Foreign Acquisitions by UK Limited Companies: Long-run Performance in the US, Continental Europe and the Rest of the World

Foreign Acquisitions by UK Limited Companies: Long-run Performance in the US, Continental Europe and the Rest of the World Alan Gregory Steve McCorriston Financial Markets Research Centre School of Business and Economics, University of Exeter October 2000

Foreign Acquisitions by UK Limited Companies: Long-run Performance in the US, Continental Europe and the Rest of the World* Alan Gregory (A.Gregory@exeter.ac.uk) Steve McCorriston (S.McCorriston@exeter.ac.uk) Abstract Based on a near-exhaustive sample of significant foreign acquisitions by UK companies over the period 1985-1994, we show that, on average, the long run performance of UK acquirers is negative, irrespective of the benchmark used. These results are also robust to the use of a conditional asset pricing model. We show that this under-performance by UK acquirers appears to be attributable to poor average returns on acquisitions in the US. Acquisitions within the EU (outside the UK) show some weak evidence of being valuereducing, while acquisitions in other parts of the world show returns not significantly different from zero. Keywords: Foreign Acquisitions; Long-Run Performance. JEL Classification: G34 Contact Author: Professor Alan Gregory, School of Business and Economics, University of Exeter, Rennes Drive. Exeter EX4 4PU, Exeter, Devon, England, UK. *The authors would like to acknowledge the helpful comments of participants at the EFM Conference in Athens, and of Mario Levis (City University, London) who acted as discussant of the paper. 2

Foreign Acquisitions by UK Limited Companies: Long-run Performance in the US, Continental Europe and the Rest of the World Executive Summary This study looks at the long run performance of UK listed firms making significant acquisitions overseas during the period 1985-1994. Using a near-exhaustive sample of such significant UK acquisitions abroad the results indicate, on the whole, negative abnormal returns to acquiring firms over the 3 years following takeover. Results from acquisitions in the US are significantly negative whatever benchmark is employed for measuring abnormal performance, while those involving the EU and the rest of the world are negative though not statistically significant. Apart from the geographical location of the target, there appears to be little that explains the poor abnormal performance apart from the relative size of the target, which is of some significance in takeovers of US targets only. For these cases there is some evidence that takeovers where targets are large relative to the acquirer are associated with weaker performance. Taken as a whole, the evidence implies that UK companies make poor acquisitions in the US, and that these acquisitions may be particularly poor when the US targets are relatively large. 3

Foreign Acquisitions by UK Limited Companies: Long-run Performance in the US, Continental Europe and the Rest of the World Introduction A key feature of the world economy since the mid-1960s has been the remarkable growth of Foreign Direct Investment (FDI). The growth of FDI has out-stripped the growth of income four-fold and trade three-fold. The growth in FDI has been particularly marked since the mid-1980s with the world economy witnessing a dramatic surge such that FDI has become the most common means of serving foreign markets. Indeed, in terms of the recent focus on globalisation, globalisation of production now exceeds globalisation through trade. Furthermore, the remarkable growth of FDI has involved most developed countries. For example, the US witnessed a surge in both FDI outflows and inflows with the latter increasing so rapidly that the US became a net importer of FDI in the late 1980s. In the case of the EU, FDI both within in the EU and between EU and non-eu countries increased rapidly. The only main exception to the simultaneous growth of FDI involving developed countries relates to Japan that witnessed considerable growth in FDI outflows while FDI inflows into Japan remained at low levels. A key characteristic of the dramatic growth in FDI since the mid-1980s is the form it has taken. FDI can take a variety of forms including the establishment of green-field sites and joint ventures. However, the most prevalent form of FDI is via cross-border acquisitions. For example, in the US, on average over the 1984-1995 period, cross-border acquisitions accounted for over 90 per cent of US FDI inflows. In the EU, cross-border 4

