they offer is the effect of revaluation of the target on the bidder. With good news, for example, the (financially constrained) bidder might not be able to increase the cash bid. Share bids, however, do not have this problem. Therefore, the cash takeover attempts for targets with a high (re-)valuation lead to an over proportional failure of cash bids.
The paper by Goergen & Renneboog (2004, pp. 28) also examines the effect of the payment form on the CAR of the target. In all time periods (explained above) used in their research, they find significant results that cash bids create higher CARs compared to share bids (or a combination of both). Over the whole period, they find a difference of 15,86% between cash and share bids. This result is consistent with Malmendier, Opp, and Saidi (2012, pp. 11). It is important to note that the magnitude of difference are similar, however, the time horizon of Malmendier, Opp, and Saidi (2012, pp. 11) is 25 days, whereas Goergen & Renneboog (2004, pp. 28) use 60 days. The same difference is also seen in the time period (-40,0), which is also consistent with the findings of Malmendier, Opp, and Saidi (2012, pp. 11). Furthermore, they argue that payment for smaller targets is mostly done in cash. In their sample, the average value of the pure cash bids is USD1,489 million, compared to USD14,225 million for the pure share bids.
Travlos (1987, pp. 944) argues that tax implications is a very important reason for the difference in returns between cash and share bids. As opposed to share bids, cash bids generate immediate tax obligations for the shareholders of the target. Any capital gains made by the target shareholders in case of a cash bid is subject to tax immediately, whereas the tax obligation for share bids is deferred to when the shares are sold. Due to this difference, the bidder must pay a higher price if it uses a cash bid to offset the tax burden. As this argument is based on the law in North America, it cannot be directly assumed for Europe. However, when looking at the Merger Directive (2009) of the EU, it can be seen that such laws also exist in the EU. For the UK this also holds, as stated in HR Revenue & Customs help sheet 297 (2014).
In all the M&A literature that reveals the difference between cash and share bids, there is no clear motive for the substantial difference for the target CAR between cash and share bids.
11 The two most stated reasons are the tax effect and a selection bias in the sample of unsuccessful takeover attempts.
2.5 The difference between the Europe and the North America, and within Europe, in the market for M&A
Looking at the market for mergers and acquisitions, there a multiple differences between the Europe and North America, and even within Europe there are difference, as stated by Moschier and Campa (2009, pp. 72).
One contrasting characteristic between the European and the North-American market is that European takeovers tend to be domestic, friendly and privately arranged, whereas the North-American takeovers are more hostile. Hostile bids are very rare in Europe (1%), although it does increase in more recent deals. The cause of the low proportion of hostile bids is the concentrated ownership in European firms, which facilitates friendly transfer of control, not hostile. Another difference is the absence of competing bids, which is also a result of few hostile bids. The likelihood of a competing bid to be completed is much lower than a non-competing bid (43% versus 64%). In Europe, most bids are cash bids, which is similar to the US. The difference, however, is that in Europe the hostile deals are mostly cash, whereas in North America cash bids are more common for non-hostile bids. A similarity between Europe and North America is that large deals are paid in shares (or a combinations of shares and cash).
As stated by Moschier and Campa (2009, pp. 73), one important difference between countries within Europe is the legislation. The European commission has tried to change and harmonize the European market for M&A by the Takeover Directive. This directive is far from ideal, as it leaves controversial provisions optional for all member states. Harmonization has been a high priority of the European Commission, and it now has a final say in all takeovers in the EU. In general, most of the deals are domestic, 81% versus 19% cross-border (Moschier and Campa. 2009, pp. 76). However, cross-border deals are becoming more important, as the industry consolidation reduces viable opportunities in the home country. Cross-border deals within the EU (in contrast to outside the EU) are attractive to firms within an EU-member state, due to the single currency and an integrated market and legislative network.
The UK is also very different compared to the rest of Europe. According to Moschier and Campa (2009, pp. 77), “British firms have a more decentralized financial decision-making process (…), better accounting standards and stronger shareholder protection (…) and operate
12 in a more active and competitive market (…) than companies in the other EU-15 countries.” Compared to firms in other countries, the UK firms also have a dispersed ownership, whereas firms in the rest of Europe (such as Germany, France, Italy) ownership is mostly concentrated to two shareholders that control the management of the firm. Hostile takeover attempts are more common in the UK, due to the ownership structure of firms. Another difference between the UK and the rest of Europe is that, in the UK, the number of cash bids is higher. This, however, may be linked to a trend in which cash-only transactions increased from 2001 to 2007 (Moschier and Campa. 2009, pp. 81). Goergen & Renneboog (2004, pp. 23) also examine the difference between the UK and the rest of continental Europe. Due to the reasons mentioned above (difference between UK and the rest of Europe) they expected a higher CAR in the UK. They find that, in their total time period (-60, +60), UK targets have a CAR of 29,32%, compared to 14,82% to the rest of continental Europe. The largest difference is in the 40 days before the announcement and the announcement day (-40,0). Here, the CAR of the UK targets is 38,30%, compared to 14,95% in continental Europe.
