The long-run performance of mergers and acquisitions : Evidence from the Canadian stock market
Paul André
Jean-François L’Her [a]
Very preliminary, please do not quote
4 November 2002
JEL Classification: G12, G14, G34
Keywords: Mergers and acquisitions, four factor pricing model, Canadian stock market
1
The long-run performance of mergers and acquisitions : Evidence from the Canadian stock market
Abstract: We study the long run performance of mergers and acquisitions from 267 Canadian firms over the period 1980 to 2000. Using various methodologies including the Fama-French three factor pricing model and a four factor pricing model, we document a significant underperformance when returns of a portfolio of Canadian acquirors are equally-weighted. This result is robust under certain conditions when returns are value-weighted
1 Introduction
The international wave of merger and acquisitions of the last decade is unprecedented notwithstanding the recent slowdown. Just in the U.S. in each of the years 1999 and 2000, there were more than 9,000 transactions worth more than a trillion dollars (Mergerstat review). Canada is no stranger to this trend. In 2000, a record breaking year, Crosbie & Co. reported close to 1,300 transactions for a total value of almost 235 billion dollars. Beyond all this activity however, there is a growing concern about the prices that are being paid and about the impact of these major transactions on future corporate performance.
Understanding the value creation process related to these transactions and the factors that drive it has been an important focus of accounting and finance research. This research can generally be divided in three different broad approaches. First, researchers (e.g., Jensen and Ruback 1983, Andrade, Mitchell and Stafford 2001) have examined the stock market reaction to mergers and acquisitions announcements. Most researchers generally agree that these transactions seem to create some value for shareholders overall. However, the gains accrue almost entirely to target shareholders, acquiring shareholders gaining nothing and sometimes losing.
More recently, researchers have been concerned with a second question, that is, the post acquisition operational performance of these transactions. Kaplan and Weisbach (1992) point out that 44% of transactions are actually undone inside a ten year period. Starting with the Healy, Palepu and Ruback (1992) paper, number of researchers have concluded that on average cash flow returns improve in the post acquisition period and that this improvement is not artificially generated by short term decisions such a cutting investments. However, more recent papers are challenging these results and suggesting that there may not be any improvements on average (Ghosh, 2001 and Bécotte, 2002).
Our paper use a third approach to examine value creation following mergers and acquisitions. We examine the long term post acquisition returns of mergers and acquisitions. Recent studies question the ability of markets to fully interpret the consequences of major transactions such as mergers and acquisitions at their announcement. Results however have been mixed, these studies being very sensitive to the methodology used. We adopt various methodologies to examine the issue and examine various explanations for the long run performance of Canadian acquirors related to firm-specific and transactions specific attributes. Overall, our results suggest that on average Canadian acquirors underperform a match control sample.
The remainder of the paper is organized as follows. In Section II, we present the literature on long-run performance. In Section III, we present the data and the methodology used to estimate the average long-run abnormal performance of Canadian acquirors. In Section IV, we expose the methodology and our results. In Section V, we present deal variables and acquiror/acquired characteristics can explained cross-sectional differences in long-run abnormal performance of acquirors. Section IV presents concluding remarks.
2Post-merger performance
Empirical research on mergers and acquisitions has generated a good deal of results as to their trends and characteristics over the last decades. A large number of event studies has demonstrated that mergers and acquisitions appear to create shareholder value but with most of the gains accruing to the target shareholders. Large scale studies in the U.S. by Jensen and Ruback (1983), Jarrell, Brickley and Netter (1988), Schwert (1996) and more recently by Andrade, Mitchell and Stafford (2001) that examine the average abnormal stock reaction in traditional short-window event studies all suggest that target shareholders are clearly winners in merger transactions. Andrade et al. (2001) examine some 3,688 transactions over the 1973-1998 period and find combined abnormal returns over the three-day period surrounding the announcement of 1.8% (statistically significant at the 5 percent level). Target shareholder abnormal returns are 16.0% while acquirer abnormal returns are not significantly different from zero. Target shareholders do even better when there is no equity financing while abnormal returns to acquiring shareholders in equity financed transactions are –1.5% and significant. These results hold for the sub-periods 1973-79, 1980-89 and 1990-98. Eckbo (1986) finds similar non significant results for successful Canadian bidders and so does Betton and Williams (2001) for a more recent sample of Canadian bidders and independent of transaction types.[b]
More recent studies have examined the long-run performance and cast some doubt on the interpretation of more traditional short-window event study findings. These studies generate a certain number of concerns with respect to market efficiency. While most authors accept that stock prices quickly adjust to incorporate the available information around merger and acquisition announcements, many believe that markets have some difficulty in properly measuring the immediate strategic fit of the combinations even more so when it is known that mergers and acquisitions are followed by important restructurings and frequent divestitures in almost half the merger and acquisition cases (Kaplan and Weisbach, 1992).
