Acquirer Performance
in Related and Unrelated Acquisitions
Master Thesis
Master in International Finance, Amsterdam Business School Prepared by: Gerben Stam 10481796
Supervisor: drs. J.F. Jullens Date: July 2016
Abstract
Even though literature is quite affluent with papers on mergers and acquisitions in
connection with relatedness, results are not consistent mainly due to different measures of relatedness. This study uses Survivor based relatedness and provides a clear description of the step by step process used, and suggests possible ways to improve Survivor based relatedness based on the results that were found carrying out this study. Results are presented both on an individual firm level as well as an aggregated level for (un)related acquisition groups including some subsamples within those groups that represent results for horizontal relatedness and a combination between Survivor and SIC based relatedness.
Table of Contents
Contents
1 Introduction ... 1
2 Literature... 5
3 Hypotheses and Method... 10
4 Data ... 13
5 Results ... 15
6 Discussion and Conclusion ... 18
7 References ... 21
1 Introduction
This study deals with the phenomenon of one firm acquiring another firm and the consequences in terms of profitability the takeover has on the acquiring company. There can be many reasons why firms would engage into acquisitions. One reason can be to expand in the same business in order to profit from economies of scale; these acquisitions are called horizontally related acquisitions (type 1). Another type of acquisition is when a firm takes over a company that either supplies to or buys from the acquiring company, also known as vertically related acquisitions (type 2). These acquisitions make firms less dependent on suppliers/buyers giving them more market power. Then diversifying acquisitions can be used as a means to get involved in different products or markets making the firm less vulnerable by not having to depend on the success of one product/market. Especially products/markets with a negative correlation are useful to a firm because their results typically don’t move in the same direction as correlated products/markets do, giving the firm a form of security. In some diversifying acquisitions sharing resources might be applicable (type 3), for instance a similar production process, required skill set, or similar research and development needed. But sometimes the target firm has really nothing in common with the acquirer (type 4) but is just used to spread risk or as a suitable investment project. All types of acquisitions are basically pursued to either increase profit or spread risk, with one of the reasons for firms choosing different types being that firms differ in their view on what type will bring them the best results, making studies on acquisition performance a potential asset for future acquisition decision making.
Many studies have investigated the effect of relatedness on acquisition added value for acquirers and targets. For acquirers some have found that unrelated acquisitions are more profitable or similar compared to related acquisitions (Singh and Montgomery 1987, Aggarwal and Baxamusa 2013) whereas others found related acquisitions being more profitable for acquirers (Kapland and Weisbach 1992, Lien and Klein 2006). This study makes a comparison with these results and in doing so will find support for one of these different schools of thought in this discussion.
The main research question for this study is thus to test for relatedness using a more recent sample (2011-2012) compared to the 1982-1985 dataset used in the Lien and Klein (2006) study, and to see if this dataset gives similar results; a positive relationship between relatedness and acquirer CAAR.
The purpose of this study is to see how related and unrelated acquisitions perform, however depending on the method used, type 2 and 3 acquisitions might be included in related acquisitions or not. Relatedness can be classified using Standard Industrial Classification (SIC) or North American Industry Classification System (NAICS) codes. SIC/NAICS describe all possible firm segments with coding and a description. When the segment codes are comparable within 1 of these classification systems, then firms are considered to be related. These methods don’t include type 2 and 3 acquisitions. A different method is the text analysis method that compared text between different segments, segments with similar texts are then considered to be related. Another method is looking at industry input-output tables comparing investment rates between different industries, industries with high input-output rates are considered to be related. This method is meant to establish vertical relatedness only. All previous measurements for relatedness are subjective either because they rely on the SIC/NAICS classification method judged by the Office of Management and Budget, and/or because their definition of relatedness itself is arguable and incomplete. Especially type 3 relatedness is neglected in these definitions. The ones that use SIC/NAICS data are also static as this data is not being updated on a yearly basis. One of the alternative methods developed to tackle some of these shortcomings is the survivor based relatedness, which has been chosen for carrying out this study. The survivor based relatedness’s strength lies in the fact that it avoids the discussion and subjectivity of what exactly relatedness is but focuses instead on what it does. Relying on companies making choices in their own best interest helping them to improve chances to survive and prosper.