In the attempt to discover the difference between Europe and the US, the difference between the UK and the rest of Europe is examined. In addition, it is also important to find the difference between the UK and North America. Goergen and Renneboog (2004, pp. 11) found various studies that found significant difference in the effects of target revaluation between the US the UK. These differences, 40% and 26% for North American targets, compared to 18 and 16% for UK targets, are substantial.
13
3 Hypotheses
Based upon the previous literature and the formulated research question, the following two hypotheses can be formulated.
3.1 Hypothesis 1
As described above, in all cases in the research based on the North American market for M&A, target revaluation after an unsuccessful cash bid is higher than with a share bid. Looking at previous literature from North America and Europe, it becomes clear that the revaluation (CAR) of cash bids is between 15 and 30 per cent, depending on what interval in the total time period used.
There is no one clear reason for this phenomenon. Travlos (1987, pp. 944) argues that tax implications is a very important reason for the difference in returns between cash and share bids. Malmendier, Opp, and Saidi (2012, pp. 11) argue it is due to a selection bias.
H1 Compared to share bids, cash bids make for a higher and positive cumulative abnormal return for the target.
3.2 Hypothesis 2
The market for M&A in the UK is more closely related to the market in North America, rather than the market for the rest of Europe. Therefore, in order to determine whether the effects of the research based in North America also holds in Europe, it is important to make a difference between the UK and the rest of Europe. There are various reasons for the difference between the UK and the rest of Europe. The most important ones are the stronger shareholder protection, better accounting standards and the market is more active and competitive. For these reasons, the expected CAR for the target is higher than in the other European countries. H2 If the target is located in the UK, the cumulative abnormal return for the target is
14
4 Data & Model
4.1 Data
The data about the failed takeover attempts is collected from Thompson ONE and dates from 2000 until 2015. There were various filters used to retrieve the data, namely deal status: Withdrawn, Method of payment: Cash or Shares, Domestic deals: European Union, Deal type: Merger and Acquisition, and listed target. The last criterion is crucial for analyzing the stock data. Furthermore, I excluded several cases from the data set, such as deals that were not solely financed with cash or shares, an unknown announcement/withdrawal date, if the deal value was below €20 million, non EU-domestic deals (EU to/from outside the EU), more than 250 trading days between the announcements (announcement of the M&A, announcement of withdrawal), consecutive failed bids (this makes filtering the effect of one specific bid not possible). After setting the filters and excluding several cases, the dataset consists of 295 failed takeover attempts in Europe. The stock prices and market indices are collected from Datastream.
Table 1
Panel A shows the summary statistics for the full sample. Panel B distinguishes the total observation between payment method and location of the target.
Source: Sample data
Panel A: Summary Statistics
Variable Obs. Mean Std. Dev. Min Max
CAR 295 0,2060 0,3487 -1,0153 1,4724
Firm size (x 1m) 295 € 1.616,95 € 3.589,49 € 8,38 € 28.511,91 Deal value (x 1m) 295 € 1.583,41 € 3.769,60 € 20,14 € 30.500,32 Days between announcements 295 53,49 43,61 1 236
Panel B: Observations for payment method and separation by country
Payment method
Cash 203 Shares 92
Countries (Target)
United Kingdom 122 France 12
Nordic1 56 Benelux 12
Southern Europe2 42 Switzerland 10
Germany 13 Other 28
1
15
2
Southern Europe: Cyprus, Greece, Spain, Portugal, Italy
The total event window in the sample is 25 days before the announcement of the takeover until 25 days after the announcement of the failure of the takeover. The reasoning behind the use of 25 days for the run-up period is based on the paper of Malmendier, Opp, and Saidi (2012, pp. 5) that use the findings of Schwert (1996), where it is said that “run-ups do not occur until 21 days before the announcement.” Malmendier, Opp, and Saidi (2012, pp. 5). In the same paper they argue that including deals that take more than 250 days to the announcement of failure are not to be included. If no upper bound is imposed than there is a chance that information is also taken into account that is unrelated to the offer.
4.2 Model
4.2.1 Cumulative Abnormal Return (CAR)
For the first part, the following model is used to compute the cumulative abnormal return:
𝐶𝐴𝑅𝑖𝑡 = ∑(𝑟𝑖𝑗− 𝑟𝑚𝑗) 𝑡
𝑗=1
,
where:
𝐶𝐴𝑅𝑖𝑡 Cumulative abnormal return of firm 𝑖 𝑟𝑖𝑗 Return of firm 𝑖 at time 𝑗
𝑟𝑚𝑗 MSCI Europe value-weighted market return 𝑚 at time j
The cumulative abnormal return is calculated on a daily basis. The total CAR is the cumulative abnormal return of firm 𝑖 at the 25th day after the announcement of failure.
Including smaller event windows within the total event window, as done by Goergen and Renneboog (2004, pp. 19), will yield a more complete picture of the effect of a failed takeover attempt. The time windows used will be (T-25, F+25), (T-25, T),(T-1,T), (T,F), (F, F+1) (F, F+25), where T is the announcement of the takeover and F is the announcement of the failure of the takeover. In the overview, the mean and the t-value of each instance will be calculated. The t-value will show whether the mean is different from zero.
4.2.2 Regression analysis
For the regression, a linear regression will be performed. To test the first hypothesis, a regression will be run with solely the dummy variable for Cash or Share bids. To test the last
16 hypothesis, the dummy of the location variable, taking 1 as UK and 0 for the rest of Europe, will be regressed against the CAR.