2.1Measure of long-run performance
Since Fama and French's (1992) article, financial literature considers that systematic risk, beta, is not a sufficient statistics to measure risk. Two easily recognizable measures: size (market capitalization of companies) and the book to market ratio explain a larger part of the future stock returns. As a consequence, methodologies used to measure the abnormal returns have to take into account these two risk factors. Several methodologies are possible to control for these forms of risk. The first used by Loughran and Vijh (1997) bases itself on the comparison of the geometrical return over 5 years on the acquiring company and on that of a comparable company in terms of size and book to market ratio. Rau and Vermaelen (1998) criticize this method and propose an alternative methodology.
The first limitation of Loughran and Vijh's procedure is that the acquiring company and the matched firm are going to evolve during the 5 years examined in terms of size and book to market ratio, and so is the level of risk. Rau and Vermaelen (1998) suggest that risk factors appropriate for two companies vary a lot after the initial matching. In their sample, only 19 % of companies are classified in the same portfolios of size and book to market ratio after 3 years. The second inconvenience of this methodology concerns statistical tests on the long-term abnormal performance. Indeed, Barber and Lyon (1997), Kothari and Warner (1997) as well as Lyon, Barber and Tsai (1999) showed the limits of tests relative to arithmetical or geometrical abnormal returns over a long periods. They favor a technique of bootstrap aiming to generate under the hypothesis of no abnormal returns so as to determine if observed abnormal returns are statistically different from zero. This technique consists essentially in classifying month after month the acquiring companies in portfolios of stocks having about the same characteristics in terms of size and book to market ratio. The returns on the acquiring companies are then compared with those of these reference portfolios. Abnormal returns are then calculated as in the case of event studies. Statistical tests consist in generating a distribution of abnormal returns under the null hypothesis, that is to make several random draws of stocks composing the reference portfolios and to keep in every iteration the long-term abnormal return on this stock with regard to its reference portfolio. p values associated to the abnormal return are then inferred from the empirical distribution of abnormal returns.
2.2Results of the long-run performance of mergers and acquisitions
Control for systematic risk and size, but not for the book to market ratio: Franks, Harris and Titman (1991) study 399 acquisitions over the period from 1975 to 1984. They find positive and significant long-term abnormal returns only for those leading to a weaker post-grouping market capitalization. On 304 mergers studied over the period from 1965 to 1986, Loderer and Martin (1992) observe a negative but not significant abnormal return over the 5 subsequent years. Abnormal returns are however significant when measured over 3 years. On 155 examined acquisitions, they obtain a positive, but not significant abnormal return. However, when they subdivide their sample in three sub-periods, they notice that long-term abnormal returns over 3 years are significant during the sixties, when the objective of high growth was most likely more important. On the contrary, Agrawal, Jaffe and Mandelker (1992) over the period from 1955 to 1987 find for 937 studied mergers a negative and significant abnormal return on 10.3 % over the five subsequent years. Abnormal returns on 227studied tender offers are positive, but not significant.
Control for systematic risk, size and book to market: Loughran and Vijh (1997) find that long-term abnormal returns (5 years) in the case of mergers are 15.9 % while in the case of tender offers they are +43 %. In the cases they call ambiguous, these abnormal returns are 20.8 %. Loughran and Vijh perform the same tests on a sample where there is no overlapping of events by the same company and they find for these 534 transactions that long-term abnormal returns in the case of mergers are 14.2 % while 61.3 % in the case of acquisitions.
Rau and Vermaelen (1998) confirm the fact that long-term abnormal returns are respectively negative and significant in the case of mergers and positive and significant in the case of acquisitions. However, they criticize the methodology used by Loughran and Vijh and obtain results sharply lower in absolute value, even though their results are over three years rather than five. When one compares the performance of the acquiring companies in the case of mergers with that of a portfolio of companies characterized by the same risk factors in terms of size and book to market value, it is lower by 4 % and significantly different from zero. On the contrary, the acquiring companies in the cases of tender offers present a return superior to that of the portfolios of companies of the same level of risk. Abnormal returns are however only 8.9 %.
Finally, Mitchell and Stafford (2000), as well as Ikenberry, Lakonishok and Vermaelen (2000) examine in a more general way long-term financial performance further to three types of events entailing a significant change of the number of shares in circulation: share repurchases, equity offerings and mergers and acquisitions. Mitchell and Stafford (2000) analyzes over the period 1961 to 1993 a sample of 2068 transactions. They do not distinguish the mergers of acquisitions but report a negative mean abnormal monthly returns over three years of –0.14 % and -0.04 % for equal weight and value weight portfolios respectively using calendar time abnormal returns based on the Fama-French three factor model. Ikenberry, Lakonishok and Vermaelen (2000) examine in Canada a sample of 27 acquisitions over the period from 1989 to 1995 for which more than a third of the financing were shares. Abnormal returns are negative, but not significantly different from zero over the subsequent three years.