acquisitions have also dominated FDI flows involving both EU and non-eu countries. While there may be some variation between countries and industries, the dominance of cross-border acquisitions is such that we can relate the surge of FDI since the mid-1980s as being synonymous with a surge in international acquisitions. The aim of this paper is to consider the long-run performance of acquiring firms during the 1984-1995 period. Although some (albeit limited) research on the returns to shareholders of bidding firms involved in international acquisitions exists, we depart from this literature in two important respects. First, by focusing on a (near exhaustive) sample of UK acquiring firms, we can assess whether there is any variation in long-run performance given the geographical destination of UK acquisitions abroad. Moreover, since the UK is a leading player in international acquisitions, the study of UK acquisitions abroad is an important aspect in determining the overall success of FDI by acquisition. For example, Healy and Palepu (1993) note that, over the late 1980s, the UK was the lead acquiring nation in international acquisitions accounting for almost 30 per cent of international corporate investments over that period. (As an acquirer nation, the US accounted for around 14 per cent of total acquisitions over the same period). The focus on the UK departs from most recent studies that look at either the US or Japan as the source of acquiring firms or, more commonly, with the US as host with the returns to bidding firms varying across the source countries. Second, most recent studies have focussed on the effect of the acquisition around the event (acquisition) date. However, positive abnormal returns (if they exist) may dissipate over the long-run particularly if the acquisition, and the premium paid, is influenced by short-run factors such as the presence 5

of multiple bidders in a given acquisition, a given level of the exchange rate, changes in legislation (e.g. the US Tax Reform Act of 1986) or the perception of increased protectionism (e.g. the creation of the single market in the EU). For example, recent studies focussing on shareholders wealth in target firms following cross-border acquisitions (around the event date) have highlighted the role of the US dollar (see Swenson, 1993, and Dewenter, 1995a and 1995b, with the latter also focussing on the impact of the US Tax Reform Act of 1986). Therefore in this paper we focus on the long-run performance of UK acquiring firms, which may differ from the immediate abnormal returns associated with the acquisition event. The paper is organised as follows. In section 1, we present a brief review of the literature on wealth effects on bidding firms involved in cross-border acquisitions. In section 2, we discuss the data that forms the basis of the sample used to assess the performance of UK firms while, in section 3, we present the methodology. In section 4, the results are reported while in section 5 we report the principal conclusions and discuss avenues for further research. 1. Literature review While cross-border acquisitions have received some attention in the finance literature (most commonly on the impact on the target firm) only a limited body of research exists on the impact of cross-border acquisitions on returns on acquiring firms. For example, Doukas and Travlos (1988) focus on US acquiring firms and find that, on average, there is no significant impact on bidders wealth. However, there is considerable variation over 6

their sample of firms with positive abnormal returns arising if the acquiring firm is entering new markets or new industries. In particular, the acquiring firm experiences (insignificant) negative returns if it already has experience of the country or industry it is entering. Morck and Yeung (1992) report positive abnormal returns for US firms from international acquisitions if the acquiring firm possesses firm-specific intangible assets as reflected in high levels of research and development expenditure and/or advertising expenditure the possession of these assets being most commonly associated with the characteristics of firms likely to invest abroad as outlined in the traditional literature on FDI. Kang (1993) investigates the abnormal returns of Japanese bidders in the US and finds positive abnormal returns to Japanese firms. More recently, Eun et al. (1996) have shown that the returns to acquiring firms are likely to vary across countries. Examining cross-border acquisitions in the US, they show that bidding firms sourced from Japan experienced positive abnormal returns while UK firms experienced considerable negative abnormal returns. Acquiring firms based in Canada experienced mildly positive abnormal returns that were considerably below those experienced by Japanese firms. These studies suggest that positive abnormal returns are likely to vary depending upon the characteristics of the investing firms, the country of origin, and the country and/or industry in which the acquiring firm is investing in. However, all these studies are linked by a similar characteristic i.e. they all measure the impact on the acquiring firm around the date of the acquisition event. In this paper, the focus is on the nature of the returns of the acquiring firm over the long-run as any positive impact may be determined by factors corresponding with the event date. Such an extension is particularly relevant if the 7

acquisition is intended to be value-creating (say, through synergy). Clearly, aside from the immediate impact of the acquisition event, identifying the long-run performance of international acquisitions is important in assessing the overall impact of FDI. 2. Data The sample is drawn from the set of all overseas acquisitions recorded by Amdata with bid values and sales of acquirer and target companies available. In order to ensure that only economically significant deals are analysed, we use the cut-off that target sales must be at least 5% of acquirer s sales in the financial year pre-acquisition. This leads to an initial sample of 365 acquisitions. Depending on the method used to calculate post-bid abnormal returns, we impose additional data requirements on this initial sample. In the case of size-matched buy-andhold returns, we require the market capitalisation of the firm from the London Business School Share Price Database (LSPD). For size-and-book-to-market matched returns, we also require the book-to-market ratio to be available from Datastream. In the case of the CAPM and Fama-French regressions described below, used to estimate Jensen alpha measures and cumulative abnormal returns (CARs), these data are not required but a minimum of 12 months post-bid returns are necessary in order to estimate the model parameters. This increases to 24 months in the case of the conditional models. In all, this results in a sample size of 338 firms and 325 firms respectively for the characteristicmatched samples, 354 firms for the unconditional pricing model samples, and 330 firms for the conditional pricing models. 8

