Cross-Border versus Domestic Acquisitions and the Impact on Shareholder Wealth

(i) Theoretical Arguments for Acquisition Cross-Border Effects

If international capital and takeover markets are perfectly integrated, one could expect there to be no systematic differences in the abnormal returns to either targets or bidders in cross-border as compared to domestic acquisitions (Harris and Ravenscraft, 1991). However, an assumption of perfectly integrated markets is arguably unrealistic, and there are both theoretical arguments and prior empirical evidence to suggest that the level of abnormal returns may differ systematically between cross-border and domestic acquisitions. The literature is, however, conflicting, with different arguments put forward as to whether cross-border acquisitions can be expected to create or destroy value, and whether the wealth effects of cross-border acquisitions will be greater or smaller than in domestic acquisitions.

Cross-border acquisitions can be expected to be more complex, and thus more costly and risky to execute, than domestic acquisitions. The potential for valuation error may be a more serious problem in cross-border than domestic acquisitions (Conn et al., 2005) if targets in foreign markets are more difficult to value than domestic targets (due to e.g., less developed capital markets, differences in accounting practices, volatile exchange rates, or less knowledge of foreign markets). If synergies are forecast with some degree of error, or if managers suffer from hubris and systematically over-estimate their ability to improve the performance of the target firm (Roll, 1986) and the successful bidder is the one with the highest target over-valuation, bidding company shareholders can be expected to lose. If there is greater scope for valuation error in cross-border than in domestic acquisitions, we may expect higher target abnormal returns, but lower bidder returns, in cross-border than in domestic acquisitions. Aybar and Ficici (2009) argue that the problems of cross-border acquisitions, such as limited market knowledge, may be exacerbated if the bidding company has no prior operations in the target country, while Conn and Connell (1990) argue that companies from more competitive takeover markets, such as the US, can be expected to have more acquisition experience and make better acquisitions.

Not only may managers suffer from hubris and over-estimate potential synergies; with the separation of ownership and control, acquisitions may be driven by managerial and not only shareholder wealth maximisation objectives (Jensen and Meckling, 1976). The scope for agency conflict may be particularly severe in acquisitions, as managers may benefit from such transactions even where they deliver no value to shareholders.⁵ If the scope for valuation error or the agency conflict between managers and shareholders is larger in cross-border than in domestic acquisitions, one can expect bidders to perform worse, but targets to gain more, in cross-border than in domestic acquisitions.

With the high cost and risk of cross-border acquisitions, why do managers increasingly pursue such transactions, and why do shareholders allow them? Unless there are significant additional benefits from cross-border as compared to domestic acquisitions, one could expect the added cost and complexity to lead to lower abnormal returns to bidders, as well as the overall wealth creation to be lower in cross-border than in domestic acquisitions.

Increasing the investment universe to also consider cross-border acquisitions may increase the scope for identifying undervalued targets. If corporate governance rules in a country are weak, foreign bidders may add value through improvement in target company management (Bris and Cabolis, 2008). Volatile exchange rates may also provide foreign bidders with a cost of capital advantage over local firms if managers are able to time acquisitions to coincide with a strong home currency compared to that of the target country (Froot and Stein, 1991; and Harris and Ravenscraft, 1991).

The internationalisation literature suggests that multinational companies (MNCs) may have the advantage of being able to exploit their intangible assets in a number of markets (Harris and Ravenscraft, 1991). If international diversification or access to new markets (Doukas and Travlos, 1988) is valuable, we may expect bidders to perform better in cross-border than in domestic acquisitions. However, with targets tending to extract most, if not all, of any merger benefit, any gain from market access or international diversification can also be expected to be observed in higher target abnormal returns.

While the discussion above relates to domestic vs. cross-border acquisitions in general, the abnormal returns can be expected to vary between cross-border acquisitions into and out of the UK, as well as with the nationality of the overseas targets or bidders. Conn and Connell (1990) argue that target abnormal returns are likely to be especially high in the US, given its highly competitive takeover market, while ‘ … returns to foreign bidders should be relatively high if the market for corporate control is relatively inefficient in [the target country] … ’ (p. 691). If the US takeover market is the most competitive, we may expect US cross-border targets to gain the most, while US cross-border bidders may also gain from acquiring into the UK if the UK takeover market is less competitive than that in the US.

If countries within the European Economic Area (EEA)⁶ are well integrated, acquisitions within the free trade area may be perceived as having lower risk, but also potentially lower diversification benefits, than other cross-border acquisitions. Indeed, if the EEA is fully integrated, one could expect the abnormal returns to be little different in intra-EEA and domestic acquisitions. Intra-EEA acquisitions may therefore be associated with lower target, and possibly also lower bidder, cross-border effects. We split our sample into UK, US, (non-UK) EEA and the Rest of the World in the empirical analysis to test these predictions.