The survivor relatedness method involves an analysis of existing company structures, the relatedness database, aimed at counting how many times industry segments are being combined within the same company which is then compared with the expected number of combinations if segment combinations would be evenly distributed in proportion to their size.
More common combinations are considered to be related, less common segment combinations are considered to be unrelated. If then the target firm holds an industry segment which is related with one of the industry segments from the acquirer, the acquisition is considered to be related. Often related combinations are type 1 acquisitions having same or very similar NAICS codes, however also types 2, and 3 acquisitions are included with the survivor relatedness method that would not have been captured using other methods. An example of a survivor related segment combination is for instance combination 212311 Dimension Stone Mining and Quarrying with 327331 Concrete Block and Brick Manufacturing. These segments according to the NAICS code don’t suggest relatedness, but they are obviously vertically related as stones are used to manufacture bricks. Survivor relatedness reflects this relatedness as a common choice for segment combinations in the relatedness database. An example of an unrelated combination is for instance 212231 Lead Ore and Zinc Mining with 611430 Professional and Management Development Training; this combination does not occur in the relatedness database making it unrelated.
The advantage of the survivor based measuring for relatedness lies in the practicality of it. Instead of needing different methods for horizontal and vertical relatedness, this method does both at the same time. It assumes that any acquisition advantage coming from relatedness is recognized by firms leading to more combinations of advantageous segment combinations and less advantageous combinations being less frequently combined. Discussions about how to measure horizontal/vertical relatedness are thus avoided.
Especially vertical relatedness can be hard to measure as this concept can involve several different types of firm similarities; for instance similar production processes, shared customers/suppliers, similar research and development, shared sales locations. Making it very hard to fully capture all different aspects of vertical relatedness. On the other hand using common segment combinations as an exit point may lead to relatedness being implied were in fact there is none except for the acquirer segment group having excess cash and the target segment group being either an interesting investment object or even just being fashionable
Having said this it is more likely that survivor based relatedness would imply relatedness where in fact there is no relatedness. The opposite holds for other methods where it is much more likely to stamp a segment combination as being unrelated, while in fact relatedness is the case.
This study uses 2011/2012 data from US based acquirers, acquiring US based target firms. As there are many different time frames as well as regions used in literature this also provides an opportunity to see if similar results can be found.
Besides measuring performance of related and unrelated acquisitions, for comparison purposes with the SIC/NAICS method of relatedness this study also uses a subsample from the related group that holds type 1/horizontally related acquisitions only. Also from the unrelated group there’s a subsample used to compare with SIC/NAICS based relatedness that has different first digit NAICS codes between the acquirer and the target firm making those acquisitions both unrelated under the survivor related measure as well as the SIC/NAICS method. Industry bias is being avoided by using pairs of acquisitions for the related and unrelated groups where the acquirer has the same primary NAICS code; this holds for every sample or subsample used in this study.
To determine acquisition performance different methods are being used in literature; the market model is most commonly used and is also used in this study. Other methods include the mean adjusted return model and the market adjusted return model. The market model uses an estimation window prior to acquisition to compare the market performance represented by an index with the stock performance of the acquirer. As a measure for market performance prices from the AMEX Composite index (XAX) were taken as this index is considered to hold a fair representation of all industries. The estimation window is used to calculate normal performance for the acquirer which is compared to the actual performance in the event window after the acquisition announcement. The difference is called Cumulative Abnormal Return (CAR); representing individual firm performance. CAR values are then combined for different sample groups using an average; indicated by Cumulative Average Abnormal Returns (CAAR). In other literature often CAR is used both for individual firm values as well as for groups of firms.
2 Literature
The literature examined in this study has focused on very different periods between 1971 and 2007, ranging from a 3 year interval to a 12 year period. Sample size varied from 72 up to 2014 observations taken from either US acquisitions, or worldwide. All studies used the market model, but the definition of diversification and the methods for establishing
relatedness differ, making results difficult to compare. Within the 4 studies that have been closely looked into (Singh and Montgomery 1987, Kaplan and Weisbach 1992, Lien and Klein 2006, Aggarwal and Baxamusa 2013) the following methods have been used to measure relatedness: SIC measurement, survivor based measurement, a text analysis method for horizontal relatedness, and industry input-output tables for vertical relatedness.