3. Method Our first and main group of tests involves the use of reference portfolios of firms with similar size and similar market-to-book ratios. For the size-matched analysis, we form decile reference portfolios sorted on market capitalisation. For the size and book-tomarket analysis we form quintile reference portfolios, the first sorted on size (market capitalisation) and the second sorted 5 x 5 portfolios sorted both on size and book-tomarket ratios as at the 31 st December of the year t-1. All share returns and market capitalisation data are from the London Business School Share Price Database (LSPD), whilst all book-to-market ratios are from Datastream. Reference portfolio and returns are calculated using the buy-and-hold method described in Lyon et al, (1999, p. 169) with returns being value-weighted: R bh ps s+ τ ( 1+ R ) V is it 1 nt t= s Vis τ = (1) nt where s is the beginning period, τ is the period of investment in months, R it is the return on security i in month t, and V is is the market capitalisation of security i in the reference portfolio in month s, the first period for the return calculation. This represents the return on a passive investment portfolio. We define the expected return on acquirer i, [E(R iτ )] as the reference portfolio buy-andhold return given by (1). Abnormal returns are then defined as: AR = R E( R ) (2) iτ iτ iτ 9

We use two sets of reference portfolios following the evidence in Loughran and Ritter (1999) that benchmark portfolios formed on size alone capture around 90% of true abnormal returns, as opposed to the approximately 80% captured by size and book-tomarket benchmarks. Returns are then calculated for up to 36 months post completion. Alternative measures of performance in event time are the CAPM or Fama-French three- factor model. These are given by: ( ) R R = α + β 1 R R + ε i (3) iτ fτ i i mτ fτ τ ( ) R R = α + β R R + γ SMB + δ HML + ε iτ fτ i i mτ fτ i τ i τ iτ (4) where R ft is the monthly return on three-month UK Treasury bills, R mτ is the return on the (value weighted) FT All-Share Index, SMB τ is the difference in return between the portfolio of small and large firms respectively, and HML τ is the difference in return between the portfolio of high and low book-to-market firms respectively. The SML and HML factor portfolios are formed using the universe of UK stocks for which market capitalisations and returns, and book-to-market ratios are available on the LSPD and Datastream respectively, following the method described in Davies et al (1999) 1. In both cases, the model is estimated on 36 months post completion returns to provide an estimate either of an equivalent to the Jensen alpha measure, or a CAR measure of performance. Although we report the latter for completeness, we do not put great 1 This study of 500 large UK firms shows that the three factor model has better explanatory power for the UK than either the CAPM or the APT. 10

emphasis on these results given that CARs may be biased measures of long-run performance (Kothari and Warner, 1997; Lyon et al, 1999). Recent work has shown that the assumption of unconditional asset pricing models may be flawed (Jaganathan and Wang, 1995). This has been recognized in the fund manager performance literature (e.g. Ferson and Schadt, 1996). If the true asset pricing process is a conditional one, a failure to take account of this will result in the bad model problem (Fama, 1998). In such circumstances, unconditional models of stock returns could confuse abnormal performance with time variations in risks or risk premia. To allow for this, we re-estimate our results under the assumption that beta coefficients are conditioned on the conditioning variables used in Ferson and Schadt (1996). These are, respectively with their UK equivalents: (1) the lagged level of the one-month Treasury bill yield; (2) a lagged dividend yield (the yield on the FT All-Share Index); (3) a lagged measure of the term structure of interest rates (the UK ten year Government Bond rate); (4) a lagged measure of quality spread in the corporate bond market given by the Datastream yield on corporate bonds minus the gilt yield, and (5) a dummy for the month of January. The conditional models can be estimated in event time as: Riτ ( R R ) β ( z [ R R ]) R fτ = αi + β 1i mτ fτ + 2i τ -1 mτ fτ + εiτ (5) Riτ R fτ = αi + β 1i + εiτ ( R R ) + γ SMB + δ HML + β ( z [ R R ]) mτ fτ i τ i τ 2i τ -1 mτ fτ (6) 11