We further explore whether differences in countries’ corporate governance systems may explain any observed variation in cross-border effects with the nationalities of the companies involved. Bidders from countries with strong investor protection can be expected to be less inclined to undertake value-destructive acquisitions than other bidders (Kuipers et al., 2009). However, agency conflict may also afflict target company management, and the weaker the corporate governance system, the greater the scope for companies to be poorly managed. In cross-border acquisitions, at least where there is a complete transfer of ownership, ‘ … the target firm becomes a national of the country of the acquiror, and consequently subject to its corporate governance system’ (Bris and Cabolis, 2008, p. 605). Thus, the greater the differences in the legal traditions and level of investor protection of the bidding and target countries, the greater the potential improvements in the governance systems of the target company after a cross-border acquisition. We expect both bidders and targets to gain more from cross-border acquisitions where the difference in the corporate governance systems is large. We explore the impact of country governance systems on cross-border effects in the empirical analysis. First, however, we review some of the prior empirical evidence on target and bidding cross-border effects and their determinants.

(ii) Prior Evidence on Target and Bidder Cross-Border Effects

Studies of domestic acquisitions⁷ tend to find that target shareholders on average earn significant abnormal returns around the time of the bid announcement, often in the region of 20% to 30%, and prior literature on cross-border acquisitions⁸ tends to find that targets gain more in cross-border than in domestic acquisitions. However, not only is the evidence for non-US targets still limited, but more importantly, the cause of the target company cross-border effect still eludes us. While some studies attribute the cross-border effect to differences in bid and company characteristics between cross-border and domestic acquisitions, others find significant cross-border effects even when controlling for bid characteristics. For example, while Wansley et al. (1983), Dewenter (1995), Campa and Hernando (2004) and Danbolt (2004) find no significant residual target cross-border effect once differences in bid and company characteristics are controlled for, and Bris and Cabolis (2008) surprisingly find the mean premium to be significantly lower for cross-border than for comparable domestic targets, Harris and Ravenscraft (1991) find the residual cross-border effect to exceed 10 percentage points even when controlling for the method of payment and other bid characteristics. There is also some evidence to suggest that the size of the target company cross-border effect varies internationally, and that it is relatively small in the UK (Danbolt, 2004) compared to that of US targets (Conn and Connell, 1990).

The evidence on bidder returns of cross-border acquisitions is even more mixed. While e.g., Doukas and Travlos (1988) and Francis et al. (2008), amongst others, find US bidders to gain in at least some cross-border acquisitions, other studies, such as Moeller and Schlingemann (2005), find US bidders to lose from cross-border acquisitions. While Kuipers et al. (2009) find significant losses to bidders from cross-border acquisitions into the US, Kang (1993) finds foreign bidders to gain significantly from acquisitions of US targets. Chari et al. (2006) find that companies from developed markets gain significantly from acquisitions into emerging markets. The evidence for UK bidders is, however, much more limited. Gregory and McCorriston (2005) and Uddin and Boateng (2009) find insignificant bid-announcement abnormal returns to UK bidders in cross-border acquisitions, while Danbolt (1995) finds overseas bidders into the UK gain insignificantly during the month of the bid announcement.⁹ Conn and Connell (1990) find that both UK and US bidders suffer negative abnormal returns following cross-border acquisitions between the two markets, with UK bidders on average performing substantially worse than US cross-border acquirers.

However, none of the above studies compares the performance of bidders in cross-border acquisitions to that of domestic acquirers, and therefore they do not address the issue of whether there are systematic differences in bidder abnormal returns in domestic and cross-border acquisitions. There are few such comparative studies. Eckbo and Thorburn (2000) find US bidders to gain less than Canadian bidders in acquisitions of Canadian firms, and Starks and Wei (2004) find foreign bidders to gain less than US bidders in acquisitions of US firms. Moeller and Schlingemann (2005) and Black et al. (2007) find lower announcement returns to US bidders in cross-border than in domestic acquisitions. Francis et al. (2008) overall also find that US bidders in general perform better in domestic than in cross-border acquisitions, although the cross-border effects for US bidders turn insignificantly positive during the late 1990s and early 2000s. In a study of European acquisitions, Campa and Hernando (2004) find bidders to perform better in domestic than in cross-border acquisitions, although the difference is only significant for a long pre-announcement window. In a large international study, Ahern et al. (2011) find no significant difference in bidder returns in cross-border and domestic acquisitions. Studying long-run abnormal returns for a small sample of UK acquiring firms in large acquisitions,¹⁰ Aw and Chatterjee (2004) find UK bidders to perform worse following cross-border than domestic acquisitions. Conn et al. (2005, p. 815), also studying UK bidders, find ‘cross-border acquisitions result in lower announcement and long-run returns than domestic acquisitions’. However, their sample is dominated by acquisitions of private firms, and in acquisitions of public targets the bidder cross-border effect is reversed.¹¹

Market access is commonly argued to be a major motive for cross-border acquis-itions, but prior evidence is mixed. While Doukas and Travlos (1988) find higher abnormal returns to US bidders when acquiring into new markets, Danbolt (2004) finds no evidence of market access having a significant impact on target abnormal returns in cross-border acquisitions into the UK. We extend prior literature by studying the impact of market access on both target and bidder cross-border effects. We also extend prior literature by exploring whether prior bidder cross-border acquisition experience has an impact on target or bidder returns.

Prior research provides mixed evidence regarding the impact of exchange rate changes on the abnormal returns in cross-border acquisitions. While Harris and Ravencraft (1991) and Kang (1993) find target abnormal returns in cross-border acquisitions to be higher when the bidding country's currency is strong relative to the currency of the target country, Dewenter (1995), Danbolt (2004) and Starks and Wei (2004) find no support for the exchange rate hypothesis on target returns. Gregory and McCorriston (2005) find exchange rates to have a statistically significant impact on bidder returns, while Conn et al. (2005) find no significant impact of exchange rate changes on bidder returns.