Conclusions for acquirer CAAR vary greatly ranging from significant higher CAAR for related acquisitions (Lien and Klein, 2006), insignificant higher CAAR for related acquisitions (Kaplan and Weisbach, 1992), to insignificant higher CAAR for unrelated acquisitions (Singh and Montgomery, 1987) and significant higher CAAR for unrelated acquisitions (Aggarwal and Baxamusa, 2013).
According to Singh and Montgomery (1987) excess resources are the main reason why
firms are looking for ways to expand their current business or diversify. Growth without acquisition is often considered to be highly demanding in terms of investments and business development. Also entry barriers could be another reason to opt for acquisition, as well as acquiring usually being the fastest way to get started with new products or in new markets. On the other hand acquisitions may be costly as a premium may have to be paid.
Related acquisitions can add value because of economies of scale/scope and market power. Unrelated acquisitions can be used to move away from core business with limited growth potential. Also a greater effort would be needed to realize gains for related acquisitions. Unrelated ones can be run as stand-alone businesses; related ones cannot, they need to be integrated in existing business units.
The market model is being used, the window used is -800, +400 days; relatedness is considered when there’s a match in either production technology, research,
products/markets. 77 Acquisitions with values above 100M$ were used between 1975 and 1980. There were both significant positive effects for the related target as well as the unrelated target; more positive for the related target.
While both were positive, unrelated CAAR was more positive for the acquirers in the total event window 0, +100 days but no significant effect was found because, as the authors concluded, either the acquisition only effected a part of the acquirer, or the
estimation/event window was distorted by other events, or potential gains were already paid to the target as a premium decreasing net present value.
Kaplan and Weisbach (1992) found a significant link between post-merger performance and
returns shortly after the announcement. Unrelated acquisitions were divested 4 times more often than related ones, but unrelated acquisitions were not found significantly less
successful. Similar as Singh and Montgomery (1987), targets have higher CAAR compared to acquirers.
To distinguish between relatedness and unrelatedness the 4 most important segments, based on sales volume, of both acquirer and target were compared based on a first 3 digit SIC code level, if at least 1 same first 3 digit code was found then the combination was considered related.
Data was taken from 271 transactions for listed companies between 1971 and 1982 with a minimum transaction size of 100M$. The market model was used in order to calculate abnormal returns using an estimation window of -300, -61 days and an event window of -5, +5 days. Significant difference in performance was found only while comparing the number of unsuccessful acquisitions between related and unrelated firms. Taking both unsuccessful and successful together, the difference was no longer found significant.
Although a count rate of successful and unsuccessful acquisitions provides some insight about the distribution within these groups, CAAR values also take the level of success into account and are thus considered to have more economic meaning. Although related
acquisitions performed better on average, this test states that the difference in performance was not significant; similar to Lien and Klein (2006).
Lien and Klein (2006) argue that previous literature usually found that relatedness does
create value, but is appropriated mostly to the target and not to the acquirer (Singh and Montgomery (1987)). Few have found a positive acquirer CAAR or positive association between relatedness and acquirer CAAR, and some have even found a negative association between relatedness and acquirer CAAR (Agrawal et al. 1992).
The main reason why previous literature does not usually support related acquisition value for acquirers is that relatedness should be measured differently. The authors propose a survivor based approach. Instead of using models describing horizontal/vertical relatedness, the survivor based approach assumes companies making choices in their best interest improving chances to survive and qualifies most common choices for industry segment combinations as being related.
They believe, in accordance with Barney (1988), that acquisitions will bring value when either synergies can be found, there is high information asymmetry, or there are more targets then acquirers. These factors are considered more likely in related acquisitions. Data is being used from US listed acquisitions between 1982 and 1985. The market model was used with a -30, +30 day window for 72 acquisitions. Acquirer CAAR was not significant with SIC relatedness, but was significant with survivor based relatedness.