where z t-1 is the vector of conditioning variables described above. Again the model is estimated on the 36 month post completion returns. The t-test statistics are the Brown and Warner (1980, p251-2) Crude Dependence Adjustment test for the CARs, and a simple cross-sectional t-test (Strong 1992, p545) for the buy-and-hold returns. In cases where the adjusted Jensen alpha is estimated from a regression (models (3) to (6)), the t-test statistics are formed from the standard error of the alpha coefficient from the regression. Under the null hypothesis that!α i = 0, i=1,n: [ α i α i ] t = (! / SE(! )) / N (7) will have an approximate (asymptotic) normal distribution (where N is the number of acquirers) under the usual assumptions about the error terms. 4. Results In Table 1 we present the overall results using each benchmark. This shows that however acquirer performance is measured, the 3-year post acquisition returns for UK foreign acquirers are significantly negative. In all cases except for the size-matched buy-and-hold benchmark, which shows the smallest negative returns, these are significant at the 5% level in a two-tailed test. The size-matched 36 month return is - 6.30% (t = -1.693, significant at the 10% level in a two-tailed test), while the size and book-to-market matched returns are 9.89% after 36 months. The CAR measures are similar, whether the CAPM or Fama-French three factor model is used, but more negative than the buy-and- 12

hold abnormal returns at -14.30% and -13.70% respectively. Inspection of the median returns in the last four rows of Table 1 suggest that the mean buy-and-hold results may be being biased upwards by outliers. The median abnormal returns are -13.67% and -15.64% from the size and size-and-book-to-market reference portfolios, well below the mean returns. A simple sign test shows that the number of negative abnormal returns significantly exceeds the number of positive abnormal returns. For the CAR-based results, the median and mean returns are reasonably close, and a sign test confirms that the number of negative CARs significantly exceeds the number of positive CARs. Table 1 about here Figure 1 shows how these returns evolve through event time for the buy-and-hold and CAR based measures. Returns are not significantly different from zero using any of the benchmarks after 12 months (see Table 1), but generally start to decline after this period. Although abnormal returns are not significantly negative after 2 years using the buy-andhold method, CARs are significantly negative. The rate of decline in CARs for the third year is roughly equivalent to the second year rate of decline. However, the buy-and-hold negative abnormal returns appear to accelerate during the third year. Figure 1 about here 13

The estimated alphas from the unconditional CAPM and Fama-French regressions from Table 1 are -0.36% and -0.33% per month, figures broadly in line with the CAR estimates of abnormal performance. Both coefficients are highly significant. When the regressions are estimated in conditional form (from (5) and (6) above), the alpha estimates become more negative in both cases, and remain highly significant, with the coefficients being -0.44% and -0.42% respectively. This suggests that time varying risk exposures or risk premia do not explain the abnormal performance of acquirers. The results described above are particularly striking, given that until very recently foreign takeovers by UK companies almost universally involved cash, rather than equity. Most studies in the US and the UK document that domestic acquisitions financed by cash exhibit returns not significantly different from zero. 2 If the pattern found for cross-border takeovers was similar to that found in domestic acquisitions, we would have expected to find zero abnormal returns for our sample. Our contrary finding of significant negative returns provides a motivation to partition the data to determine whether these returns are time, geographic or merger-type specific. Geographically, we partition our data according to whether the takeover is of a US, non-uk European Union (EU) or other ( Rest of the World, RoW) domiciled company. Last, we partition our sample into conglomerate and non-conglomerate takeovers, our definition being based upon whether the two-digit SIC code of acquirer and target are coincident. All partitioned results use buy-and-hold (mean and median) results, with all t-tests being simple cross-sectional tests. 2 Although Gregory (1997) in a study of UK domestic acquirers for 1984-92 provides weak evidence that cash acquirers may under-perform, in that returns are just significantly negative under some benchmarks. 14

First, we show the results by announcement year in Table 2; the data are presented graphically in Figure 2. Although there is considerable variation in the event time returns by cohort year, the majority of years show negative abnormal returns. Although the intertemporal variation of the abnormal buy-and-hold returns differs according to whether size or size-and-book-to-market matching is employed, there is no suggestion of any clear pattern emerging through time. The analysis by geographical region highlights the fact that for both buy-and-hold measures of abnormal returns, mean US-target abnormal returns are significantly negative for both size and size and book-to-market matched returns 3. Using size-matching, UK firms acquiring in the US experience negative abnormal buy-and-hold returns of 12.04%, and 13.17% on a size and book-to-market matched basis. Both mean abnormal returns are highly significant. Furthermore, the sign test on the number of negative versus the number of positive abnormal returns is significant in both cases. For size matching, 72 firms experience positive abnormal returns against 130 that experience negative abnormal returns. With size and book-to-market matching the numbers are 75 positive and 118 negative. In the case of continental European acquisitions, the evidence is more mixed. For both benchmarks, abnormal returns are negative (-3.47% sizematched; -10.87% size and book-to-market matched), but not significantly so. However, the numbers of positive: negative abnormal returns show a tendency to exhibit significance (at the 10% level on a size-adjusted basis, but at the 5% level on a size and 3 Although not reported in the sub-analyses, these results are robust to the measurement of performance using the conditional and unconditional CAPM and FF three factor models, all of which exhibit a significantly negative Jensen-alpha for US acquisitions. 15