Bris and Cabolis (2008) find differences in accounting quality between the bidder and target countries to have a significant impact on the target company bid premium, with target gains higher where the bidder comes from a nation with better accounting quality than the target. Prior literature suggests national differences in country governance characteristics may also have a significant impact on the level of cross-border effects. Black et al. (2007) and Francis et al. (2008), studying US bidders, Starks and Wei (2004) and Kuipers et al. (2009), studying foreign bidders acquiring into the US, and Martynova and Renneboog (2008), studying European bidders, all find higher bidder returns where the bidder comes from a country with better corporate governance standards than those of the target country, although the impact on bidder returns is not significant in the study by Starks and Wei. While their focus is mainly on the impact of international variations of laws and regulations on the volume of mergers and acquisitions activity, Rossi and Volpin (2004) also find the level of the bid premium to be higher in countries with strong shareholder protection, although their results seem to be driven by returns for US and UK targets. Bris and Cabolis (2008) similarly argue that differences between bidder and target countries in terms of investor protection may have a significant impact on the bid premium. Studying cross-border acquisitions in 39 countries, Bris and Cabolis (2008) find the cross-border effect on the target bid premium to increase where the cross-border bidders come from countries with better shareholder protection than those in the target country. Studying European acquisitions, Martynova and Renneboog (2008) similarly find higher target returns where their country's corporate governance standards are lower than those of the bidder. However, studying cross-border acquisitions by US firms, Black et al. (2007) find the foreign targets to gain less when they are based in countries with low accounting quality, and Starks and Wei (2004) find US targets to gain less when the foreign acquirer comes from countries with strong corporate governance. The prior evidence is thus mixed.

The evidence on the impact of accounting quality on target returns is mixed, and does not incorporate the possible impact of accounting harmonisation following the widespread adoption of IFRS. There is also a general lack of evidence on whether differences in the level of shareholder protection in the bidder and target countries also affect the level of bidder returns. Following Bris and Cabolis (2008), we analyse the impact of differences in accounting quality, legal origin, the levels of anti-director rights, the quality of the rule of law, and the level of shareholder protection between the bidder and target countries on the level of abnormal returns, while also controlling for the effect of market access and bidders’ prior acquisition experience. While the prior literature has mostly focused on target shareholders, we analyse the impact of country characteristics on both target and bidder cross-border effects.

We next turn to a discussion of our research design, including a discussion of our data, sample and our model for calculating abnormal returns.

(i) Data Sources and Sample

We obtain information on announcement dates and bid characteristics from the Thomson Financial SDC Mergers Database obtained through Thomson ONE Banker.¹² We focus on changes in control, and therefore limit the sample to successful acquisitions where the bidder held less than 50% of the target prior to the bid announcement, and where the bidder holds more than 50% of the shares in the target after the acquisition. The acquisitions were announced between 1 January, 1980 and 31 December, 2008, although very few transactions are included in the SDC database prior to 1986. Share and index returns, as well as market value data, are obtained from Thomson Datastream.

A sample of 535 cross-border acquisitions – 305 acquisitions into the UK and 230 cross-border acquisitions out of the UK – match our initial sample criteria. However, returns data is missing for 114 cross-border targets and 154 cross-border bidders, as detailed in Table 1. Following Bris and Cabolis (2008), we match the target and bidding companies in cross-border acquisitions with targets and bidders in comparable domestic acquisitions. Acquisitions are matched based on country, year of acquisition, industry, and size.¹³ Our final sample consists of 251 cross-border targets matched to 251 target companies in domestic acquisitions, and 146 cross-border bidders, again matched to bidders in comparable domestic acquisitions. In the final sample of cross-border acquisitions, we have data for 174 UK targets and 77 overseas targets acquired by UK companies, and for 81 UK and 65 overseas cross-border bidders. While the US dominates as the target for UK cross-border acquisition and is the most frequent overseas acquirer of UK companies, our sample also includes companies from a number of other countries. The main countries represented are specified in Table 1. In the analysis we group the countries into the UK, other European Economic Area countries, the US, and the Rest of the World.

Table 2 provides further information on the sample and the matching between cross-border and domestic acquisitions. The majority of acquisitions in the sample took place during the merger boom of the late 1990s, and the sample is small in certain years. In order to control for possible time-variation in returns, we match our sample of cross-border acquisitions to comparable companies in domestic acquisitions in the same year. We also use time dummies in the cross-sectional regressions.¹⁴ Our third matching criterion is industry. The largest proportion of cross-border acquisitions involves companies in the manufacturing sector, although service sector firms are also well represented in the sample. Finally, we match cross-border and domestic acquisitions based on total assets, restricting the value of the domestic company to be between 50% and 200% of the size of the cross-border company.