Aggarwal and Baxamusa (2013) found that unrelated acquisitions perform better compared
to related acquisitions. Both measured in cumulative abnormal returns as well as in operating performance. While e.g. Lamont (1997) found that diversification creates an internal capital market within company divisions and makes the allocation of capital inefficient, the authors found internal capital markets can be beneficial in accordance with Campa and Kedia (2002), Graham, Lemmon, and Wolf (2002), Villalonga (2004), and Hund, Monk and Tice (2010).
When in unrelated acquisitions operational improvements can be made or information asymmetry is high, unrelated acquisitions adds value. Also unrelated targets are more often mature firms with more positive cash flows.
Data was taken from a large set of acquisitions between 1997 and 2007 from 2014 deals exceeding 1M$ between listed companies. The estimation window is set to -210, -20 days before the announcement date. A text analysis method is used in order to determine if firm segments are horizontally related. For vertical relatedness industry input-output tables were used comparing investment rates between different industries. For performance
measurement the market model was used.
In contrast to Schleifer and Vishny (1991) who concluded in a similar study that unrelated acquisitions destroy value, the authors concluded that unrelated acquisitions are value adding. For different event windows of 3, 5, and 14 days all CAAR values were significantly more positive for the unrelated acquirer.
In The Economist, Economies of Scale and Scope (2008) it is argued that related
acquisitions are relatively inefficient due to their massive size. Unrelated acquisitions are more likely to have the advantage of sharing centralised functions, cross-selling, and using the outputs of one business as the inputs of another.
In The Economist, The next big bet (2011) one of the main arguments in favour of unrelated acquisitions is that companies in fast moving markets have to keep diversifying in order to keep the high NPV’s from emerging products. Another pro is the avoidance of innovation risk. However emerging products are risky as they might be overtaken by other products or not become as successful as expected leaving the acquirer with overcapacity.
Other arguments about diversification include: corporate diversification would usually be
unsuccessful due to a lack of a good value adding strategy (Porter 1987), managements with better networks would perform better in acquisitions (Schonlau and Singh 2009), small strategic alliance partners, and non-IT partners would perform better (Lee and Lim 2006), Seth (1990) concluded that in unrelated acquisitions economies of scale/scope and market power play no role unlike with related acquisitions; however it is associated with the
coinsurance effect and it is shown that financial diversification plays no role in either unrelated/related acquisitions.
Concluding: In most literature there is more evidence of positive target CAAR then acquirer
CAAR due to the presence of multiple (potential) bidders increasing the premium being paid. Usually favouring related target CAAR, in line with the argument of more (prominent) economic value. Results on acquirer CAAR are mixed, some studies found an (insignificant) positive relation between relatedness and acquirer CAAR; but some also found a negative (insignificant) relation.
One of the most interesting points of discussion in literature is that on one hand the total
value creation in related deals is expected to be higher because of the more prominent economic advantages such as economies of scale/scope and increased market power. As such there is more to be profited from. Unrelated firms are unlikely for instance to combine their purchasing efforts in order to save both personnel expenses as well as achieve high quantity discounts.
On the other hand it’s not always easy to combine related firms. Cultural differences, data migration from 1 system to the other, and resistance to a new way of working are often higher between related firms. Unrelated target firms are often run as standalone businesses without integration problems. Unrelated firms can also bring new skills into the acquiring company and launch the firm into new upcoming markets with high potential gains. Lien and Klein (2006)’s method of survivor relatedness is expected to classify some of the more successful unrelated deals from other relatedness methods as related because for instance obvious choices for new useful skills from unrelated deals are commonly made and are thus considered related under the survivor based method. This would lead to related deals being more successful under the survivor based method compared to other methods.
3 Hypotheses and Method
Similar to Lien and Klein (2006) this study uses the survivor based principle for relatedness in order to set a base for industry segment combinations and then comparing that base with the acquisitions being financially completed the next year to determine whether relatedness is the case. Survivor Relatedness has been chosen because it should reflect logical acquisition choices counting on acquiring firms making choices in their own best interest.
