Acquisitions made during “times of crisis”; what are
the long term implications on acquirers’ profitability.
Event study on American blue chips during the recent
financial crisis (2009 – 2012).
Amsterdam Business School
Name Nils Nienaber
Student number 10630457
Program Economics & Business Specialization Finance & Organization Number of ECTS 12
Supervisor Ilko Naaborg
Completion Monday, February 1, 2016
Abstract
This paper investigates the long term effects (up to 4 years) of acquisitions performed during a severe “crisis period” on company performance, using a sample of about 100 major US American companies from January 2009 to December 2012. Compared to the pre crisis period takeovers have been done at significant lower transaction values during the financial crisis. The impact of acquisitions on companies’ return on assets ratios as well as on cumulative stock returns is investigated in this paper. Our results indicate that acquisitions made during the financial crisis have a positive effect on companies’ return on asset ratios. However, in terms of accumulative stock returns our results do not show any statistical significant impact of acquisitions on stock performance.
Keywords:
Table of Contents
1. Introduction ... 3
2. Literature review ... 5
2.1. ROA and CR literature ... 5
2.2 Short and medium term returns ... 6
2.3 Long term returns ... 7
2.4 Summary and alternative ROA approach ... 8
3. Methodology and Data ... 9
3.1. Methodology ... 9
3.2. Data and descriptive statistics ... 11
4. Empirical result ... 14
4.1. Empirical Results ... 14
4.2 Robustness check, using a CR model ... 16
5. Conclusion and discussion ... 19
1. Introduction
An acquisition is a corporate action of one company buying most or all of a target company’s shares in order to gain control over it. Acquirers can pay the
shareholders of the target firm in different ways, e.g. in cash, in acquirer’s stocks or in a combination of both. There are various reasons for acquisitions. Firstly, acquirers might aim to gain a competitive advantage or increase their market share by acquiring a company in a market where the purchaser is not yet active, or by making use of the target’s distributions or marketing network. Another reason to engage in an acquisition is the acquirer’s wish to diversify its product portfolio in order to do risk diversification. A company may also try to achieve economies of scale and scope to obtain a competitive advantage. This means that companies are able to decrease average production costs. Finally, there are also reasons for a target firm to be taken over. If a company finds itself in financial stress, the board may recommend its current investors to tender their shares for the company to survive under new ownership.
With the bankruptcy of Lehman brothers in September 2008 the financial crisis had fully evolved and hit the global economy. The credit crunch, as well as the prevailing uncertainty surrounding the situation, had an impact on Mergers and Acquisitions (M&A) activity. Global M&A activity had seen a decline in 2008 compared to its record year in 2007, and an even sharper decline in 2009. According to Reddy et al., M&A activity in the United States of America had decreased along with the global trend (2014).
The acquisition of Motorola Mobility by Google (today Alphabet) has certainly been one of the biggest acquisitions on the American market within the wake of the financial crisis. In January 2011, Motorola split into two separate entities; Motorola Solutions, focusing on the company’s radios and client specific
commercial needs, while Motorola Mobility concentrated on Motorola’s mobile handset division (mobile/smart phones). In August 2011, Google publicly announced the acquisition of Motorola Mobility segment for 12.5 billion Dollars
in order to obtain its portfolio of patents. By obtaining the patents, Google intended to increase its product portfolio as well as improve its Android
operating system to further establish its position in the market. However, there has also been public criticism about the acquisition. For example, the premium Google paid was viewed by some as too high considering Motorola’s lack of profitability. In regards of acquisitions it is very interesting to find out whether the purchases actually increase the profitability of the acquirers like Google, and therefore justify the high costs of the takeover. Contrary to that, these
acquisitions could have occurred partially due to managerial hubris leading to the harm of the shareholder’s interests.
In our study, we want to analyze whether acquisition-decisions made by American blue chip companies in the financial crisis were in the long term interest of its shareholders. In order to test this, we investigate about 100
acquisitions and the effect they had on the acquirer’s return on assets (ROA) and their cumulated stock return (CR).