To test the first hypothesis, the regression, in formula form, will look as follows: 𝐶𝐴𝑅 = 𝛽0 + 𝛽1𝐷𝑐𝑎𝑠ℎ+ 𝛽2𝑋1+ ⋯ + 𝛽𝑖+1𝑋𝑖
First, the regression will be done using only the variable Cash. This will function as a benchmark, for more regressions will be done, including control variables for relative size, ln(Firm value), and period. Dummies will be used to indicate the time periods as in table 2, with period 2003-2007 as the omitted variable. As stated by Moschier and Campa (2009, pp. 82), Goergen and Renneboog (2004, pp. 24) claim cash is generally used for smaller targets. Therefore, the relative firm value is also important to include as a control variable. Moreover, as relative deal value will also be included as a control variable. As the relative deal value and relative firm value have a high correlation (large firms often require large takeover bids), an interaction term will be included, namely ln(firm value) multiplied by ln(deal value). Furthermore, as stated by Moschier and Campa (2009, pp. 78), share bids take longer to complete due to approval by authorities. Therefore, the relative difference from the average completion time of the sample will be included. Just like with the relative deal value and relative firm value, an interaction term will be used for cash and relative difference from the average completion time as they are also correlated (cash deals are often completed quicker compared to share bids).
To test the second hypothesis, the regression, in formula form, will look as follows: 𝐶𝐴𝑅 = 𝛽0+ 𝛽1,2𝐷𝑙𝑜𝑐𝑎𝑡𝑖𝑜𝑛+ 𝛽2𝑋1+ ⋯ + 𝛽𝑖+1𝑋𝑖
As with the first hypothesis,first, a regression with only the variable for location will be done, this regression will also function as a benchmark. Here, too, control variables will be added to control for various other effects. As shown by Moschier and Campa (2009, pp. 81), in the UK cash bids are used more often than the European average (30% versus 33%). Even though this difference is not that great, it is important to note that in their sample the UK is overrepresented and pulls the average of Europe up. Therefore, it is important to include cash as a dummy as a control variable. Furthermore, Goergen and Renneboog (2004, pp. 38) find that domestic takeovers yield a higher CAR compared to cross-border takeovers. Combined with the more efficient market and higher expected CAR for the UK market for M&A, it is also important to include a variable for domestic. Just as with the first hypothesis, it is also
17 important to control for the time period. A variable for the period 2005-2007 will also be included.
5 Results
5.1 General results
The initial sample consisted of 371 unique cases. However, various restrictions that were put on the dataset reduced the sample to 295. The first restriction, as mentioned before, is that the time between the merger announcement and the failure announcement should be at most 250 days. Other restrictions are ownership sought (+50%, as stated by Malmendier, Opp, and Saidi (2012, pp. 5)), deal size (+€20 mil), date (later than 2000), available equity values (total equity value and share price through Datastream), and pure cash or share bids. In the table below, there are various characteristics of the sample used in the analysis.
Table 2
Sample Characteristics.
This table shows the sample characteristic in periods of four years, starting at the year 2000 and ending at 2015. Panel A shows the total and average deal value. Panel B shows the amount of observation, both in absolute amounts and as a percentage of the total amount, and average deal size per payment method. Panel C displays the total amount of failed takeover attempts per country/region.
Source: Sample data
2000-2003 2004-2007 2008-2011 2012-2015 Total
Sample (Obs.) 74 112 66 43 295
Panel A: Deal value characteristics
Deal value (DV)
Total (x 1.000m) € 59,607 € 249,119 € 82,305 € 31,075 € 467,105 Average (x 1.000m) € 0,806 € 2,626 € 1,247 € 0,723 € 1,583
Panel B: Characteristics by payment method
Payment method
Cash (Obs.) 52 78 48 25 203
% of total 70% 70% 73% 58% 69%
18
Shares (Obs.) 22 34 18 18 92
% of total 30% 30% 27% 42% 32%
Average DV (x 1.000m) € 1,167 € 3,378 €2,862 € 0,783 € 2,241
Panel C: Failed takeovers by country
Location (Target) United Kingdom 23 51 32 16 122 Nordic 25 16 13 5 59 Southern Europe 13 20 3 6 42 Benelux 0 5 6 2 13 Germany 0 6 1 4 11 Switzerland 3 4 1 2 10 France 4 2 1 2 9 Other 6 8 9 6 29
Table 2 describes the main characteristics of the sample. As can be seen, the majority of the deals are cash bids, consistent with various literatures about the North American M&A market. An interesting finding here, also consistent with the remarks of Moschier and Campa (2009, pp. 82), during times of a recession (and of low M&A activity), bids tend to move away from share bids, and more into cash bids. The reason for this is that during periods of high stock prices (i.e. not in a recession), firms tend to move more to share bids, as they expect the shares to increase in value. During recessions, however, there is a tendency to move away from share bids. This effect can be seen in the third period, from 2008-2011. In this time period, bids were more done with cash, and less with shares, compared to the other periods. In the last period, from 2012-2015, the stock markets climbed up again. This could explain the large amount of share bids, compared to the cash bids. The boom in M&A activity is also evident from table 2. In the second period, from 2004 – 2007, the increase in the amount of failed takeover attempts compared to the period before (2000 – 2003) is 51,35%. Also, the total and average deal value in the same period is substantially higher compared to the other periods.