book-to-market adjusted basis). Last, we conducted simple t-tests for differences between US and EU acquisitions; these fail to show any significant differences between the two geographical areas in either case. 4 For the rest of the world, abnormal returns are positive but not significantly so for both models; the number of positive and negative abnormal returns is almost evenly split. T-tests for differences show that at the 10% level, RoW acquirers abnormal returns are significantly better than US acquirers returns, although there are no significant differences compared to EU acquirers returns. However, given the small number of RoW acquisitions (39 cases), it is difficult to make any strong claims for the success or otherwise of these takeovers. We also partition our acquirers returns on the basis of whether they are in the same industry (same 2-digit SIC code) or not. These figures are also reported in Table 3. The results may be seen as somewhat surprising, but are in line with the finding of Agrawal et al. (1992) who investigated domestic US acquisitions and found that non-conglomerate acquirers have poorer post-acquisition performance. In our sample, same industry acquirers show significant negative abnormal returns for both benchmarks (-9.05% and 11.32% for size and size-and-book-to-market), whereas out-of-industry acquirers show insignificant performance of -0.01% and 6.72% respectively. However, when numbers of positive and negative performance are looked at, the sign test for the size and book-tomarket benchmark is almost significantly negative at the 5% level. 4 Further analysis shows that only the unconditional FF alpha is significant for EU acquisitions. Despite the significant negative performance of US acquisitions in all cases (see footnote 2), no t-test for differences between the US and the UK is significant. 16

One feature of the above analysis is that the partitioned results are somewhat sensitive to the benchmark used. Whilst the post-acquisition performance of UK acquirers in the US is unambiguously negative, and that of acquirers in the RoW seems not significantly different from zero, the significance of the post-acquisition performance of EU acquirers and, to some extent at least, out-of-industry acquirers, shows some signs of being sensitive to the benchmark used. Furthermore, the relative scale of any performance is also sensitive to the benchmark chosen, with EU and different industry acquisitions faring comparatively worse when book-to-market matching is added to size matching. For this reason, we ran additional regression tests with buy-and-hold abnormal returns as dependent variables on the following independent variables: EU = dummy variable equal to one if the acquisition is made in the EU ROW = dummy variable equal to one if the acquisition is made outside the EU or US SAME = dummy variable equal to one if the acquisition is made in an industry with the same 2-digit SIC code MCAP = natural logarithm of the acquirer s market capitalisation in the January preceding the takeover RELSIZE = amount paid for the target divided by the market capitalisation of the acquirer In many studies, heteroscedasticity is controlled for by using weighted least-squares, with weights being the inverse of the standard prediction error in the case of CARs being used as the dependent variable. This is not possible when using the characteristic matched buy-and-hold method to estimate abnormal returns, so instead we report t-ratios that are 17

adjusted for heteroscedasticity using the White (1980) method. The results of these regressions are reported in Columns 2-2 and 6-7 of Table 4. Unfortunately, the F-statistic of the size-matched regression is only significant at the 10% level, whilst that of the sizeand-book-to-market matched regression is not significant. 5 This weak result is somewhat puzzling and led us to explore further whether the control variables have a significantly different effect on our geographical areas. Preliminary regressions (not reported) to investigate this matter suggested that RELSIZE has a significant negative association with abnormal performance in the case of US acquisitions but a weakly positive association outside the US. MCAP appeared to have no explanatory power in these regressions. We therefore re-ran the regression tests including a slope shifting dummy, USRELS which is equal to RELSIZE if the takeover was of a US target, and zero otherwise. The results from these regressions are reported in Columns 4-5 and 8-9 of Table 4. These regressions provide weak evidence that relatively large US acquisitions are associated with poor post acquisition abnormal performance when the size-adjusted benchmark is employed. The co-efficient on USRELS is negative but significant at only the 10% level in a two-tailed test. The regression itself is significant at the 5% level although the adjusted R-squared is only 0.026. When the size and book-to-market benchmark is employed, USRELS is again significantly negative at the 5% level, although the regression itself is only significant at the 10.8% level. It would be inappropriate to make any strong claims on the basis of such regressions, but they do at least confirm that the size of the acquirer does not have a 5 We also ran both weighted (with weights equal to 1/standard error of the alpha estimate) and unweighted regressions with our four Jensen alpha measures as dependent variables. These had significance levels less than 10% and so are not reported here. 18