The given sample consists of the constituents of the S&P 100 index between January 2009 and December 2012, which are described as the years of the financial crisis. Due to missing data, some companies are removed resulting in our sample consisting of 106 companies in total. Within the given time period, 67 of the 106 companies acquired at least one other company. In order to
investigate the significance of the dummy variable acquirer on ROA, we use a model developed by Andrew P. Dickerson, Heather D.Gibson and Euclid Tsakalotos. Their theory was published in the Oxford Economic Paper (1997). The model predicts the ROA by using independent acquirer variables like past ROA, company size, leverage and net asset growth. In the robustness check, we slightly change the model, by replacing ROA and past ROA by CR and past CR respectively. In our opinion it is very interesting to see whether the model can also predict actual stock returns, which would be of greater value for investors than predicted ROA ratios.
To our knowledge there has been no similar study performed that researches the effect of acquisitions on ROAs and CRs within the given time period after the breakdown of Lehman brothers and the full evolvement of the financial crisis. Furthermore, there is no knowledge of a study performed using accumulated stock returns for the given test period as well.
The overall conclusions of the first inquiry are not aligned with the results of Dickerson et al., and other research about the effect of acquisitions on
profitability. We find significant support that acquisitions had a positive effect on acquirers’ ROAs during the financial crisis between 2009 and 2012, whereas Dickerson et al. found a significant negative effect of acquisitions on ROA in the UK between 1948 and 1977.
In regards to our robustness check, there is a vast array of research being done on post-acquisition CRs (stock return) during different time windows that conclude with very ambiguous results. We are therefore not surprised by the non-significance of our test results.
Following this introduction, we are going to review the current literature and explain the two current strands of research on our research question. In part three we are going to present and explain our sample and sample data. Furthermore, we will investigate the possible relationships within the data sample and present our ROA based model. In the analysis part, we will present and describe our empirical results. Following this, we will apply a CR based robustness check in order to check the applicability of the model. To conclude our study we will discuss our results as well as the limits of our research.
2. Literature review 2.1 ROA and CR literature
According to Dickerson, Gibson and Tsakalotos, there are two main strands of measuring the success and effect of acquisitions on acquirer performance (1997). Firstly, there is the more common strand of the stock price approach, which measures the success of acquisitions in terms of cumulative abnormal stock
returns (CR). Secondly, there is the more direct approach applied by Dickerson, Gibson and Tsakalotos, who measure the success of acquisitions by evaluating the companies’ return on assets (ROA). Apart from Dickerson, Gibson, and Tsakalotos, all authors in this review have applied the CR based model. As stated by Conn and Cosh, there have been a great number of event studies, which investigate the effects of mergers and acquisitions on shareholder’s short term as well as long term stock returns (2001). According to the authors, stock data for the U.S and the U.K are “readily available”, that is why most research done on this topic concerns either one of these two markets (2001).
2.2 Short and medium term returns
We are aware that our research focuses on the more medium to long term effects of acquisitions on company performance. However, as we do not know how long it takes until the effects of an acquisition are fully reflected in the companies’ ROAs or CRs we decided to also review literature about short term returns. In terms of short term U.S market returns Jensen and Ruback stated in their CR based research that returns around announcement and completion date are either zero or negative for acquiring firms (1988). Conn and Cosh analyzed a sample of 2195 U.K. domestic as well as 1065 U.K. cross boarder acquisitions between November 1984 and July 2000, using a similar model (2001). The authors confirmed Ruback’s and Jensen’s findings, and argued that there is a “conventional wisdom within financial economics” that short-term abnormal return for acquirers are negative. However, Conn and Cosh also compared several event studies with according to them similar methodologies and concluded that out of nine conducted studies on U.S acquirer post-acquisition performance two argue that short term returns are indeed positive. The authors conclude that, because of these two “outlier studies” there seem to be ambiguous effects, which they were not able to explain. Next to the U.S and the U.K there has been research done on the case of Japan. According to Charlie and Conn, there have been three studies, which all support positive short term returns on acquirer stocks in Japan (2001). However, the reason why this is the case, according to the authors, were “bidder specific firm characteristics” and that
returns were observed in US Dollar, while the YEN appreciated, which was not controlled for in the analysis (2001).
Conn and Connell, as well as Aw and Chatterjee conducted research using the CR model on monthly returns from the period of acquisition up to one year after deal completion (1990; 2004) for U.K companies. Both teams found significant negative cumulative abnormal returns.