(ii) Estimation of Abnormal Returns

We measure shareholder wealth effects around the period of the bid announcement using standard event study methodology, with abnormal returns estimated using the conventional market model (Brown and Warner, 1985), as specified in equation (1):

(1)

Log returns are calculated from Total Returns Index (TRI) data obtained from Datastream, and R_it, capturing the log return on share i on day t, is estimated as in equation (2):

(2)

R_mt captures the log return on the market portfolio on day t, calculated from the return on the various home country stock market indices. β_i captures the systematic risk of the share, while α_i is the intercept. We estimate the market model parameters over a period of 220 days, from day t–260 to t–41, where t= 0 refers to the date of the bid announcement.¹⁵

While the market model (MM) is the most widely used event study model, we acknowledge that this model has been subject to critique, including possible problems of missing variable bias and other measurement problems. Benchmarking problems can be significant, although focusing on the short-run announcement effect, the choice of benchmark has only a small impact on the estimated abnormal returns (Kothari and Warner, 2007).¹⁶ Still, as a robustness check, we also undertake the analysis using the market adjusted returns model (MAR), ¹⁷ with α= 0, and β= 1, as specified in equation (3):

(3)

The choice of model has limited impact and the results are consistent. For brevity, we only report market model results in the tables.¹⁸

Following prior literature, we focus on a short, 3-day event window, from day t–1 to day t+1. However, such a short event window, while commonly used in prior literature, may not capture the full impact of acquisitions if there is bid speculation or information leakage prior to the formal bid announcement, or if new price-sensitive information is released after the initial bid announcement.¹⁹ We therefore also analyse cumulative abnormal returns (CAR) over a slightly longer 11-day event period, from t–5 to t+5.²⁰

The levels of statistical significance of the cumulative abnormal returns are calculated using a cross-sectional t–test of the mean (Strong, 1992). However, to allow for non-normality in the distribution of abnormal returns, we also test the significance using the nonparametric Wilcoxon rank sum test of the median, as well as a simple sign test based on the proportion of sample companies with non-negative abnormal returns (Siegel and Castellan, 1988). The significance of the regression coefficients are calculated using heteroscedasticity-consistent standard errors (White, 1980), controlling for clustering (Petersen, 2009).²¹ As a further robustness test, we also estimate the regressions using robust rank regressions, and the results are consistent.

(i) Target Abnormal Returns

Target abnormal returns are reported in Panel A of Table 3. Target company shareholders on average earn significant positive abnormal returns around the time of the bid announcement, with mean cumulative abnormal returns to targets in cross-border acquisitions amounting to a highly significant 20.9% over the 3-day period from day t–1 to t+1. More than 93% of sample firms earn positive cumulative abnormal returns. Target companies in comparable domestic acquisitions on average also earn positive abnormal returns, but the gains are significantly smaller, averaging 10.9%. The difference in target returns between cross-border and matched domestic acquisitions – the target company cross-border effect – amounts to a highly significant 10.1 percentage points, with almost two thirds of the cross-border targets earning higher 3-day CAR than shareholders in comparable domestic acquisitions.

A 3-day event window may arguably be too short to capture the full impact of acquisitions, and we also analyse abnormal returns over an 11-day event window, from five days prior to five days after the day of the bid announcement. Over this extended event window, target abnormal returns in cross-border acquisitions are even higher, at 26.4%, and the cross-border effect rises to 13.1 percentage points. Indeed, as can be seen from Figure 2, Panel A, target company share prices on average start rising long before the bid announcement, with cumulative abnormal returns over the time period from t–40 to t–6 averaging 9.4% for cross-border and 9.7% for domestic targets.²² Over an extended 81-day period centred on the bid announcement, targets on average earn 40.1% cumulative abnormal returns in cross-border acquisitions, compared to 26.8% in domestic acquisitions, leading to a significant cross-border effect of 13.3%– similar to that captured by the 11-day event window.²³ Thus, regardless of what event window we analyse, we find evidence of significant cross-border effects in target company abnormal returns.

As can be seen from Panel C of Table 3, the level of target abnormal returns in cross-border acquisitions has increased substantially over time, from an average of 14.1% during the 1980s to 26.8% during the 2005–2008 period. The level of target returns in domestic acquisitions has not shown a similar increase, and the target company cross-border effect has been rising substantially over time, from 7.9 percentage points during the 1980s (and an insignificant negative cross-border effect for the small number of transactions during the early 1990s), to 13.7 and 14.3 percentage points, respectively, for the first and second half of the 2000s.

However, further analysis reveals that the level of the target company cross-border effect also varies significantly with the nationality of the target and bidding companies. As displayed in Panel D of Table 3, while the average 3-day target company cross-border effect for the sample as a whole amounts to 10.1 percentage points, the cross-border effect for UK targets of 4.6 percentage points is substantially smaller than the average cross-border effect of 22.5 percentage points to overseas targets acquired by UK firms. The cross-border effect is particularly high for US and non-UK EEA targets, at 24.2 and 19.1 percentage points, respectively, although the target company cross-border effect is also high for other overseas targets, averaging 9.9 percentage points. The low cross-border effect to UK targets is particularly pronounced in acquisitions by bidders from other EEA countries, where the target cross-border effect is an insignificant 0.7 percentage points. Indeed, it is only in acquisitions by US bidders that we observe a significant cross-border effect for UK targets, at 8.9 percentage points. The target company cross-border effect is thus consistently lower in acquisitions into the UK than in cross-border acquisitions out of the UK. We analyse in Section 5 the extent to which the differences in abnormal returns in domestic and cross-border acquisitions are attributable to differences in company and bid characteristics, while in Section 6 we explore why the target company cross-border effect varies with target and bidder nationality. The high abnormal returns to targets, particularly in cross-border acquisitions, raise questions as to whether such transactions are valuable investments for the acquiring companies, or whether bidders tend to overpay. We explore this next.