2.3A last word on long term performance
Loughran and Vijh ( 1997 ) is the only study, to our knowledge, to examine collectively the short-term and long-term performance of the acquisitions using US data. It allows us to attempt to answer the following question: do the acquisitions of companies create of the value for the shareholders? While the answer was not ambiguous in the short term, one can imagine that long-term loss of value in the case of mergers can cancel short-term returns. Loughran and Vijh take the point of view of a shareholder of the target company and make the following hypotheses: 1) in the case of mergers, the short-term abnormal returns on the shareholders are calculated as in all event studies and they calculated long run returns by making the hypothesis that the shareholder keeps the shares during the 5 subsequent years; 2) in the case of tender offers, the short-term abnormal return for the shareholders are calculated as in all event studies and the long run returns are calculated by making the hypothesis that the shareholders reinvest the cash in the shares of the acquiring company.
In the case of mergers, short-term return is 25.8 % while that in the long run is of 14.8 %. Combined abnormal return would be 29.6 % from the date of the announcement to 5 years after. In the case of tender offers, short-term return is 24.5 % while that in the long run is of 61.5 %. Combined abnormal return would be 126.9 % from the date of announcement to 5 years after. Abnormal performance remains positive for mergers, but this last one is more of 4 times weaker than that of the tender offers. There is average value creation for the shareholders when one observes an acquisition. Loughran and Vijh (1997) note only one exception: in the case of the acquisitions characterized by high ratio of the size of the target over the acquiror, they find negative abnormal returns of 47.4 %on the full period considered (pre and post acquisition periods).
In conclusion, most of studies suggest abnormal long-term returns are negative in the case of mergers. However, in the case of tender offers and with the exception of the Loughran and Vijh ( 1997) study, abnormal returns are positive but hardly significant.
However, these differences show clearly that methodologies used to determine abnormal long-term returns have a major impact on results. We therefore examine long-run performance of mergers and acquisitions in Canada using a variety of methodologies.
3Data
3.1Data
Our data sets of mergers and acquisitions of assets was obtained from the Securities Data Corporation Worldwide Mergers and Acquisitions database. The data meet the following criteria:
1) Given that the subject of this research is post acquisition performance of merger and acquisition in Canada we begin by considering all transactions made by Canadian firms.
2) Collected observations are for 1980 – 2000 period inclusively.
3) Mergers should be completed to analyze an effect to its full extent.
4) Deal must be a merger, exchange offer or acquisition of majority interest.
5) We do not reject companies with several merger or acquisition announcements made during period.
6) We only include transactions greater than 10 million $ Canadian.
Selection criteriaRequest / Hits / Request Description
0 / - / DATABASES: All Mergers (MA, OMA, IMA)
1 / - / Date Announced: 1/1/1962 to 31/12/2000 (Custom)
2 / 18656 / Acquiror Nation : CA
3 / 4277 / Deal Value ($ Mil): 10 to HI
4 / 2752 / Deal Status : C
5 / 2208 / Deal Type : 1,2,4
6 / 971 / Form of the Deal : A, AM, EO, M
7 / 497 / Target Public Status : P
8 / 383 / Acquiror Public Status : P
9 / 373 / Percent of Shares Owned after Transaction: 50 to 100
As a result from the approximate 18,000 acquisitions (completed and uncompleted, public and private of all types including acquisitions of partial and non controlling interests) by Canadian companies, we obtained an initial sample of 373 transactions. We retain the companies which had a correspondence in the Research Insight Compustat database over the April 1980-April 2001 period. The final sample comprises 267 events (176 companies). We impose the following criteria to keep the event in the final sample: to have at least 24 monthly returns available in the post acquisition period.
Descriptive statistics
[Insert Table 1 and Figure 1]
Panel A of Table 1 reports the annual number , aggregate value and mean value of target stock for acquisitions completed during 1980-2000. Our sample includes a total of 267 acquisitions with a market value over 100 billion dollars. A noticeable trend is present in the data. Few transactions took place before 1994. During the late 1990s, the number of acquisition has continuously increased. Figure 1 plots the number of acquisitions and the total dollar value of the transactions by year. Panel B presents a breakdown by primary SIC code. As could be expected, the large majority (over 38%)of the 267 transactions and 176 firms are in the resource industries (SIC 1000). The rest of the transaction are fairly distributed across industries. Panel C presents the Top 10 transactions in our sample.
Panel D categorizes the sample by different characteristics. We find 124 (46.4%) tender offers. The smaller number of tender offers is consistent with prior studies (Loughran and Vijh 1997, Rau and Vermalean 1998). Also consistent with prior studies is the breakdown between Friendly and Hostile transactions. Only 20 (7.5%) transactions are classified by SDC Thomson Financial as being hostile or neutral, whereas, 247 (92.5%) are classified as friendly. Schwert (2000) cautions researcher of the difficulties in properly identifying the targets attitude toward the transaction. We also find 120 (44.5%) cash payments, 55 stock payments (20.6%) and 92 (34.5%) mixed payments. Further, out of the 90 (33.7%) cross-border transactions, the vast majority, 68 (25.5%), involved a US target as could be expected while 22 are with firms in other countries. We also note that few transactions, 8 (4.5%) are accounted for using the pooling of interest method of accounting for acquisitions given the much stricter rules in Canada as compared to the US. Finally, 199 (74.4%) of all transactions are between firms with the same primary SIC code, i.e., most of the transactions are in related business with 110 (41.2%) in the same 4 digit SIC Code.