2.3 Long term returns
According to Charlie and Conn, there have been two studies about the long-term returns of acquirers in the U.S market (2001). Loughran and Vijh applied a CR model for their research and found negative returns for their entire sample of 947 U.S domestic acquisitions between 1970 and 1989; however their results are not statistically significant (1997). Rau and Vermaehlen analyzed a sample of 2997 U.S acquisitions between 1980 and 1991, using a CR model as well, but could also not provide general statistical significant results either (1998). However, that does not mean that their research has not been successful. Both studies found significant results that acquirers’ long term returns depend on whether deals were done in form of mergers or tender offers. In contrast to these two teams, Black et al. found significant negative returns for U.S acquirers
engaging in cross-border deals (2007). The authors analyzed the returns of 361 acquirers in the time from 1985 to 1995. Charlie and Conn assess Black et al.’s paper as the “most extensive analysis up to date” on cross border acquisitions. The authors appreciate the fact that Black et al. included a benchmark group of domestic takeovers and the inclusion of not just private, but also public
companies (2007).
There are several studies about the long term acquisition effects on acquirer’s post-acquisition stock performance on the U.K market as well. In their CR based study on long term effects Alan Gregory and Steve McCorristion consider 343 acquisitions within a time span of nine years (1985-1994).The authors found statistically significant results, supporting the notion that an acquisition has a negative effect on long term acquirer’s returns (1997). These results are supported by Higson and Elliot, who analyzed 830 U.K mergers between 1975
and 1990, using a similar CR methodology as Gregory and McCorriston (1998). However, Cosh and Guest found post acquisition returns which were not different from zero for their sample from 1985 to 1996 (2001). According to Charlie’s and Conn’s study, acquisitions have a significant negative effect “on long term share returns”. The negative abnormal returns persist up to two years following the acquisition for domestic takeovers. However, in terms of cross-border takeovers the authors did not find any significant result.
2.4 Summary and alternative ROA approach
All in all, it can be said that in terms of short term returns the majority of CR based studies show significant negative effects of acquisitions on stock performance. Therefore, we agree with Conn’s and Gosh’s statement of a “conventional wisdom” regarding negative short term returns in the industry. However, regarding long term returns, the picture is somewhat more ambiguous, on the one hand there is some research supporting the notion of negative long-term stock returns (Black et. al, Gregory and McCorriston, Higson and Elliot), on the other hand there is also research suggesting that the effect of acquisitions on returns is insignificant (Loughran and Vijh, Rau and Vermaehlen, Cosh and Guest). Thus, in terms of long term effects, there is a prevailing disagreement about the existence of a negative effect on stock returns or no effect at all. Next to the CR based model, there is also a second strand of research, which advocates measuring the success of acquisitions differently. The approach of using cumulative stock returns (CRs) has been criticized by Schiller (1989). As reported by the author, the stock price at the time of takeover could not only represent a fair valuation of the companies’ value, but also prevailing over- or undervaluation due to various reasons like e.g. imminent takeover rumors (1989). Moreover, Scherer argued, that these “random valuation errors” could basically persist at any point in time and that a rising acquirer’s stock price could simply reflect a market correction of the target’s stock price instead of a gain in efficiency (1988). Hence, Scherer supports the notion of measuring the success of an acquisition by evaluating companies’ profitability directly, instead of stock returns. Due to this problem, we choose to follow the second strand of research,
which focuses on company profitability (measured by ROAs). Subsequently, we also incorporate the first strand of research by changing our model to a CR based stock price model, in order to check our model’s robustness.
Using the profitability ROA approach Dickerson et al. report negative effects on acquirer’s company profitability (1997).
To our knowledge there has been no research published yet about the effect of acquisitions on acquirer’s profitability during the financial crisis from 2009 to 2012. All the studies mentioned above focus on different time periods. By
focusing solely on the recent financial crisis our research is no exception in these regards. However, since the above researchers did not limit their models to specific parts in the business circle we expect to see either negative or insignificant effects of acquisitions on ROAs and CRs in our tests.
3. Methodology and Data 3.1. Methodology
We want to analyze the effect of acquisitions made by American companies during the financial crisis on their profits. Therefore, we gather data of all companies which have been constituents of the S&P 100 at any point in time between 01.01.2009 and 31.12.2012. The S&P 100 constituents are chosen in order to guarantee an unbiased sample of leading American blue chip companies. According to the official S&P 500 factsheet (December 2015), released by S&P Dow Jones Indices LLC, the S&P 500 covers about 80% of the total American equity market with a market capitalization of 18.774 billion US Dollars. According to the S&P 100 factsheet, all constituents of the S&P 100 are selected in the light of sector balance and represent equity worth about 11.836 billion US Dollar (December 2015). Hence, we infer that the S&P 100 represents about 50% of the total American equity market.