The more a segment combination is chosen relative to the size of these segments, the more logical that combination is deemed to be. Hence a more positive outcome can be expected. Contrasting, if an acquisition is unrelated then the expected outcome would be less positive otherwise more companies would have engaged in similar acquisitions. This reasoning leads to the following hypotheses:
H1: Related acquisitions have positive acquirer CAAR.
H2: Related acquisitions have higher acquirer CAAR than unrelated acquisitions.
For examining relatedness the North American Industry Classification (NAICS) coding system is used to distinguish between industry segments. In accordance with Lien and Klein (2006) and Teece et al. 1994 below model was used to calculate survivor relatedness.
Where is the number of times that segments i and j are combined, being the mean or expected value of combinations between i and j if combinations were random and following a hypergeometric distribution, and being the standard deviation. The mean and standard deviation were calculated as followed:
With K being the number of diversified firms. To determine significant relatedness an Alpha of 5% was used, as survivor relatedness is following a standard normalized function this is equivalent with a positive survivor relatedness value of 1,65 or bigger. For 1256 industry segments each possible combination was formed resulting in 788140 unique combinations. Each possible industry segment combination was indicated as either related or unrelated forming the relatedness database.
Acquisition data from Thomson One was used stating (primary) NAICS codes for both acquirer and target. Segments had to be checked until one combination was found that was related according to the relatedness database, only if all combinations were checked and none of them was related then the acquisition was marked as unrelated. For the sample selection process the suitable acquisitions were aggregated per industry group and handpicked per pair of the same primary NAICS code to avoid industry bias. Industry groups were chosen in proportion to the suitable acquisitions industry group sizes to give a fair representation. All stock prices were obtained from Yahoo finance (Kahn et al, 2016). As a comparison to determine abnormal returns, prices from the AMEX Composite index (XAX) were taken as this index is considered to hold a fair representation of all industries. In order to measure acquisition performance the market model is being used comparing within the estimation window the stock performance to the performance of the XAX index. The market model follows the model
In which is the return of firm i at time t estimated by a linear equation
where is the intercept, the slope/fit between the firm and the market performance, and representing the abnormal return.
Unlike Lien and Klein (2006) this study does not use both Ebitda, target and acquirer performance as estimators for abnormal return levels, but focusses only on acquirer
In similar fashion to Lien and Klein (2006) the main testing window is between 30 days before the announcement day and 30 days after announcement. However the event window (from the announcement day onwards) is split further into 1, 5, and 10 days after announcement. Also 250 days after announcement is added to look for any long term effects.
Statistical methods (Müller 2016) are then used for hypotheses testing, for aggregation across time the following formulas were used for individual firm analysis:
With being the cumulative abnormal return for firm i, being the last day of the estimation window, being the last day of the event window, and being the abnormal return of firm i at time t. Significance was found by performing:
with , and as the number of returns in the estimation window. For aggregation across firms:
with being the cumulative average abnormal return across firms. Testing for significance was done with:
All tests were done for the total sample of 66 acquisitions as well as for the subsamples of 33 related and 33 unrelated acquisitions. Different alpha’s were chosen for analysis on individual firm level, CAR, changing between a more exploring approach with an alpha of 25% to the standard alpha of 5%, and finally testing for highly significant values with an alpha of 1%. For the aggregated CAAR level a subsample was used consisting of 19 acquisitions with the same primary NAICS code in order to check for differences within the related acquisition group. For the unrelated group a subsample was chosen with acquisitions where primary acquirer NAICS and primary target NAICS were from a different industry group: have a different first digit code, in order for this group to be considered unrelated under the principle of the classical SIC/NAICS measurement method for relatedness.
4 Data
The data for this study was taken from stock market listed and reported firm acquisitions from Thomson One, financially completed in 2012 where both the acquirer and target had their base in the United States of America. The choice not to take foreign acquisitions into account is based on the logic that every country has its own structure of relatedness what would mean that every listed company’s relatedness worldwide would have to be examined making the analysis too extensive and unfeasible.