5.2 Event windows
19 5.2.1 Overall sample
Table 3
Cumulative abnormal returns of the target.
This table shows the cumulative abnormal return for six of the event windows used.
Source: Calculations in Appendix 1 Event window Overall sample Mean t-value (T-25, F+25) 0,206024 10,13126 *** (T-25, T-2) 0,067063 5,49806 *** (T-1,T) 0,112909 13,92190 *** (T,F) 0,044245 3,29249 *** (F, F+1) -0,010710 -1,79439 * (F+2, F+25) -0,007704 -0,92538 Obs. 295
*/**/*** denotes significance at a 10%/5%/1% level, resp.
Table 3 shows several event windows from the whole sample. On the day of the announcement of the takeover and the day before (T-1, T), an abnormal return of 11,29% is realized, compared to 6,71% which is realized over the whole month prior to the takeover announcement, excluding the last two days of the announcement. A reason could be that rumors of the takeover drove up the share price even before it was publically announced. In between the two announcements, a positive abnormal return of 4,42% is realized. All these values are statistically significant at a 1% significance level. On the day of the announcement of takeover failure, and the day after (F, F+1), a small negative abnormal return is realized, as in the month following the failed takeover attempt (F+2, F+25). These returns are -1,07% and -0,77% respectively, however, only the event window (F, F+1) is statistically significant at a 5% level.
5.2.2 Cash versus Share Bids
Table 4
Cumulative abnormal returns of the target (cash versus share bids).
This table shows the cumulative abnormal return for six of the event windows used, distinguished by cash and share bids.
20 Source: Calculations in Appendix 2
Event window
Cash Shares Difference
Mean t-value Mean t-value Mean t-value
(T-25, F+25) 0,2437 9,936 *** 0,1228 3,488 *** 0,1209 2,791 *** (T-25, T-2) 0,0819 5,037 *** 0,0343 2,250 ** 0,0477 1,821 * (T-1,T) 0,1218 11,734 *** 0,0933 7,631 *** 0,0285 1,633 * (T,F) 0,0645 4,083 *** -0,0005 -0,020 0,0650 2,260 ** (F, F+1) -0,0142 -1,684 * -0,0030 -0,659 -0,0111 -0,866 (F+2, F+25) -0,0113 -1,279 0,0003 0,014 -0,0116 -0,644 Obs. 122 173
*/**/*** denotes significance at a 10%/5%/1% level, resp.
In table 4 the difference between cash and share bids in various event windows is made apparent. An important note is that in the sample of failed takeover events, there are more cash bids than share bids. And, conform Goergen and Renneboog (2004, pp. 24) and as seen in table 2, the share bids are relatively large in deal size compared to cash bids. Over the whole time period, the average cash deal value was €1.285 million, compared to the average share deal value of € 2.240 million. Over the whole time period, and especially in the days before the announcement of the takeover, there is strong evidence that the target share price increases more with cash bids. Over the whole period, this difference is 24,37% compared to 12,28% (means are significantly different at 1%). In the event windows for the days before the announcement, this effect is also apparent. In the even window (T-25, T-2) the cumulative abnormal return is 8,19% for cash bids and 3,43% for share bids (means significantly different at 5%). The first three are statistically significantly different from zero at 1%, whereas the last is significant at a 5% level. The day before the announcement of the takeover and the day itself, cash bids also outperform the share bids, 12,17% versus 9,33% respectively, both statistically significant at 1%.
The strong takeover announcement returns are also observed in North American studies. For cash bids, Malmendier, Opp, and Saidi (2012, pp. 9) find that the cumulative abnormal return is 25% and for share bids this is 15%. Values of similar magnitude are also found by Huang and Walkling (1987, pp. 344). These values, however, are larger than the values found here. In the sample of Goergen and Renneboog, which consists of European takeovers, similar values are found as in this sample. As stated in section 2.5, the market for M&A is less competitive in Europe, amongst others, and therefore the returns are lower.
21 An important side note to these results is the finding of Goergen and Renneboog (2004, pp. 24), that small deals are paid with cash, and the finding of Malmendier, Opp, and Saidi (2012, pp. 11) that small targets are revalued higher due to more information becoming available on the market. Combined, this would imply that cash bids are used more for smaller deals, which in turn could explain the higher revaluation due to more new information. The result that smaller bids are predominantly done with cash bids can be seen in table 2. This explanation could be added to the already existing tax effect and selection bias.
5.2.3 UK versus the rest of Europe
Table 5
Cumulative abnormal returns of the target (UK versus the rest of Europe).
This table shows the cumulative abnormal return for six of the event windows used, distinguished by location (UK versus the rest of Europe).