We compare firms, which did at least one acquisition in the given time period with firms which didn’t acquire any other company in the test period.
We retrieved our data from the WRDS campustat database. According to the database, there are 124 companies, which have been constituents of the index between 2009 and 2012. A list of all acquisitions and therefore all acquirers has been retrieved from the
Zephyr database. Some of the entries of the acquisitions list had to be deleted, due to not disclosed deal figurers or not completed deal statuses.
As a result of double entries and missing data we are able to collect data for 106 companies, of which 39 companies have not acquired any other company within the defined time period, whereas the remaining 67 have at least acquired one other company in the test period.
The basic model proposed by Dickerson et al., which is used in this study, is the following:
Dickerson, Gibson and Tsakalotos included the latter variables in their model (1997):
is the average profitability of company i of the time period 2009 to 2012, measured by rate return on assets (ROA). The ROA is computed by dividing the net income over total assets. The first variable and its coefficient capture the degree of persistence of profits. is the average ROA of the years from 2005 to 2008. Size of the firm is measured as the natural logarithm of total assets at the end of the test period (2012) in order to compare relative sizes. Dickerson et al. used a five-stage ranking system with five dummy variables to capture size. However, since our sample size is “only” 106, we would have obtained relatively small numbers of companies in each category using a category approach. Leverage is represented by the debt over net asset ratio at the end of the test period (Year 2012). The Net Asset Growth variable is measured by the growth of net assets over the 4 year time period. Dickerson et al. constructed the growth variable slightly differently by adding a “lag θ(L) to allow for delayed effects of company growth”. However, we show in the analysis part that the effect of lagged growth in Dickerson et al.’s results is very close to zero. Therefore, we do not include lagged growth into our model. The fourth variable Acquirer is a dummy variable which can either take on the value of zero or one. Hence, in case the company did at least one acquisition in the time between 01.01.2009 and 31.12.2012 Acquirer value is going to be one. Otherwise the Acquirer value is going to be zero. The variable is the constant. According to Dickerson et al.
the chosen variables are “standard and command wide support in the industrial organization literature”.
There are some issues which are not taken into account in the model. Firstly as stated by Dickerson et al, the model does not consider “unobservable differences between companies”, like “managerial quality and/or industry-specific influences”. Secondly, the variables Leverage and Net Asset Growth might be endogenous. Concerning this endogeneity, the authors stress, that current positive profitability figures signal a healthy state of the company to capital markets, which could then be reflected by increased loaning (Leverage), which can eventually lead to increased net asset growth through higher earnings (1997).
In terms of causal relationships, we assume that the decision to acquire another company has a significant effect on the acquirers’ profitability after the deal. This common assumption is tested by a vast number of studies, of which we discussed the most relevant ones in the literature review.
3.2. Data and descriptive statistics
In order to get a full list of all companies which had been part of the S&P 100 in the years of the test period (2009 – 2012) we used the campustat database in WRDS. We obtained the data for the variable Acquirer (dummy) from the Zephyr database, which contains information on mergers and acquisition. The only relevant part for our research was thereby to find out, which companies have been active on the acquisition market. How active they were and what sort of transactions (i.e. horizontal or vertical) they conducted does not play a role for this research. The relevant stock data like net income, total assets and total debt has been retrieved from the campustat database as well. Table 1 represents a summary of important information about the variables data, like observation numbers, means, standard deviations, and minimums as well as maximums.
Table 1
Summary statistics for the independent and all dependent variables
ROA Average 09-12 is the independent variable whereas ROA Average 05-08, Ln (Total Assets), Leverage, Net Asset growth and Acquirer (dummy) are the independent variables. The regression uses average annual data of the years 01/2005 to 12/2008 and 01/2009 to 12/2012 respectively for the two ROA variables. The variable Ln (Total Assets) is reported as the logarithm of total assets (12/2012). Leverage is the debt over net assets ratio (12/2012). Net Asset Growth is the growth of net assets over the test period from 01/2009 to 12/2012. The variable Acquirer (dummy) is a binary variable, it is equal to one, if the company has acquired at least one other company between 01/2008 and 12/2012. Next to the number of observations the table reports the variables means, its standard deviations, as well as the variables smallest and highest observation.