particular influence on our conclusions, whilst it appears that the relative size of the acquisition has a different effect depending on whether or not the target is a US company. 5. Conclusions The back-drop to this paper has been the dramatic surge in FDI that has occurred in the world economy since the mid-1980s and has involved, in the main, cross-border acquisitions. The specific aim of this paper has been to consider the long-run performance of acquiring firms. Using a near-exhaustive sample of significant UK acquisitions abroad the results indicate, on the whole, negative abnormal returns to acquiring firms over the long-run with results from acquisitions in the US being significantly negative while those involving the EU and the rest of the world are negative though not statistically significant. The results indicate that geography seems to matter though investing in a related industry does not. There is some evidence that there is a relationship, albeit a weak one, between the relative size of the target and acquirer performance that differs between US and non- US acquisitions. Relatively large US acquisitions tend to be associated with poor post acquisition performance, whereas there is no such relationship for acquisitions outside the US. Overall, the results appear to be consistent with those of Eun et al. (op. cit.) that suggest negative abnormal returns for UK firms investing abroad. However, the results presented in this paper differ in an important way from those of Eun et al, and other studies more generally, in that the emphasis here has been on measuring the long-run performance of acquiring firms. Re-visiting Figure 1 shows that UK firms experienced positive abnormal returns around the acquisition date. This is consistent, to 19

varying degrees, with all benchmarks used. The results reported in this paper indicate that that performance measures can change in direction and certainly in degree over the longrun. In contrast, Eun et al. reported a negative performance around the acquisition event only. This paper represents a useful first cut of the performance of UK firms over the long-run. There are obvious extensions to be pursued. First, the multinational characteristics of UK firms could be explored further. Most studied of FDI highlight the importance of firm-specific assets as measured by, for example, research and development expenditure. Given the widely varying experience of firms in the sample, it would be worth exploring whether certain firm-specific characteristics determine the long-run performance of acquiring firms 6. Second, much previous research has focused on the impact of foreign acquisitions on the target firm and, in some cases, has highlighted the distribution of wealth gains between bidding and target firms. While positive abnormal returns are commonly reported for target firms, it would be worthwhile extending the current analysis to measure the performance of target firms and to assess the distribution of the wealth effects over the long-run. REFERENCES: Agrawal, A., Jaffe, J.F. and Mandelker, G.N. (1992), 'The Post-Merger Performance of Acquiring Firms; A Re-Examination of an Anomaly', Journal of Finance, 47 pp 1605-22. 6 Although such an analysis of the current sample is not possible as disclosure of research and development expenditure only became mandatory in the UK for accounting periods beginning on or after 1 st January 1989 (SSAP 13, revised). 20

Barber, B.M. and J.D. Lyon (1997), Detecting Long-Run Abnormal Stock Returns: The Empirical Power and Specification of Test Statistics, Journal of Financial Economics, 43, 341-372. Brown, S. and J.B. Warner (1980), Measuring Security Price Performance, Journal of Financial Economics, 8, 205-258. Chistopherson, J.A., Ferson, W.E. and Glassman, D.E.. (1996), Conditioning Manager Alphas on Economic Information: Another Look at the Persistence of Performance, National Bureau of Economic Research, Working Paper 5830 Conn, R.C., and Connell, F. (1990), International Mergers; Returns to US and British Firms, Journal of Business Finance and Accounting, Winter 1990, pp. 689-711. Davies, R., S. Unni, P. Draper and K. Paudyal (1999), The Cost of Equity Capital, Chartered Institute of Management Accountants, London. Dewenter, K.L. (1995a), Do Exchange Rate Changes Drive Foreign Direct Investment? Journal of Business, 68, pp. 405-433. Dewenter, K.L. (1995b), Does the Market React Differently to Domestic and Foreign Takeover Announcements? Evidence from US Chemical and Retail Industries, Journal of Financial Economics, 37, pp.421-441. Doukas, J. and N.G. Travlos (1988), The Effect of Corporate Multinationalism on Shareholders Wealth: Evidence from International Acquisitions. Journal of Finance, 43, pp. 401-417. Eun, C.S., R. Kolodny and C. Scheraga (1996), Cross-border Acquisitions and Shareholder Wealth: Tests of the Synergy and Internalization Hypthesis. Journal of Banking and Finance, 20, pp. 1559-1582. Fama, E.F. and French, K.R. (1993), Common Risk Factors in Returns on Stocks and Bonds, Journal of Financial Economics, 33, pp 3-56. Fama, E. and K. French. (1995), Size and Book-to-Market Factors in Earnings and Returns, Journal of Finance, L, 1 (March), pp. 131-155. Fama, E.F. and French, K.R. (1996), Multifactor Explanations of Asset Pricing Anomalies, Journal of Finance, 50, pp 131-155. Ferson, W.E. and Schadt, R.W. (1996), Measuring Fund Strategy and Performance in Changing Economic Conditions, Journal of Finance, 51:2, June, 425-61. 21