Segment data per company was obtained from the Compustat database, stating all (primary) NAICS codes per firm. Starting with 5990 acquisitions financially completed in 2012, the data was then filtered, beginning with filtering on stock price availability starting 30 days before the time of the acquisition announcement until 250 days after announcement. 30 days before announcement was used to set a normal return window consistent with Lien and Klein (2006). To avoid industry bias as much as possible, acquiring firms being active in a primary segments
The reason for this being that these acquisitions are only unrelated acquisitions making it impossible to compare same primary segments between unrelated and related subsamples which would result in company bias.
Furthermore acquisitions with acquiring companies having letters in the primary segment code were excluded as the segments in the relatedness database do not contain letter codes. Announcement dates before 2011 were taken out to ensure that acquisition announcement dates are comparable with the relatedness timeframe. After the data reduction process 1444 acquisitions were qualified as suitable for the sample process.
The definition for unrelatedness being used is quite narrow: when one possible segment combination is related between acquirer and target, the acquisition is marked as related. This results in most acquisitions being related and thus making it harder to pick same primary segment pairs for the acquirer between related and unrelated acquisitions. In the end 66 acquisitions were picked as a sample, 33 related acquisitions and 33 unrelated acquisitions. This accounts for 5% of the cleaned data set.
Looking at the descriptive statistics on individual CAR values in Table 1 and the above graph: “Change in USD value per type of acquisition”, all mean values are positive for both subsamples, but higher for the unrelated group.
However, running single factor ANOVA’s to compare related and unrelated CAR values per event window produces no significant difference in mean. Standard deviation is larger for the unrelated group in the event windows until 10 days after announcement, making positive CAR values less significant, but smaller for the event window of 250 days. Skewness is for all event windows more positive for the unrelated group, with the related group being negative in all event windows except 0-1 days, this together with the standard deviation declares the larger number of individual firm CAR values for the related group while the mean CAR is actually lower. On an aggregated CAAR level however the unrelated group is more significant except for the 0-1 day event window, as shown in table 3A further down in the results section.
5 Results
Table 2B shows a count of significant individual firm CAR values under the t-statistic with Alpha set to 5% for CAR as described in the method section. Both positive as well as negative significant values were found, except for the 250 day event window the pattern seems to be that more positive significant values were found in the related group compared to the
The same pattern from Table 2B but less pronounced can be seen in Table 2A testing with an Alpha of 25%, this table is included in the appendix: tables and figures section at the end of this study.
Table 3A results show that all t-stat values for CAAR are positive, but more positive for the unrelated group except for the 0-1 day event window, mainly caused by the 3 times larger standard deviation. As for the significance of CAAR values is concerned, for the related group 1 out of 5 event windows is significant (T: 0-1) at the standard 5% level. The average p-Value for the related event windows is 0,256. For the unrelated group also just one window (T: 0-250) is significant, but the average p-Value of 0,101 for these values lays a lot closer to the 5% significance level.
Previously mentioned results in the context of the hypotheses imply that support was found for hypothesis 1 (H1: Related acquisitions have positive acquirer CAAR). For hypothesis 2 (H2: Related acquisitions have higher acquirer CAAR than unrelated acquisitions), no support was found. Although results point into the opposite direction, this was not significant for all event windows.
Overall looking at p-Values for CAAR for the whole sample group the average p-Value is significant with 0,048; also each event window is significant by itself. CAAR values and standard deviation lie in between those of the related and unrelated groups. The difference in p-Value between the whole sample and the subsamples is mainly caused by scaling up from 33 to 66 causing the p-Value to move towards significance by a factor sqrt(N).
The subsample with same primary NAICS code related acquisitions shown in Table 3B shows remarkable results compared to the complete related group. Average CAAR for all event windows is 0,118 compared to 0,079. Average p-Value is 0,062 compared to 0,256; when the T: 0-250 event window is taken out of the average p-Value that would improve significance for the same NAICS acquisition group further to 0,031 making it significant.