Source: Calculations in Appendix 3 Event
window
UK Rest of Europe Difference
Mean t-value Mean t-value Mean t-value
(T-25, F+25) 0,2484 7,363 *** 0,1761 7,004 *** 0,0723 1,76099 * (T-25, T-2) 0,0765 3,778 *** 0,0604 3,973 *** 0,0160 0,64798 (T-1,T) 0,1189 8,928 *** 0,1087 10,649 *** 0,0101 0,61672 (T,F) 0,0642 3,038 *** 0,0302 1,731 * 0,0340 1,25001 (F, F+1) -0,0137 -1,983 ** -0,0086 -0,959 -0,0051 -0,42199 (F+2, F+25) -0,0012 -0,073 -0,0123 -1,440 0,0111 0,65919 Obs 122 173
*/**/*** denotes significance at a 10%/5%/1% level, resp.
Table 5 shows the difference between the UK and the rest of Europe. It is important to note that, even though targets in the UK have, on average, a higher cumulative abnormal return, inside Europe there are also differences between countries. As can be seen, the biggest difference is with the whole sample event window (F-25, F+25), with 24,84% for the UK and 17,61% for the rest of Europe, both significant at a 1% level. The difference between the two means, in this case, is also significantly different at 5%. The event windows around the takeover announcement show a greater CAR for UK targets, however, the difference is not significant. In the event window (T-25, T-2) this CARs are 7,64% versus 6,04%, and at (T-1, T) the CARs are 11,89% and 10,87%, for the UK and the rest of Europe, respectively. All
22 these findings are significant at a 1% level, however, the differences between the means are not. The inter announcement event window (T, F) shows that UK targets have a CAR of 6,42% and European a CAR of 3,02%, with a significance level of 1% and 5%, respectively. The event windows of post failure announcement do show negative CARs, but these results are of less (or no) statistical significance.
Even though it is evident from table 5 that target is the UK have a higher CAR in this sample, statistically there is only evidence this is the case over the whole event window.
When compared to findings in related literature about North American, the UK shows no more resemblance than the rest of Europe. Malmendier, Opp, and Saidi (2012, pp. 9) find that at the end of the event window, the CAR of cash bids is around 20% and share bids is 5%. These values resemble the results of the rest of Europe more than the results of the UK. However, when looking at other event windows, it becomes clear that the UK resembles the results of North America more. In this case, the resemblance depends on what event window is chosen for comparison. Huang and Walkling (1987, pp. 343) show that the takeover announcement (for successful and unsuccessful takeovers) generates an average abnormal return of 23,4%. This result resembles the results found for the UK. Important to note is that in their sample, Huang and Walkling (1987, pp. 343) used both successful and unsuccessful takeovers. However, Goergen and Renneboog (2004, pp. 30) show that targets of failed takeover bids have an even higher CAR (until the announcement of the takeover) than targets of successful bids.
As there is no strong evidence for the difference between the UK and Europe, the statement that the UK is different from the rest of Europe does not always hold. Moreover, as found by a literature review by Goergen and Renneboog (2004, pp. 11), North America and the UK differ substantially. Combining the evidence found in the sample and previous results from related literature, it can be said that Europe (including the UK) is different from North America in the magnitude of the revaluation of the target.
5.3 Regression analysis
Below are the results from the regressions done to test the two hypotheses. Table 6
Regression analysis.
23 that equals one if there was a cash bid, and equals zero if there was a share bid. Location UK is a dummy variable that equals 1 if the target is located in the UK, and equals zero if it is located anywhere else in Europe. The variable ln(Firm Value) is the relative size of the firm, measured in total equity. The variable ln(Deal value) is the relative proposed transaction value. The term ln(Firm value) x ln(Deal value) is the interaction term between the relative firm size and the relative transaction value. The variable for the period is T(2000-2003) is a dummy which equals whether the takeover attempt took place in the year 2000 – 2003 (1) or not (0). The same goes for T(2008-2011) and T(2012-2015). The omitted variable is T(2004-2007). The variable Difference from Average is the percentage deviation from the average of the sample of the time from the announcement until the announcement of failure. The variable Difference from Average x Cash is the interaction term between the variables Cash and Difference from Average. The variable Dometic indicates whether the takeover bid comes from a bidder that is of the same nationality as the target (1) or not (0).
Source: Sample data
Target CAR 1 2 3 4 Variables Cash 𝛽1,1 0,1094*** (0,04) 0,0929** (0,04) 0,0893** (0,04) Location UK 𝛽1,2 0,0723* (0,04) 0,0863** (0,04) ln(Firm value) 0,3310** (0,15) 0,3545** (0,15) ln(Deal value) 0,1397 (0,15) 0,1216 (0,15) ln(Firm value) × ln(Deal value)
-0,0118* (0,01) -0,0119* (0,01) T(2000 − 2003) 0,1461*** (0,05) 0,1595*** (0,05) T(2008 − 2011) 0,1778*** (0,05) 0,1781*** (0,05) T(2012 − 2015) 0,0665 (0,06) 0,0806 (0,06)
24 Difference from average
0,0014 (0,00)
0,0014 (0,00) Difference from average × cash
0,0031 (0,01) 0,0028 (0,01) Domestic -0,0415 (0,04) N 295 295 295 295 R2 0,0259 0,1002 0,0105 0,1158
*/**/*** denotes significance at a 10%/5%/1% level, resp.
Table 6 shows the regression coefficients of the regressions that are done to test the two hypotheses. The omitted variable is one of the four periods in the sample period. The first hypothesis is tested in column 1 and 2, column 3 and 4 test the second hypothesis.