(ii) Bidder Abnormal Returns

As reported in Panel B of Table 3, bidders in cross-border acquisitions on average earn abnormal returns insignificantly different from zero, amounting to −0.3% over both the 3-day and 11-day event windows. However, bidders in comparable domestic acquisitions on average earn significant negative abnormal returns, of −1.8% and −1.3%, respectively, over the two event windows. Thus, while target shareholders gain significantly more from cross-border than from domestic acquisitions, bidders on average also perform significantly better – or more accurately, significantly less badly – in cross-border than in domestic acquisitions. The bidding company cross-border effect amounts to a significant 1.5 percentage points over the 3-day event window, although the cross-border effect, of 1.0 percentage point, is no longer significant for the longer 11-day event window. The movements in bidder abnormal returns over the full period from t–40 to t+40 days are depicted in Figure 2, Panel B. While the abnormal returns to bidders in cross-border acquisitions fall marginally over the post-acquisition period, the bidding company cross-border effect remains positive.

The bidder cross-border effect is further analysed in Panels C and D of Table 3, which splits the results based on time period and the nationality of the targets and bidders, respectively. Average bidder returns in cross-border acquisitions are insignificantly different from zero in all periods, while domestic bidders earn significant negative abnormal returns in three of the five time periods. The bidder cross-border effect is particularly large during the early 2000s, when it reaches a significant 3.8 percentage points. The bidding company cross-border effect also varies significantly with the nationality of the firms involved. The cross-border effect is higher for UK cross-border bidders, at a significant 1.9 percentage points, and is particularly high, at 3.8 percentage points, when UK companies acquire into other EEA countries. On the other hand, for US bidders the cross-border effect is only 0.8 percentage points and not statistically significant, and for UK bidders acquiring in the US the cross-border effect is 0.7 percentage points and again not significant. The small number of bidders from EEA and other countries also earn higher abnormal returns in cross-border than in domestic acquisitions, although the bidding company cross-border effects are not significant for the EEA and the Rest of the World regions.

The analysis above suggests that both targets and bidders on average perform better in cross-border than in domestic acquisitions. While the target company cross-border effect is particularly high in cross-border acquisitions by UK companies, UK bidders perform significantly better in cross-border than in domestic acquisitions. However, prior research suggests bid characteristics may have a significant impact on target and bidder returns, and if there are systematic differences in the characteristics of targets and bidders in cross-border and domestic acquisitions, the cross-border effects may be attributable to such differences rather than to the different nationalities of the targets or bidders. In the next section we therefore analyse the differences in the characteristics of companies involved in cross-border and domestic acquisitions, and the impact of such differences on the cross-border effects.

(i) Bid Characteristics and the Impact on Abnormal Returns

Sample characteristics are reported in Panel A of Table 4, while the Pearson correlation coefficients between the 3-day cumulative abnormal returns and sample characteristics are reported in Panel B. Data sources and detailed variable definitions for the cross-sectional variables are provided in the Appendix.

Prior research generally finds both target and bidding shareholders to gain significantly more in cash than in equity offers,²⁴ and we follow prior literature in controlling for the method of payment. We introduce dummy variables for cash-only and equity-only payment, with mixed payment offers as the residual category. As can be seen from Panel A, significant differences between domestic and cross-border acquisitions are observed in the method of payment. While cash-only is offered in more than 61% of cross-border acquisitions, such payment is used in approximately 52% of domestic acquisitions based on the sample of targets, and 34% based on the bidder sample. The difference between cross-border and domestic acquisitions in the prevalence of cash payment is substantial and statistically significant. Full equity offers are relatively rare, accounting for approximately 10% of domestic and less than 9% of cross-border acquisitions.

In Panel B we explore the relationship between the method of payment and cumulative abnormal returns, and our results are generally consistent with the prior literature. We find target abnormal returns to be significantly negatively correlated with equity payment, although the method of payment is not found to have a significant impact on target returns in domestic acquisitions. Domestic bidders also perform significantly better in cash-financed acquisitions than in transactions with other forms of payment. The higher proportion of cross-border than domestic acquisitions with cash offers and the higher returns to targets in cash than in equity offers may contribute to the target company cross-border effect. We explore this further in the cross-sectional regression analysis below.

We next control for the effects of company size and the relative size of the targets and bidders.²⁵ We measure company size by the market value of the company 41 days prior to the date of the bid announcement. Due to the non-normality of company size, we use a log transformation of market values in the cross-sectional analysis. We measure relative size by the ratio of total assets of the target to the total assets of the bidder.²⁶ While we match cross-border and domestic acquisitions based on total assets, the market values are on average higher for cross-border targets and bidders than their matched domestic counterparts. However, Panel B suggests target abnormal returns are significantly negatively related to firm size, which would seem to work against the observed target company cross-border effect. We find no significant differences in the relative size of targets and bidders in cross-border and domestic acquisitions, nor does relative size seem to have a significant correlation with the cross-border effects in the univariate analysis.