Variable Obs Mean Std. Dev. Min Max
ROA Average 09-12 106 0.0650 0.0509 -0.0634 0.2049 ROA Average 05-08 106 0.0743 0.0583 -0.1196 0.2092 Ln (Total Assets) 106 11.0624 1.3612 6.6513 14.6739
Leverage 106 0.3354 0.3561 -2.0500 1.5264
Net Asset Growth 106 -0.1216 0.5605 -0.8560 4.0744 Acquirer (dummy) 106 0.6321 0.4845 0.0000 1.0000
Regarding the data it is worth mentioning that the ROA (return on total assets) has been about 1% higher per year in the period 2005-2008. However, the standard deviation is also slightly higher than in the period from 2009-2012. Leverage varies between 0 and 1.5264, which means that there is one company in the sample which has no long-term liabilities and one company which has long term liabilities 1.5264 times as high as their equity. However, there is one
exception; Anadarko Petroleum Corp was practically insolvent in the end of 2012. Anadarko’s equity value was -6474 million, which caused the negative leverage figure of -2.0500 in our leverage computations. On average, net assets dropped by 12.1577% within the test period from 2009 to 2012. The average for the Acquirer dummy has been 0.6321, which means that about 63% of the sample companies have not realized any acquisition.
All variables, except of the dummy variable Acquirer are normally distributed. The dummy variable can only take on two different values and the data can therefore not be normally distributed. Hence, a transformation of the data is not necessary. The outcome of the Shapiro-Wilk test is shown in the following table 2.
Table 2
Shapiro-Wilk test for normality for all variables except of the Acquirer (dummy)
The table reports the variable’s Shapiro-Wilk test value W, its z-value and in the last column the variable’s p-value. Variable W z Prob>z ROA Average 09-12 0.9625 2.625 0.0043 ROA Average 05-08 0.9752 1.705 0.0441 Ln (Total Assets) 0.9665 2.374 0.0088 Leverage 0.7537 6.811 0.0000
Net Asset Growth 0.6139 7.813 0.0000 Acquirer (dummy) 0.9943 -1.561 0.9407
All p-values are below five percent, which means that the probability of the variables being normally distributed is higher than 95 percent. Thus, we can infer that the sample data is normally distributed.
Table 3 shows the correlations of variables with each other:
Table 3
Correlations table of all variables
The table displays the correlation between the different variables. The correlations of 1 simply show that the variables correlation with itself is 1, which must be the case.
ROA 05-08 Ln (Tot As) Leverage Net As G Aqc Dummy ROA Average 05-08 1.0000
ROA Average 09-12 0.7311 1.0000
Ln (Total Assets) -0.3342 -0.3290 1.0000
Leverage 0.0262 -0.1592 0.0458 1.0000
Net Asset Growth -0.0699 -0.3013 -0.2006 0.1812 1.0000 Acquirer (dummy) 0.2100 0.1999 0.3951 -0.1038 -0.2578 The correlations between our variables are relatively low. Only the ROAs of 2005 to 2008 correlate quite strongly with the ROAs of 2009 to 2012. ROAs are
apparently relatively consisting (0.7311), which means that there is a trend that if past returns decrease\increase, then current returns decrease\increase as well.
However, Dickerson et al. do not discuss this correlation in their paper, so they either didn’t encounter this problem of relatively high correlation, or they didn’t see it as a potential problem. Furthermore, multicollinearity is considered to be a correlation between different independent variables, not a correlation between the dependant variable with one independent variable. Therefore, we can infer that there is no multicollinearity between the independent variables.
4. Analysis
4.1. Empirical Results
It is important to note that this regression is done in order to investigate what effects the different independent variables have on American blue-chip
companies’ ROAs in the time of a financial crisis, like the one from 2009 to 2012. Other than in the paper of Dickerson, Gibson and Tsakalotos (1997), our results should not be generalized to non American blue chip companies nor should they be used to forecast ROAs in non-crisis times.
Table 4
Regression of average yearly ROA from 2009 - 2012 on explanatory variables (Basic model of Dickerson et al. applied.)