Table 3C shows CAAR statistics from a selection of different first digit NAICS codes from the unrelated group. The CAAR values are remarkably higher than those from the complete unrelated group. Unfortunately for both Table 3B and 3C holds that the sample size has become too small to draw significant inferences from, however had the sample size been 66 like the whole sample and CAAR values and standard deviation would be the same, then all P-Values for every event window would be significant.
6 Discussion and Conclusion
This study shows support for both related as well as unrelated acquisition success. Unrelated acquisitions having even higher CAAR values and coming closer to significance at the standard 5% level. The method used in this study is similar to the method used by Lien and Klein (2006), however unlike in their study, results do not support the association between Survivor based relatedness and acquirer CAAR. Results are more comparable with Singh and Montgomery (1987) and Aggarwal and Baxamusa (2013) although their methods to measure relatedness are different.
As suggested mainly by Singh and Montgomery (1987), one possible reason for unrelated acquirer CAAR to be higher is that even though the total expected value creation for related deals is higher due to more apparent economic advantages, these advantages result in higher premiums to be paid to the target firms compared to the premiums paid in unrelated deals.
Economic value creation for unrelated deals is less obvious, and might only become visible after the acquisition has unfolded, possibly explaining an increase in CAAR for the unrelated acquirer.
As discussed in the literature section there might also be downsides to related deals.
Cultural differences, data migration from 1 system to the other, and resistance to a new way of working to name a few are often higher between related firms. Unrelated target firms on the other hand are often run as standalone businesses not hindered by these integration issues, and still benefitting from skills being brought into the acquiring firm.
One of the limitations of this study lies in how survivor relatedness is being measured. 5% significance is chosen to suffice for any combination of segments between acquirer and target making the acquisition related. However there’s no ranking system applied between different segments for how relevant they are to the firm. Also separate relatedness values were not added to 1 final value. Future studies could explore a weighted measuring system for all segment combinations added to 1 final relatedness value based on sales, net profit, or other grounds in order to make survivor relatedness a more reliable measure.
Another limitation lies in the size effect of the deals. Even though only listed acquiring firms were taken and thus excluding relatively small firms, they may still vary greatly in size. As for the targets they were not necessarily listed making them possibly even more different. Then again Lien and Klein (2006) had found no size effect in their similar study. Some studies suggest that effects should be corrected for differences in size proportions between acquirer and target by multiplying/dividing to arrive at the same size proportion. But then again some relative small targets that have a specific skill might be able to share this skill with the complete acquiring firm having a huge impact whereas the value of other skills acquired might be limited to the target alone. Straightforward dividing/multiplying between different acquirer/target size proportions therefor does not seem to make sense. More research should be done on the effects of different acquirer/target size proportions to make for a better comparison.
Future studies could follow up on the results in this study as shown in tables 3b and 3c by dividing related acquisitions into even more subsamples as there seem to be a different CAAR levels for different types of relatedness. Also using larger samples would further improve significance.
A final question that arises is how to deal with different time frames of measurement between the relatedness database and the acquisition database. In this study the relatedness data was taken from 2011 whereas the acquisitions data was taken from acquisitions financially finished in 2012 with an announcement date varying between the start of 2011 and the end of 2012. However relatedness will vary from year to year, making the comparison between different years unreliable. Future studies could explore better methods for more accurate comparison between relatedness data and acquisition data.
7 References
Aggarwal, Rajesh K. and Baxamusa, Mufaddal H., Unrelated Acquisitions (June 28, 2013). AFA 2013 San Diego Meetings Paper.
Agrawal, A., Jaffe, J. F., & Mandelker, G. N. (1992). The post-merger performance of acquiring firms – a reexamination of an anomaly. Journal of Finance, 47(4), 1605–1621. Barney, J. B. (1988). Returns to bidding firms in mergers and acquisitions – reconsidering the relatedness hypothesis, Strategic Management Journal, 9, 71–78.
Bruton G.D., Oviatt B.M., White M.A. (Aug 1994). Performance of Acquisitions of Distressed Firms, Academy of Management, vol. 37 no. 4, 972-989.
Campa, J. and S. Kedia, 2002, Explaining the Diversification Discount, Journal of Finance, 57: 1731-62.