Prior research has suggested the method of acquisition matters, and we control for whether the acquisitions are undertaken through a tender offer or a merger,²⁷ with a dummy variable taking the value 1 in tender offers. As reported in Panel B, the results suggest cross-border targets and domestic bidders perform better in tender offers than in mergers, although the correlation coefficients are insignificant and negative for domestic targets and cross-border bidders. However, as can be seen from Panel A, there are only marginal differences in the proportion of cross-border and domestic acquisitions that are undertaken through tender offers. The variable is thus unlikely to account for the observed target or bidder cross-border effects.

We find target abnormal returns to be significantly negatively related to the size of the stake held by the bidder in the target prior to the acquisition. Bidders on average hold significantly higher pre-bid stakes in the target in domestic than in cross-border acquisitions, and this may contribute to the observed target company cross-border effect. We control for the size of any stake by the bidder in the target prior to the acquisition²⁸ in the cross-sectional analysis.

Finally, we control for industrial diversification, with a dummy variable taking the value 1 where the targets and bidders are operating in different primary industries.²⁹ A higher proportion of domestic than cross-border acquisitions involves industrial diversification. However, the correlation coefficients in Panel B of Table 4 suggest that whether the acquisition is focused or diversifying appears to have limited impact on the level of either target or bidding company abnormal returns.

We next explore the relationship between bid characteristics and abnormal returns in a multivariate setting.

(ii) Cross-Sectional Analysis of Abnormal Returns

In order to test whether the cross-border effects are robust to controlling for differences in the characteristics of cross-border and domestic acquisitions, we run cross-sectional regressions of the 3-day cumulative abnormal returns (CAR) for targets and bidders on the combined samples of cross-border and matched domestic acquisitions against a cross-border dummy and controls for the different bid characteristics.³⁰ Given that the cross-border effects appear to vary over time (see Table 1, Panel C), we also include four time dummies in the regressions: Early ‘90s (1990–1994), Late ‘90s (1995–1999), Early ‘00s (2000–2004) and Late ‘00s (2005–2008), with acquisitions during the 1980s (1980–1989) captured by the intercept. The regression model is as specified in equation (4) below:

(4)

Given Relative Size is missing for approximately half the target firms, we do not include this variable in the reported results for target firms.³¹ In a second set of regressions, we explore whether the cross-border effects still vary with the nationality of the target or bidding companies once we also control for bid characteristics. The regression model is specified as follows (equation (5)):

(5)

EEA refers to countries within the European Economic Area (other than the UK), and RoW to the Rest of the World.

Regression output from the cross-sectional analysis of target company abnormal returns is presented in columns 1 and 2 of Table 5. We find bid characteristics to have a significant impact on target abnormal returns, with targets gaining less when paid in shares, if their company is large, or where the acquiring company owns shares in the target company prior to the bid. Target returns are also significantly higher during the 2000–2008 period than during the 1980s. However, even when controlling for bid characteristics, the target company cross-border effect remains highly significant, at almost 9.5 percentage points. In the second regression we also control for the location of the target firm. We find the cross-border effect still to be higher for US than for UK targets, although after controlling for bid characteristics the cross-border effect for the small sample of targets from the rest of the world (i.e., non-US or European) turns insignificantly negative. Thus, while differences in the bid characteristics of cross-border and domestic acquisitions have a significant impact on target returns, they do not appear to fully explain the target company cross-border effect or why the cross-border effect varies with the location of the target firms.

Results for the bidding company shareholders are reported in the last two columns of Table 5. We find bidder CARs to be significantly higher in cash-financed acquisitions than in acquisitions with other forms of payment, while none of the other bid characteristics appear to have a significant impact on the level of bidder cross-border effects. Controlling for bid characteristics, the residual bidding company cross-border effect is still above 1 percentage point, but loses statistical significance. In the final regression, we observe that the bidding company cross-border effect is significantly positive for US bidders acquiring into the UK (at 1.8 percentage points), and particularly high for European bidders acquiring into the UK, for whom the mean cross-border effect is a significant 6.2 percentage points. The bidder cross-border effect for UK bidders and for bidders from the Rest of the World is indistinguishable from zero. Thus, differences in bid characteristics – and in particular the method of payment – appear to have a significant impact on the bidding company cross-border effect. However, we still observe differences in the bidder cross-border effect depending on the nationality of the bidder. We next analyse the cross-border acquisitions in more detail, to ascertain whether bidders’ international experience, exchange rate effects or national corporate governance characteristics may help explain the cross-border effects.

(i) Sample Characteristics of Cross-Border Acquisitions

Descriptive statistics on the characteristics of the cross-border acquisitions are reported in Table 6. We study the impact of market access on abnormal returns by controlling for whether the bidding company had operations in the target country prior to the acquisition. We hand-collect this data from annual reports from the period prior to the date of the bid announcement. We are able to confirm that 56.2% of bidders (or 46.6% of bidders, based on the target sample) already had operations in the target country prior to the acquisition, while 10.9% (7.6%) of bidders entered a new market through the acquisition. For 32.9% (45.8%) of bidders we are unable to ascertain whether they had operations in the target country prior to the acquisition, and we treat this as a separate, residual category. We also check whether cross-border bidders have previously undertaken a cross-border acquisition, using data from Thomson ONE Banker; 70.6% (50.6%) of bidders had prior cross-border acquisition experience.