The table describes what effect different variables have on a company’s average ROA in the time between 01/2009 and 12/2012. Next to the variables’ coefficients β the table also reports standard errors of the variables, its t-test values, and more importantly its p-values. Acquirer (dummy) represents the only binary variable in this regression. A constant is included into the regression and is listed in the very last row. The 95% confidence interval is also included. The regression uses average annual data of the years 01/2005 to 12/2008 and 01/2009 to 12/2012, respectively for the two ROA variables. The recorded values for Ln (Total Assets) and Leverage are from the 31.12.2012. Net Asset growth is the computed growth over the period from 01/2009 to 12/2012. The Acquirer (dummy) variable is equal to one, if the firm has conducted at least one acquisition between 01/2009 and 12/2012.
ROA Average 09-12 Coef. Std. Err. t P>t 95% Conf. Interval ROA Average 05-08 0.6189 0.0677 9.14 0.000 0.4846 0.7533 Ln (Total Assets) -0.0056 0.0030 -1.87 0.064 -0.0116 0.0003 Leverage 0.0200 0.0093 2.16 0.034 0.0016 0.0384 Net Asset Growth 0.0121 0.0065 1.88 0.063 -0.0007 0.0250 Acquirer (dummy) 0.0186 0.0079 2.36 0.020 0.0030 0.0341 Constant 0.0643 0.0338 1.90 0.060 -0.0028 0.1314
Our results indicate that acquisitions have a positive effect on profitability measured by return on assets (ROAs). The coefficient for the acquirer dummy is 0.0186, which means that companies which have acquired at least one company in the test period have on average a 1.86 percentage point higher ROA then other “non-acquirer” companies. Dickerson’s et al.’s coefficient however is -0.0138, which indicates the opposite. Thus, according to our outcomes, acquisitions had a positive effect on ROAs of American blue chip companies during the financial crisis.
Past ROAs are very persisting over the tested period. Our coefficient is 0.6189 which is similar to Dickerson et al.’s result of 0.5242. According to our data, a one percentage point increase in past returns causes the average ROA in the test period (2009-2012) to increase by 0.6189 percentage points. Hence, past returns have a very significant effect on the ROA in times of financial turmoil too.
The coefficient for company size (Ln (Total Assets)), measured as the natural logarithm of total assets is very close to zero. It is in fact -.0056, hence we infer, that an increase in total assets has only a minor effect on the ROA. Dickerson et al. captured company size with the aid of five dummy variables, four being slightly positive and one being slightly negative in their analysis. Therefore, our coefficient is still aligned with their results. Therefore, company size has a
similar effect on ROAs during the financial crisis, as it has in general according to Dickerson et al.
Our coefficient for Leverage is 0.0200; it is statistically significant in our regression, which means that e.g. a one percentage point increase in leverage brings about a 0.02 percentage point increase in ROA in times of a financial crisis (holding all other variables constant). Still Dickerson found a statistical
significant negative coefficient for leverage (-0.0150). One could suspect our positive coefficient to be related to the problem of endogeneity, which means that the variable might be correlated to another variable and the error term (omitted variable). However, by reviewing table 3 it becomes clear that there is no multicolinearity between the two variables. Hence, we infer that in times of a financial crisis leverage simply has the opposite effect on ROA than in general. Our Net Asset Gowth coefficient is 0.0121, which captures growth during the test period solely. If net asset growth increases by 1%, the ROA will increase by 0.0121%. Dickerson et al. obtained two coefficients for growth, a positive
coefficient for growth in the test period of 0.0733 and a coefficient of -0.0051 for lagged growth. It seems like growth had a significantly higher effect on ROAs in the time between 1949 and 1977 than during the financial crisis from 2009 to 2012.
With 0.6423 our constant is slightly smaller than Dickerson’s Gibson’s and Tsakalotos’ constant of 0.1054, which however, should have no effect on the quality of our results.
4.2 Robustness check, using a CR model
In order to confirm the validity of our results we slightly adjusted the model of Dickerson et al. by replacing the ROA by period cumulative stock return (CR). The new model looks as following:
and represent the cumulative stock returns from 2009 to 2012 and 2005 to 2008 respectively. Cumulative means that potential dividends would be reinvested into the company again. The summary statistics for our two new variables look as following:
Table 5
Comparison of cumulative stock returns of 2005-2008 and 2009-2012
The table shows observation numbers, means, standard deviations as well as minimums and maximums for the two new variables CR 05-08 and CR 09-12. Cumulative stock returns assume, that all proceeds from the stock in form of i.e. dividends are immediately reinvested the company in form of stock purchases. CR stands for the cumulative stock returns for the time periods of 01/2005 to 12/2008 and 01/2009 to 12/2012 respectively.