The first hypothesis states that cash bids make for a higher CAR than share bids. Column 1 and 2 indicate this. Without any control variables, the target CAR is 10,94% higher with a cash bid, compared to a share bid, and is statistically significant at 1%. This result remains almost the same when various control variables are added. As can be seen, the target CAR is 9,29% with cash and significant at 5%. Therefore, in both cases, there is a strong positive relationship between the target CAR and cash bids. Interesting to see is that the relative firm size has a significant effect on the target CAR in this sample. For every 1% change in the asset value of the target, the CAR changes with 33,10%. This is contrary to existing literature based on the M&A market in North America and in Europe, where no significant statistical significance is found. Goergen and Renneboog (2004, pp. 36) did not find a statistically significant result of firm value on the target CAR for Europe. However, their sample consisted of successful and unsuccessful takeovers. Another interesting find is that in the periods 2000-2003 and 2008-2011, there were higher abnormal returns compared to 2004-2007, with 14,61% and 17,78%, respectively, in column 2. The same effect can be seen in column 4. In general, the regression yields strong evidence for a difference between cash and share bids, with the target CAR being higher with cash bids.
Combining the results from the event windows and the regression output, there is strong evidence that cash bids make for a higher CAR than share bids. In the event window for the period after the failure announcement, however, the evidence is less strong.
25 The second hypothesis states that if the target is located in the UK, the cumulative abnormal return for the target is higher than in other European countries. In column 3, where only the location of the target is regressed against CAR, there is weak evidence, significant at 10%, that indeed the cumulative abnormal return is higher in the UK compared to the rest of Europe (7,23%). When the control variables are added, the evidence becomes stronger, and the difference rises to 8,63%, with a statistical significance of 5%. Here, as in the event window analysis, there is no strong evidence that the location of the target in the UK has a positive relationship with the target CAR.
As there is no control for industry, the difference between the UK and the rest of Europe can be explained by that, amongst others. Another explanation can be the differences that were discussed in part 2.5, as also stated by Goergen and Renneboog (2004, pp. 38).
Combining the results from the event windows and the output of the regression, there is weak evidence that when the target is located in the UK, the CAR is higher, compared to the rest of Europe. Even though in the regression there is weak evidence for this result, the event windows do show that there is a difference between the UK and the rest of Europe, however only evidence is found for the difference over the whole period, no smaller event windows find evidence.
26
6 Discussion & Conclusion
This thesis shows that there certain effects that are observed in research based on the North American market for M&A. Strong evidence is found for the difference between cash and share bids for trying to explain the cumulative abnormal return differences in target firms. The effects are the same as in the North American market. The magnitude, however, is less. There are various reasons for the effect of cash and share bids on the target CAR. First, tax has to be paid on any capital gains realized by investors. This effect has been found in the North American market, and also holds for the European market. Capital gain tax can be deferred when a target is taken over with shares. This is not the case with cash. For this reason, investors want to be compensated. Second, the (financially constrained) bidder might not be able to increase a cash bid in case good news about the target becomes public. Therefore, the takeover attempt will fail. Share bids do not have this problem. The cash takeover attempts for targets with a high (re-)valuation lead to an over proportional failure of cash bids. Lastly, as cash bids are mostly done when the deal size is relatively low, and small targets have the higher revaluation than larger targets (more new information comes to the market when the bid comes out), cash bids are inherently linked to a higher revaluation. Also (weak) evidence is found for the effect of location on the target revaluation. Over the whole sample event window, there is evidence that there is a difference between the location of the target, either in the UK or the rest of Europe. Smaller event windows do not show this statistical significance. Even though the evidence for the difference in this sample is apparent, it is not very strong. Other, earlier, studies do find strong evidence for the difference between the UK and the rest of Europe (Goergen and Renneboog. 2004, pp. 37). There are various reasons for the difference between the UK and the rest of Europe. The most important reasons are a more decentralized financial decision-making process, better accounting standards, stronger shareholder protection, a more active and competitive market, and firms have more dispersed ownership. It is very common in the rest of Europe that there are only one or two majority shareholders that have effective control over firm management.
Combining the findings of the two hypotheses, it is clear that there is a difference in the market for M&A when comparing Europe and North America. Certain characteristics of the markets are similar, even though the magnitudes of certain effects differ between Europe and North America. Previous literature stated that the market in Europe is fundamentally different from the North American market. There results in this thesis, however, indicate that the
27 differences are not that pronounced, and that the relationships between the causes and effects do show similarities.
It is apparent that there are similarities and differences between the market for M&A in North America and Europe. And even within Europe there are large differences. There are various limitations to the results presented in this thesis. The targets that have been analyzed are all listed, which may have yielded different results compared to when all the targets would have been used in the sample, no listed and unlisted. The difficulty with this, however, is that it is difficult to find reliable and comparable information for unlisted targets. Share prices are a good means of comparison. Moreover, only pure cash and pure share bids have been used in the sample. To discover whether the results found in this thesis hold for every deal made, it is interesting to investigate all deals, including a hybrid of cash and shares, but also other forms of payment. Furthermore, in this thesis, the only geographical distinction made is the difference between the UK and the rest of Europe. As the UK is more similar to North America than to the rest of Europe, it would be interesting to discover whether the results within Europe also hold, when distinguishing between various regions in Europe, and leaving the UK out of the analysis. This way, a better picture can be formed of how the regions in Europe perform.