Several prior studies suggest exchange rate changes may have an impact on abnormal returns from cross-border acquisitions, although the evidence from prior literature, as discussed in Section 2, is mixed. While both Harris and Ravenscraft (1991) and Conn et al. (2005)‘ … measure the strength of the buyer's home currency relative to the [target's home currency] as the proportionate deviation from the average exchange rate for the sample period’ (Harris and Ravenscraft, 1991, p. 832), we would argue this measure is unsatisfactory. By comparing the exchange rate to the average for the sample period, observations during the early parts of the sample period will be compared to predominately future exchange rates, while observations during the latter parts of the sample period will be compared to historic exchange rates. There is thus an inconsistency in this approach. Furthermore, this approach seems to implicitly assume that managers in the early parts of the sample period are able to forecast exchange rate movements, and time their acquisitions to when their currency is strong relative to future exchange rates. In order to avoid such look-ahead bias, we use the change in the exchange rate over the 12 months prior to the date of the bid announcement as our measure of exchange rate change.³² Contrary to the predictions of Froot and Stein (1991) of companies acquiring abroad when their currency is strong, for the sample of targets, the currency of the bidder on average fell by a significant 1.43% relative to the currency of the target during the year prior to the acquisition. However, for the sample of bidders, the mean exchange rate change is positive but insignificant.

We next explore the differences in the governance characteristics of the bidder and target countries, analysing the impact of legal origin, the level of anti-director rights, the rule of law and the overall level of shareholder protection, which combines the level of anti-director rights and the rule of law. Following Bris and Cabolis (2008) and Kuipers et al. (2009), we obtain data on international variations in corporate governance regimes from La Porta et al. (1998). We also control for the quality of accounting information, using data from La Porta et al. (1998),³³ with variable definitions given in the Appendix. In the bottom section of Table 6, we report descriptive statistics for the governance characteristics of the main countries. The UK, with its English legal origin, scores 5 (out of 6) on anti-director rights, 8.57 (out of 10) on the rule of law and 4.29 (out of 6) on shareholder protection. The Centre for International Financial Analysis and Research, as reported in La Porta et al., rate the accounting quality in the UK at 78 (out of 90).

Descriptive statistics for the differences in country corporate governance characteristics between bidder and target countries are reported in the middle section of Table 6. For the target sample we generally find the bidders to have lower levels of anti-director rights and shareholder protection overall, but a higher level of rule of law than the targets, while for the bidder sample the results are reversed. This reflects the sample construction, focusing on cross-border acquisitions into and out of the UK. A large proportion of our cross-border targets are UK companies, and with the UK having high governance standards, it is not surprising that, in our sample, cross-border bidders on average come from countries with lower governance scores than the targets. The results are reversed for the sample of bidders, where again UK companies account for a large proportion of the sample. As not all overseas target countries have similarly high corporate governance standards, we observe that the cross-border bidders in the sample on average have higher governance scores than the targets. The exception is the rule of law, where the UK is rated lower than several of the other countries included in the study.

Before analysing the impact of these cross-border characteristics on the target and bidding company cross-border effects, we analyse the correlations between the various variables. The correlation matrices for targets and bidders are reported in Panels A and B of Table 7, respectively. We find no significant correlations between either bidder prior cross-border acquisition experience or exchange rate changes and the cross-border effects for either targets or bidders. We do, however, find target cross-border effects to be higher where the bidder already has operations in the target country. As suggested by Aybar and Ficici (2009), companies may make better acquisitions with local market knowledge, and there may be more scope for synergies if a company already has operations in the market. This may account for the higher target cross-border effects when bidders already have operations in the target country. However, the correlation matrix for bidders in Panel B suggests the bidder cross-border effects are higher when the company enters a new market, consistent with prior evidence of e.g., Doukas and Travlos (1988). Market access thus appears to be valuable to cross-border bidders.

We also find target cross-border effects to be higher in English legal origin countries and to be positively correlated with the strength of anti-director rights, shareholder protection and accounting quality in the bidder country relative to that of the target country, though targets gain less where the rule of law is stronger in the bidder than in the target country. However, none of the country corporate governance characteristics appear to be significantly correlated with bidder cross-border effects. Not surprisingly, we find the various governance variables to be highly correlated, with several of the correlation coefficients exceeding 0.6. Including all the country variables in the same regression analysis could lead to problems of collinearity, and we therefore analyse the impact of the country governance variables one at a time. The analysis in Section 5 suggested bid characteristics significantly affect the level of target and bidder abnormal returns, and we therefore also control for these factors in the cross-sectional analysis of the cross-border effects below.