Variable Obs Mean Std. Dev. Min Max
CR 05-08 106 -0.2467 0.5313 -0.9734 1.923
CR 09-12 106 0.8790 0.9519 -0.5836 5.288
From this table it becomes clear that the average CR differs greatly between the two periods. In the period before the financial crisis the average yearly CR was -24.67%, during the financial crisis the average CR was however 87.90%. The reason why the CRs for the time prior to the crisis were so low is that the first market drops on the financial markets started occurring from summer 2007 onwards. From autumn\winter 2008 on, the markets started to recover and our sample companies’ stock values increased by 87.90% until 2012. The following table compares ROAs to CRs.
Table 6
Comparison average yearly ROA with average yearly CR
ROA is an abbreviation for return on assets, whereas CR is an abbreviation for cumulative return
Here it becomes very apparent that from simply looking at ROAs, one cannot predict future accumulative stock returns reliably. From 2005 to 2008 ROA was on average 6.50%, while average CR was -6.84%. Subsequently, from 2009 to 2012 the average yearly CR was 17.08%, while the average ROA only increased to 7.43%.
In our opinion, properly predicted CRs are of higher value for an investor than properly predicted ROAs. The reason for this is, that by investing in the stock the investor participates from the development of the stock price and therefore only indirectly from the development of the ROAs. Therefore, we runanother
regression, using cumulative stock returns (CRs). The following table displays the results of our adjusted regression:
Table 7
Regression of CR from 2009 - 2012 on explanatory variables
The table shows what effect different variables have on a company’s CR in the time between 01/2009 and 12/2012. Next to the variables coefficients β the table also reports standard errors of the variables, its t-test values and its p-values. Acquirer (dummy) represents the only binary variable in this regression. A constant is included into the regression and listed at the very last row. The 95% confidence interval is also included. The regression uses cumulative returns of the years 01/2005 to 12/2008 and 01/2009 to 12/2012, respectively for the two CR variables. The recorded values for Ln (Total Assets) and Leverage are from the 31.12.2012. Net Asset growth is the computed growth over the period from 2009 to 2012. The Acquirer (dummy) variable is equal to one, if the firm has conducted at least one acquisition between 2009 and 2012.
CR 09-12 Coef. Std. Err. t P>t 95% Conf. Interval CR 05-08 -0.2358 0.1879 -1.25 0.212 -0.6087 0.1370 Ln (Total Assets) -0.1714 0.0767 -2.23 0.028 -0.3236 -0.0192 Leverage 0.1617 0.2665 0.61 0.545 -0.3670 0.6903 Net Asset Growth -0.0305 0.1798 -0.17 0.866 -0.3873 0.3263 Acquirer (dummy) 0.1311 0.2171 0.60 0.547 -0.2996 0.5619 Constant 2.6284 0.8015 3.28 0.001 1.0383 4.2185
Years ROA yearly (AVG) CR yearly (AVG)
2005 - 2008 0.0650 -0.0684
Except of the Ln (Total Assets) variable all variables are statistically insignificant. The R-squared value for this regression is 0.0602, while it had been 0.6000 for our first regression.
Firstly, we can infer that adjusting Dickerson’s, Gibson’s and Tsakalotos’ model by using CRs does not help to predict stock returns in times of financial crisis. Our adjusted model is therefore not useful.
Secondly, the results of our robustness test reveal weaknesses in our first model as well; therefore our first model, which did not “pass” its robustness check, should only be used with caution, when applied to other crises.
5. Conclusion and discussion
Throughout our study we used two different approaches to explain the effect of conducted acquisitions during the financial crisis on acquirer profitability. On one hand, we applied Dickerson et al.’s model based on return on assets figures, while on the other we applied a stock price model based on the common strand of research.
Firstly, in terms of the model based on ROA we found results, which contradict the results of Dickerson, Gibson and Tsakalotos (1997). However, we cannot stress enough that our research is solely focused on the impact of an acquisition during a financial crisis, like the one from 2009 to 2012. By contrast Dickerson et al. tried to construct a general model, therefore they used a sample time period from 1948 to 1977. Thus, it would be very interesting to compare our findings on our “crisis period” topic to other “crisis studies”. Subsequently, one could
confirm the validity of our results, that acquisitions during a financial crisis have a positive effect on ROAs. However, to our knowledge there has been not a single similar study on this topic up to this date.