Gregory, A. (1987) Divisional Performance Measurement with Divisions as Lessees of Head Office Assets, Accounting and Business Research, Summer, pp. 241-6. Healy, P.M. and K. G. Palepu (1993), 'International Corporate Equity Acquisitions: Who, Where and Why?' in K.A.Froot (ed.) Foreign Direct Investment. University of Chicago Press. Chicago. Kang, J-K. (1993), The International Market for Corporate Control, Journal of Financial Economics, 34, pp. 345-371. Kothari, S.P. and J.B. Warner (1997), Measuring Long-Horizon Security Price Performance, Journal of Financial Economics, 43, 301-339. Lyon, J. Barber, B. and C.-L. Tsai (1999) Improved Methods for Tests of Long-Run Abnormal Stock Returns, Journal of Finance, 54(1), February, 165-201. Morck, R. and B. Yeung (1992), Internalization: An Event Study, Journal of International Economics, 33, pp.41-56. Strong, N. (1992), Modelling Abnormal Returns: A Review Article, Journal of Business Finance and Accounting, June, pp. 533-553. Swenson, D.L. (1993), Foreign Mergers and Acquisitions in the United States in K.A. Froot (ed.) Foreign Direct Investment. University of Chicago Press. Chicago. White, H. (1980), A Heteroscedasticity-Consistent Covariance Matrix Estimator and a Direct Test for Heteroscedasticity, Econometrica, Vol. 48, 1980, pp. 817-838. 22

Table1: Overall results. The table shows 36 month event time mean and median buyand-hold abnormal returns and cumulative abnormal returns, and Jensen alpha-type measures of performance calculated as described in the text BUY-AND-HOLD CUMULATIVE ABNORMAL RETURN ABNORMAL RETURNS (AR) (CAR) Size Size & BTM CAPM FF 12 month return -0.0043-0.0076-0.0179-0.0202 T-test -0.265-0.470-0.860-0.998 24 month return -0.0234-0.0357-0.0823-0.0800 T-test -0.865-1.308-2.906-2.897 36 month return -0.0630-0.0989-0.1430-0.1370 T-test -1.693-2.600-4.176-4.106 Jensen alpha Unconditional -0.0036-0.0033 T-test -2.928-3.255 Conditional -0.0044-0.0042 T-test -4.634-4.787 Median Return -0.1367-0.1564-0.1446-0.1314 No. Positive 132 127 115 132 No. Negative 206 198 239 222 Sign test -4.025-3.938-6.591-4.783 23

Table2: Results by cohort year. The table shows 36 month event time mean and median buy-and-hold abnormal returns for each year of announcement Year No. obs SIZE ADJUSTED BUY & HOLD ABNORMAL RETURNS Mean AR t-test Median AR No. pos. No. neg. SIZE & BOOK-TO-MARKET ADJUSTED BUY & HOLD ABNORMAL RETURNS Sign test No. Mean AR t-test Median No. No. Sign test obs AR pos. neg. 1984 4 0.0654 0.452 0.0419 2 2 0.000 4 0.0501 0.434 0.0112 2 2 0.000 1985 3-0.2157 0.000-0.1469 0 3-1.732 3 0.1341 0.000 0.2488 2 1 0.577 1986 26-0.1941-1.686-0.3461 7 19-2.353 26-0.1190-1.043-0.2840 10 16-1.177 1987 32-0.2037-2.751-0.1898 10 22-2.121 32-0.2090-2.979-0.1520 10 22-2.121 1988 63 0.0094 0.146-0.0453 29 34-0.630 58-0.0866-1.188-0.0844 25 33-1.050 1989 38-0.0793-0.861-0.1014 16 22-0.973 33-0.2680-2.674-0.2200 9 24-2.611 1990 36-0.1908-1.934-0.1098 14 22-1.333 36-0.1909-1.959-0.2469 15 21-1.000 1991 22 0.0780 0.302-0.1591 10 12-0.426 21 0.1279 0.457 0.0488 11 10 0.218 1992 26-0.1442-1.461-0.1942 9 17-1.569 26-0.0754-0.811-0.0884 11 15-0.784 1993 41-0.0312-0.328-0.2000 17 24-1.093 40-0.0003-0.003-0.1343 17 23-0.949 1994 47 0.0695 0.470-0.2013 18 29-1.605 46-0.0636-0.439-0.2638 15 31-2.359 24