(ii) Cross-Sectional Analysis of the Impact of Cross-Border Characteristics on Cross-Border Effects

We analyse the impact of whether the bidder has prior operations in the target country or not, of bidders’ prior cross-border acquisition experience, of exchange rate effects, and of the differences in the quality of accounting and corporate governance systems in the bidder and target countries using a regression model specified as follows (equation (7)):

(7)

where β₁ captures cases where the bidder had operations in the target country prior to the acquisition, β₂, cases where the bidder had no prior operations in the target country, β₃, cases where the bidder had prior cross-border acquisition experience, and β₄ captures the change in the exchange rate between the bidding and target countries’ currencies over the 12 months prior to the date of the bid announcement. Relative size is missing for a large number of targets (and is not significant in any model for targets), and we therefore report results for targets excluding this variable, in order to maintain a larger sample size.³⁴ We incorporate dummy variables for the location of the companies (non-UK EEA, US or Rest of the World) and time periods (Early ‘90s, Late ‘90s, Early ‘00s and Late ‘00s), with UK companies in acquisitions during the 1980s captured by the intercept. The other bid characteristics controlled for are as in Section 5 above.

In columns 2 to 6 of Table 8 for targets, and columns 8 to 12 for bidders, we expand the model to incorporate accounting and corporate governance variables, capturing the differences between bidder and target country characteristics in terms of accounting quality, English legal origin, anti-director rights, rule of law and the overall level of shareholder protection, respectively.

The results for targets are reported in columns 1 to 6 of Table 8. While the correlation matrix in Table 7 suggested target returns are higher where the bidder has existing operations in the target country prior to the acquisition, the coefficient, while positive, is not statistically significant in the multiple regression analysis. Similarly, neither the bidders’ prior cross-border acquisition experience nor exchange rate changes are found to have a significant impact on target cross-border effects. However, consistent with the results in Table 1, we find significant differences in target cross-border effects depending on the nationality of the target, with particularly high gains for US targets, and to a lesser extent non-UK EEA targets. We also find significant time trends in the cross-border effects, with target abnormal returns significantly higher during the early 2000s than during the 1980s. We also find bid characteristics to significantly impact the target company cross-border effects, with significantly lower gains to target company shareholders when they are paid in shares, where the target company is large, and where the bidder owns shares in the target prior to the bid announcement.

In column 2, we incorporate the effect of accounting quality. Consistent with Bris and Cabolis (2008), we find the target company cross-border effect to be significantly higher where the accounting quality in the bidder country exceeds that of the target country. Bris and Cabolis argue that ‘ … improvements in accounting standards induced by consolidation in cross-border mergers are associated to larger premia’ (pp. 631–32). If targets are more likely to be undervalued in countries with low accounting quality, the positive impact on the target cross-border effect may also be a result of the targets having had low valuations prior to the bids, resulting in high returns when these undervaluations are removed.

The accounting quality proxy used in the analysis reported in column 2 is based on CIFAR data obtained from La Porta et al. (1998). To test whether the results hold when we control for the recent adoption of IFRS, we introduce two robustness tests. Firstly, we adjust the CIFAR data for the adoption of IFRS, and secondly, we use the Financial Reporting Quality Index (FRQI) developed by Tang et al. (2008) as alternative proxies for accounting quality. With both proxies, the results are virtually identical to those obtained using the CIFAR scores, with accounting quality significant at the 95% level in the analysis of target cross-border effects.³⁵

Turning to the impact of differences in bidder and target country corporate governance systems, we find non-robust results for English legal origin and the rule of law. As argued by Martynova and Renneboog (2008, p. 206), English legal origin countries provide the highest quality of shareholder protection. However, while the correlation matrix in Table 7 suggests target returns are significantly higher if the bidder comes from an English legal origin country, the coefficient on legal origin is not significant in the multivariate analysis (Table 8, column 3) when we also control for target company nationality. Similarly, the impact of rule of law is not robust, switching from significantly negative in the univariate analysis to insignificantly positive in column 5 of Table 8.

We do, however, find strong results for the impact of anti-director rights and the overall level of shareholder protection (which combines the effect of anti-director rights and the rule of law), with target cross-border effects higher where the bidder country offers shareholders stronger rights and protection than are available in the target country. Our results are consistent with those of Bris and Cabolis (2008), who argue that:

As can be seen from columns 4 and 6 of Table 8, these results hold even when we control for target nationality and other company characteristics. Overall, our model can explain up to 10.3% of the cross-sectional variation in target returns. However, while we find differences in the levels of shareholder protection to have a significant impact on target cross-border effects, we still find large country effects in the abnormal returns. The cross-border effect is significantly higher for US than for UK targets. Thus, consistent with Conn and Connell (1990) and Rossi and Volpin (2004), we find US targets to be able to extract very high premia from UK bidders.

The cross-sectional regressions of bidder returns are reported in columns 7 to 12 of Table 8. While the univariate analysis suggests a significant positive correlation between the bidder cross-border effect and new market access, the coefficient is small and no longer significant in the multiple regression. Indeed, we find none of the country governance variables to have a significant impact on bidder returns. Our findings are consistent with those of Martynova and Renneboog (2008), who in their study of European acquisitions similarly do not find corporate governance factors to have a significant impact on bidder returns. Our results are consistent with cross-border acquisitions creating more value when the bidder comes from a country with stronger anti-director rights and shareholder protection than that of the target country, but with the additional gains being reflected in target returns, rather than in the abnormal returns of bidders.

We do, however, find the relative size of the target and bidder to have a significant impact on bidder returns, with the bidder cross-border effect higher the larger the target relative to the size of the bidder. Large cross-border acquisitions thus appear to have a more positive impact on bidder returns.