Secondly, we adjusted our model to a stock price based model (CR) for our robustness check. Thereby, we obtained very insignificant results, which however is in line with existing literature. As described in the literature review there have been several studies about the effects of acquisitions on stock performance with very different and ambiguous results.
One limitation of our research is certainly the chosen time period. There is obviously no official date, when the financial crisis started and when it ended (if it has already ended). By using other time horizons we think it is indeed possible to obtain results different from ours. We furthermore limited our time period to 4 years. Nonetheless we do not know how long it actually takes until the effects of an acquisition are fully reflected in a companies’ return on assets ratio or its stock price respectively. Up to this date there is no consensus about this time span among the major scientists mentioned in this paper. Thirdly, we have only reviewed a relatively small sample of 106 S&P 100 constituents. Therefore, our main results are limited to the American blue chip companies and should only be applied to other sectors with outmost caution. Fourthly, we used the financial crisis of 2009 to 2012 as our sample period in order to learn something about the effects of acquisitions conducted during a crisis on acquirer’s profitability. However, since every crisis is different, it would be reasonable to identify more similar crisis in the past. Subsequently one could verify the robustness of the ROA model and, if positively verified apply it for the future.
Conclusively, we think that there is enough statistical significance to believe that acquisitions during the recent financial crisis have helped acquirers to improve their return on assets ratios (ROAs). Relatively low company valuations during the crisis and therefore “cheap” transaction values can be the reason, why acquirers were able to boost their ROAs by acquiring other companies. This is opposed to Dickerson et al, who reported general negative effects on acquirers ROA. Therefore, the historical low market capitalizations of target companies can be the reason why acquisitions during the crisis paid off and increased
companies’ ROAs. Hence, it might well be in the interest of shareholders to encourage their CEOs to engage in acquisitions during the next financial crisis. However, we advise to not make these decisions based on our study, until more research has been done on the topic and our mentioned limitations.
References
Aw, M., Chatterjee, R., 2004, The performance of UK firms acquiring large cross-border and domestic takeover targets, Applied Financial Economics 02, 337-349.
Black, E.L., Carnes, T.A., Jandik, T., 2007; The Relevance of Target Accounting Quality to the Long-Term Success of Cross-Border Mergers Journal of Business Finance & Accounting, 34, 139–168
Conn C., Cosh A., 2001, Long-run share performance of U.K. firms engaging in cross-border acquisitions, ESRC Centre for Business Research, University of Cambridge, Working Paper No. 214.
Cosh, A.D., Guest, P.M., 2001, The long-run performance of hostile and friendly takeovers: U.K evidence’, Working Paper, Judge Institute for Management Studies, Cambridge University.
Dickerson, Hether, Gibson, 1997, The impact of acquisitions on company
performance: Evidence from a large panel of U.K. firms, Oxford Economic Papers 49, 344-361.
Gregory, A., 1997, An examination of the long run performance of UK acquiring firms, Journal of Business Finance & Accounting, 24, 7, 971-1007
Higson, C., Elliot, J., 1998, Post-takeover returns: the UK evidence, Journal of Empirical Finance, 5, 27-46.
Jensen, M.C., Ruback, R.S., 1983, The market for corporate control: The scientific evidence, Journal of Financial Economics, 11, 5-50.
Loughran, T., Vijh, A.M., 1997. ‘Do long-run shareholders benefit from corporate acquisitions?’, Journal of Finance, Volume 52, 1765-1790.
Rau, P.R., Vermaelen, T., 1998. ‘Glamour, value, and post-acquisition performance of acquiring firms’, Journal of Financial Economics, 49, 223-253.
Reddy, Srinivasa, Nangia, Kumar, Rajat (2014): The 2007-2008 global financial crisis, and cross-border mergers and acquisitions: A 26-nation exploratory study, Global Journal of Emerging Market Economies, Vol. 6, No. 3 (2014): pp. 257-281. Scherer, F. M. (1988). Corporate takeovers: the efficiency arguments', Journal of Economic Perspectives, 2, 69-82.
Schiller, R. J., 1989, Fashions, fads, and bubbles in financial markets, chapter 2 R. J. Shiller, Market Volatility, MIT Press, Cambridge, MA.
S&P Dow Jones Indices LLC , www.spdji.com, Both factsheets can be easily retrieved by using the companies’ index finder via the following link: www.us.spindices.com/index-finder/