Clark P. and Mills R. (2013), Masterminding the Deal, 3rd August 2013.
Fan, J., and L. Lang, 2000. The Measurement of Relatedness: An Application to Corporate Diversification, Journal of Business, 73: 629-660
Graham, J., M. Lemmon, and J. Wolf, 2002, Does Corporate Diversification Destroy Value? Journal of Finance, 57: 695-720.
Hoberg, G. and G. Phillips, 2010, Product Market Synergies and Competition in Mergers and Acquisitions: A Text-Based Analysis, Review of Financial Studies, 23: 3773-3811.
Hubbard, R. G., & Palia, D. (1999). A reexamination of the conglomerate merger wave in the 1960s: An internal capital markets view. Journal of Finance, 54(3), 1131–1152.
Hund, J., D. Monk, and S. Tice, 2010, Uncertainty about Average Profitability and Diversification Discount, Journal of Financial Economics, 96: 463-484.
Kahn S., stock price downloader, http://investexcel.net/author/admin/, (2016) Kaplan, S. N., & Weisbach, M. S. (1992). The success of acquisitions – evidence from divestitures, Journal of Finance, 47(1), 107–138.
King D.R., Daltan D.R., Daily C.M., Covin J.G. (2004), Meta-analyses of post-acquisition performance: indications of unidentified moderators, Strategic Management Journal, Volume 25, Issue 2, pages 187–200.
Lee S.Y., Lim K.S. (Sep 2006), The impact of M&A and joint ventures on the value of IT and non-IT firms, Review of Quantitative Finance and Accounting, September 2006, Volume 27, Issue 2, pp 111-123.
Lien Lasse B. and Peter G. Klein (2006), RELATEDNESS AND ACQUIRER PERFORMANCE. Elsevier Ltd. Volume 5, 9–23.
Moeller S.B., Schlingmann F.P., Stulz R.M. (Aug 2004), Firm size and the gains from
acquisitions, Journal of Financial Economics, Volume 73, Issue 2, August 2004, Pages 201– 228.
Montgomery C.A. and Wilson V.A. (Jan.1986), Strategic Management Journal Vol. 7, No. 1, pp. 91-96
Porter M.E. (May 1987), From Competitive Advantage to Corporate Strategy, Harvard Business Review may 1987.
Prahalad, C. K., & Hamel, G. (1990). The core competence of the corporation. Harvard Business Review, 68(3), 79–91.
Roll R. (Apr 1986), The Hubris Hypothesis of Corporate Takeovers, The Journal of Business, Vol. 59, No. 2.
Schonlau R, Singh P. (2009) Board networks and merger performance, Carnegie Mellon University working paper.
Schumpeter (Oct 2013), The Economist, Riding The Wave.
Seth A. (Oct 1990), Sources of value creation in acquisitions: An empirical investigation, Strategic Management Journal, Volume 11, Issue 6, pages 431–446.
Shin, H., and R. Stulz, 1998, Are Internal Capital Markets Efficient? Quarterly Journal of Economics, 113: 531-552.
Shleifer, A., and R. Vishny, 1991, Takeovers in the ‘60’s and the ‘80’s: Evidence and Implications, Strategic Management Journal, 12: 51-9.
Müller S., statistical methods (www.eventstudytools.com/significance-tests) (2016). Singh H. and Montgomery C.A. (July 1987), Corporate acquisition strategies and economic performance, Strategic Management Journal, Volume 8, Issue 4, pages 377–386.
Shleifer, A., and R. Vishny, 1991, Takeovers in the ‘60’s and the ‘80’s: Evidence and Implications, Strategic Management Journal, 12: 51-9.
Teece, D. J., Rumelt, R., Dosi, G., & Winter, S. (1994). Understanding corporate coherence – theory and evidence. Journal of Economic Behavior & Organization, 23(1), 1–30.
The Economist (Oct 2008); Economies of scale and scope, online edition. The Economist (Sep 2011), The Next Big Bet, online edition.
Villalonga, B., 2004, Diversification Discount or Premium? New Evidence from the Business Information Tracking Series, Journal of Finance, 59: 479-506.