28
References
Block, S.B. (1968). The Effect of Mergers and Aqcuisitions on the Market Value of Common Stock. Journal of Finance. 23 (5), pp.889-890
Davidson, W.N., Dutia, D., Cheng, L. (1989). A Re-Examination of the Market Reaction to Failed Mergers. Journal of Finance. 44 (4), pp.1077-1089.
Dealogiccom. (2015). Market Insight M&A Statshot. Retrieved 01 July, 2015, from http://www.dealogic.com/media/market-insights/ma-statshot/
Dodd, P. (1980), Merger proposals, management discretion and stockholder wealth, Journal
of Financial Economics 8, 105-137.
Europa.eu. (2015). Council Directive 2009/133/EC. Retrieved 07 July, 2015, from http://eur-lex.europa.eu/LexUriServ/LexUriServ.do?uri=OJ:L:2009:310:0034:0046:EN:PDF Goergen, M & Renneboog, L. (2004). Shareholder Wealth Effects of European Domestic and
Cross-border Takeover Bids. European Financial Management, 10(1), 9-45.
Huang, Y.S., Walkling, R.A., 1987. Target abnormal returns associated with acquisition announcements: Payment, acquisition form, and managerial resistance. Journal of
Financial Economics. 19, pp. 329-349.
Jensen, M.C & Ruback, R.S. (1983). The Market for Corporate Control. The Journal of
Financial Economics, 11(1), 5-50.
Malmendier, U., Opp, M., Saidi, F. (2012). Target Revaluation after Failed Takeover
Attempts – Cash versus Stock (Working Paper No. 18211). Retrieved from National
Bureau of Economic Research website: http://www.nber.org/papers/w18211
Moschier, C., Campa, J.M. (2009). The European M&A Industry: A Market in the Process of Construction. Academy of Management Perspectives. 23 (4), pp. 71-87
Shick, R.A. . (1971). The Analysis of Mergers and Acquisitions. The Journal of Finance, 27(2), 495-502.
Travlos, N.G. (1987). Corporate Takeover Bids, Methods of Payment, and Bidding Firms' Stock Returns. The Journal of Finance, 42(4), 943-963.
29 Www.gov.uk. (2015). Enterprise Investment Scheme and Capital Gains Tax. Retrieved 07
July, 2015, from https://www.gov.uk/government/publications/enterprise-investment-scheme-and-capital-gains-tax-hs297-self-assessment-helpsheet
30
Appendix 1
Calculation table 3
Table 7
Overview values for calculations.
This table shows an overview of the values that are used in table 3. The p-value indicates whether the mean is statistically significant from zero. The test for significance is a two-sided t-test.
Event window 𝑪𝑨𝑹̅̅̅̅̅̅ 𝒔𝟐 Df t-value p-value
Overall (T-25, F+25) 0,206024 0,12158 294 10,13126 0,00 (T-25, T-2) 0,067063 0,04374 294 5,49806 0,00 (T-1,T) 0,112909 0,01934 294 13,92190 0,00 (T,F) 0,044245 0,05309 294 3,29249 0,00 (F, F+1) -0,010710 0,01047 294 -1,79439 0,07 (F+2, F+25) -0,007704 0,02038 294 -0,92538 0,36 N 295
Formula for variance and t-value
For the overall sample, the variance is calculated by the following formula:
𝑠2 = ∑(𝑥𝑖 − 𝑥̅) 2 𝑛 − 1 𝑛
𝑖=1 And the t-value by:
𝑡 = 𝐶𝐴𝑅̅̅̅̅̅̅ − 0𝑠 𝑛 − 1 Where 𝑠 = √𝑠2.
31
Appendix 2
Calculation table 4
Table 8
Overview values for calculations.
This table shows an overview of the values that are used in table 4. The p-value indicates whether the mean is statistically significant from zero. The test for significance is a two-sided t-test. Panel A shows the values for cash bids. Panel B shows the values for share bids. Panel C shows the values for the difference between the CAR of the cash and share bids.