Table 3: Results by country of target and by degree of diversity. The table shows 36 month event time mean and median buy-andhold abnormal returns for each year of announcement SIZE ADJUSTED BUY & HOLD ABNORMAL RETURNS SIZE & BOOK-TO-MARKET ADJUSTED BUY & HOLD ABNORMAL RETURNS Class No. Mean AR t-test Median No. No. Sign test No. Mean AR t-test Median No. No. Sign test obs AR pos. neg. obs AR pos. neg. by country: US 202-0.1204-2.983-0.1474 72 130-4.081 193-0.1317-3.156-0.1564 75 118-3.095 EU 98-0.0347-0.444-0.1458 40 58-1.818 94-0.1087-1.348-0.1795 34 60-2.682 RoW 38 0.1696 1.153 0.0293 20 18 0.324 38 0.0923 0.640-0.0480 18 20-0.324 By 2- digit SIC code: Same 234-0.0905-1.954-0.1621 86 148-4.053 224-0.1132-2.427-0.1583 86 138-3.474 Diverse 104-0.0010-0.016-0.1034 46 58-1.177 101-0.0672-1.024-0.1508 41 60-1.891 25

Table 4: Regressions of buy-and-hold abnormal returns. The Table shows the results of regressing Columns 2-3 of the Table show the regressions of buy-and-hold abnormal returns on the following on the following independent variables: EU = dummy variable equal to one if the acquisition is made in the EU; ROW = dummy variable equal to one if the acquisition is made outside the EU or US; SAME = dummy variable equal to one if the acquisition is made in an industry with the same 2-digit SIC code; MCAP = natural logarithm of the acquirer s market capitalisation in the January preceding the takeover; RELSIZE = amount paid for the target divided by the market capitalisation of the acquirer and; USRELS = amount paid for the target divided by the market capitalisation of the acquirer if the acquisition is of a US company, zero otherwise. Columns 2-5 show the results with size-matched buy-andhold abnormal returns as the dependent variable, while Columns 6-9 show the results with size-and-book-to-market matched buy-and-hold abnormal returns as the dependent variable. VARIABLE SIZE-ADJUSTED BENCHMARK: SIZE AND BOOK-TO-MARKET ADJUSTED BENCHMARK Coefficient T-RATIO Coefficient T-RATIO Coefficient T-RATIO Coefficient T-RATIO EU 0.0848 0.985-0.0368-0.344 0.0208 0.233-0.0978-0.896 ROW 0.3136 1.976 0.1921 1.501 0.2465 1.590 0.1187 0.934 SAME -0.1180-1.589-0.1267-1.717-0.0675-0.854-0.0730-0.937 MCAP -0.0236-0.799-0.0371-1.206 RELSIZE -0.0651-1.773 0.5183 1.605-0.0655-1.471 0.5288 1.523 USRELS -0.5916-1.824-0.5907-1.688 CONSTANT 0.4431 0.752-0.007-0.113 0.6633 1.067-0.058-0.893 Adjusted R- 0.014 0.026 0.004 0.013 SQUARE F - test (probability) 1.957 (0.085) 2.811 (0.017) 1.280 (0.272) 1.824 (0.108) 27

36 34 32 30 28 26 24 0.02 0-0.02-0.04-0.06-0.08-0.1-0.12-0.14-0.16 CAR Plot 18 20 22 Months CAPM FF B&H - size B&H S&BM 16 14 12 10 8 6 4 2 Ann CARs / B&H

Figure 2: Buy-and-hold Abnormal Returns by Cohort Year 0.2 0.15 0.1 0.05 B&H Abnormal return 0-0.05-0.1 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 Size mean Size&BM mean -0.15-0.2-0.25-0.3 Year 29

Figure 3: Abnormal returns by region ofacquired firm 0.2 0.15 36 month equivalent abnormal return 0.1 0.05 0-0.05 US EU RoW Size mean Size&BM mean -0.1-0.15 Region 30