Event window 𝑪𝑨𝑹̅̅̅̅̅̅ 𝒔𝟐 Df t-value p-value
Panel A: Values for cash bids
Cash bids (T-25, F+25) 0,243734 0,12156 202 9,93583 0,00 (T-25, T-2) 0,081933 0,05345 202 5,03664 0,00 (T-1,T) 0,121786 0,02176 202 11,73389 0,00 (T,F) 0,064518 0,05043 202 4,08332 0,00 (F, F+1) -0,014185 0,01433 202 -1,68407 0,09 (F+2, F+25) -0,011311 0,01579 202 -1,27945 0,20 N 203
Panel B: Values for share bids
Share bids (T-25, F+25) 0,122818 0,11279 91 3,48849 0,00 (T-25, T-2) 0,034254 0,02108 91 2,25050 0,03 (T-1,T) 0,093320 0,01361 91 7,63105 0,00 (T,F) -0,000489 0,05664 91 -0,01962 0,98 (F, F+1) -0,003044 0,00194 91 -0,65901 0,51 (F+2, F+25) 0,000255 0,03070 91 0,01386 0,99 N 92
Panel C: Values for the difference between cash and share bids
Difference (T-25, F+25) 0,120917 0,01412 293 2,79091 0,01 (T-25, T-2) 0,047679 0,00188 293 1,82102 0,07 (T-1,T) 0,028467 0,00037 293 1,63341 0,10 (T,F) 0,065008 0,00274 293 2,26047 0,02 (F, F+1) -0,011141 0,00011 293 -0,86580 0,39 (F+2, F+25) -0,011566 0,00042 293 -0,64400 0,52
32 Formula for variance and t-value
For the cash bids and share bids, the variance is calculated by the following formula:
𝑠2 = ∑(𝑥𝑖 − 𝑥̅) 2 𝑛 − 1 𝑛
𝑖=1 And the t-value by:
𝑡 = 𝐶𝐴𝑅̅̅̅̅̅̅ − 0𝑠 𝑛 − 1 Where 𝑠 = √𝑠2.
For the difference between cash and share bids, the pooled variance is used:
𝑠𝑝2 = 𝑠𝐶𝑎𝑠ℎ 2 (𝑛
𝐶𝑎𝑠ℎ− 1) + 𝑠𝑆ℎ𝑎𝑟𝑒𝑠2 (𝑛𝑆ℎ𝑎𝑟𝑒𝑠− 1) (𝑛𝐶𝑎𝑠ℎ− 1) + (𝑛𝑆ℎ𝑎𝑟𝑒𝑠− 1) And the t-value is calculated by:
𝑡 =𝐶𝐴𝑅̅̅̅̅̅̅𝑐𝑎𝑠ℎ − 𝐶𝐴𝑅̅̅̅̅̅̅𝑠ℎ𝑎𝑟𝑒𝑠− 0 𝑠𝑝2√ 1
𝑛𝑐𝑎𝑠ℎ + 1 𝑛𝑠ℎ𝑎𝑟𝑒𝑠
33
Appendix 3
Calculation table 5
Table 9
Overview values for calculations.
This table shows an overview of the values that are used in table 5. The p-value indicates whether the mean is statistically significant from zero. The test for significance is a two-sided t-test. Panel A shows the values for location UK. Panel B shows the values for location rest of Europe. Panel C shows the values for the difference between the CAR of targets from the UK and the rest of Europe.
Event window 𝑪𝑨𝑹̅̅̅̅̅̅ 𝒔𝟐 Df t-value p-value
Panel A: Values for the UK
UK (T-25, F+25) 0,248445 0,13776 121 7,36323 0,00 (T-25, T-2) 0,076469 0,04957 121 3,77795 0,00 (T-1,T) 0,118861 0,02145 121 8,92828 0,00 (T,F) 0,064195 0,05402 121 3,03810 0,00 (F, F+1) -0,013709 0,00578 121 -1,98292 0,05 (F+2, F+25) -0,001174 0,03158 121 -0,07266 0,94 N 122
Panel B: Values for the rest of Europe
Rest of Europe (T-25, F+25) 0,176110 0,10873 172 7,00445 0,00 (T-25, T-2) 0,060431 0,03979 172 3,97326 0,00 (T-1,T) 0,108711 0,01792 172 10,64907 0,00 (T,F) 0,030176 0,05226 172 1,73112 0,09 (F, F+1) -0,008596 0,01382 172 -0,95881 0,34 (F+2, F+25) -0,012310 0,01256 172 -1,44028 0,15 N 173
Panel C: Values for share the difference between UK and the rest of Europe
Difference (T-25, F+25) -0,072335 0,01457 293 -1,76099 0,08 (T-25, T-2) -0,016038 0,00192 293 -0,64798 0,52 (T-1,T) -0,010150 0,00038 293 -0,61672 0,54 (T,F) -0,034019 0,00281 293 -1,25001 0,21 (F, F+1) 0,005113 0,00011 293 0,42199 0,67 (F+2, F+25) -0,011136 0,00042 293 -0,65919 0,51
34 Formula for variance and t-value
For the UK and the rest of Europe, the variance is calculated by the following formula:
𝑠2 = ∑(𝑥𝑖 − 𝑥̅) 2 𝑛 − 1 𝑛
𝑖=1 And the t-value by:
𝑡 = 𝐶𝐴𝑅̅̅̅̅̅̅ − 0𝑠 𝑛 − 1 Where 𝑠 = √𝑠2.
For the difference between UK and the rest of Europe (RoE), the pooled variance is used:
𝑠𝑝2 =
𝑠𝑈𝐾2 (𝑛𝑈𝐾 − 1) + 𝑠𝑅𝑜𝐸2 (𝑛𝑅𝑜𝐸− 1) (𝑛𝑈𝐾− 1) + (𝑛𝑅𝑜𝐸− 1) And the t-value is calculated by:
𝑡 =𝐶𝐴𝑅̅̅̅̅̅̅𝑈𝐾 − 𝐶𝐴𝑅̅̅̅̅̅̅𝑅𝑜𝐸− 0 𝑠𝑝2√ 1
𝑛𝑈𝐾 + 1 𝑛𝑅𝑜𝐸
