THE EFFECT OF ENVIRONMENTAL, SOCIAL, AND GOVERNANCE
PERFORMANCE ON MERGERS AND ACQUISITIONS
A.P. HOEKSTRA S2969327 J.J. BOSMA Supervisor I. SOUROPANIS Second supervisor Abstract
Mergers and acquisitions are a consideration for many firms. They can provide access to new technologies or shorter supply trains, resulting in better performance. However, there are large risks involved in merging with or acquiring another company. This is frequently seen in the stock for the acquirer. Using the Fama French 3 factor model and two variants with two Environmental, Social, and Governance performance indicators, this thesis aims to answer the question of whether firms that are rated higher along these dimensions provide higher returns after acquiring or merging with another company. The answer to this question is that acquirers with better ESG performance have better financial performance after the acquisition.
1 INTRODUCTION
There are many studies done on the subject of merging and/or acquiring companies and the effect it has in the stock market. The generally accepted idea is that stocks of the target company rise as the acquirer pays a premium to convince shareholders of letting go of their shares, for which the acquirer has to give up either cash, shares, or credit which decreases their opportunities for other business projects. Add to that the fact that many mergers fail to deliver the synergies initially promised to justify the merger, it is no surprise that investors are hesitant around mergers and the share price of the acquirer either stays the same or falls. This leads to the question of whether there are factors that can increase the returns of firms after the merger or acquisition. One such an area could be better Corporate Social Responsibility (CSR), which is what this thesis focuses on.
With the ever increasingly competitive nature of business, companies are looking broader and broader for avenues to distinguish themselves from their competitor in search of a competitive edge. This could, for example, be a strict focus on health and safety protocols to reduce the rate of accidents and product wastage, leading to more efficient use of resources with as drawback that production might be slower compared to a competitor with a lesser focus on health and safety but with an increased focus on high rates of production. Most firms, no matter what they distinguish themselves in, aim to make money. Some would even say it is their primary purpose, such as Friedman (1970). This would be categorised as a shareholder perspective: firms should aim to maximize the wealth of their shareholders (Yen and André, 2019, p. 114). One possibility is through merging with or acquiring another company in order to capture a larger consumer base, get access to better technologies, or simply being able to reduce costs in the supply chain.
Investors too are looking with a broader view on the financial markets. They are increasingly looking towards Socially Responsible Investment (SRI) for either increased returns or choosing not to invest in companies that do not align with their view of the world. Typical industries that fall in these categories are firms in the oil industry, guns, adult entertainment; the so-called “sin industries” in short. For investors that are looking at more than only the highest returns, companies regularly publish reports on their CSR activities and performance that investors can use to judge how well their vision aligns with that of the company. In order to help investors make a decision on these issues, a new branch of financial rating agencies has emerged, focused on rating the efforts of a firm in being more socially responsible. This rating typically consists of three main areas: Environment, Social, and Governance (ESG). This can be used as an argument for companies to invest in CSR. By investing in this area they may become part of one of the several CSR indices, increasing their exposure to CSR conscious investors.
2 better returns after merging than firms with low ESG/Governance rating. The hypothesis going with this research question is that higher governance rating leads to better financial performance after the merger. The choice for governance is based on it having the most effect on the operations of a company, and that out of the three parts of ESG, governance stands the closest to the financial markets. The avenue that companies choose to distinguish themselves by is selected by the board, governing decisions made by middle managers affecting the operations of a firm. Additionally, shareholder relations are considered in the judgement of a firms’ governance, adding a stock market angle to the factor that the other two factors do not consider.
This thesis is set up in the following way: first there is an overview of what can be found on the subject in the literature. Following this is a look at the data and the methods used to answer the research question. This continues to the results and a discussion about these results. All this leads to a conclusion in the conclusion.
LITERATURE REVIEW
In most papers about CSR, the two opposing stakeholder and shareholder theories play a role. On the one hand, there is the stakeholder view that proposes that “corporations should consider the effects of their actions upon customers, suppliers, the general public, employees, and others who have a stake or interest in the corporation” (Yen and André, 2019, p. 116). In other words, a broad view of the company encompassing all that are affected by the company. On the other hand, there is the shareholder view that proposes that businesses benefit society through making money. Any money spent on CSR that does not directly help the business would then be a waste. In mergers the opposing views might clash more, as on one hand there is what some would call unnecessary spending on CSR and would wish to reduce this to increase financial performance (the shareholder view). While on the other hand, increased CSR performance and its related decreased information asymmetry increase the chance of a successful merger (the stakeholder view).
EFFECT OF CSR ON FINANCIAL PERFORMANCE
3 stock-market based data to predict social responsibility, since accounting-based data may capture more company specific features that get drowned out in stock-market based data.
Based on their research they suggest several conclusions. The first being that managers could look towards using CSR programs to lower their risk to increase their attractiveness to the market rather than only focusing on pursuing the highest returns. Then, as their risk lowers, they can feasibly invest in more opportunities that they may have passed up on before to make money. The second conclusion they draw is, as mentioned, that accounting-based data was a better measure to use to predict CSR. Their third conclusion is that “associations found between concurrent social responsibility and performance may partially be artifacts of previous high financial performance.” (McGuire et al., 1988, p. 869). Some drawbacks of this study compared to more modern studies are the limited sample size (131 firms), and lack of consistency in scoring CSR performance. Nowadays there are various companies that specialize in rating a firms’ social performance, which brings more consistency to the ratings given to companies. Additionally, the amount of companies rated compared to back then is higher as well, further increasing data reliability and research conclusions.
Beck, Frost, and Jones (2018) looked into the effect CSR has on financial performance in three countries, all subject to differing regulations surrounding CSR reporting. The research focused on the United Kingdom (mature regulations), Australia (developed regulations), and Hong Kong (beginning with regulations). Firstly, they found a positive relation between CSR disclosure, CSR performance, and firm size, suggesting that firm size is a useful control variable for CSR engagement. Secondly, they found that their model “shows some quite strong industry-level fixed effects” (Beck et al., 2018, p. 528). This shows that industry can play quite a role in CSR engagement, and thus could be a viable control variable to better isolate the effect CSR has. Furthermore, they “find that our financial performance metric, pre-tax ROE, is again positive and significant” (Beck et al., 2018 p. 530), giving an early indication of this thesis’ result.
4 target and the acquirer ratings.” (Aktas et al., 2011, p. 1760). In other words, the greater the difference in rating between the target and the acquirer before the merger/acquisition, the more learning takes place. Having better governance may help this learning process, as they could make the transfer of information between the target and the acquirer more effective.
El Ghoul, Guedhami, Kwok, and Mishra (2011) research whether CSR performance can affect financial performance through the cost of capital. They argue that firms that invest more into CSR have lower information symmetry, due to the extra releases for CSR performance, increased coverage by analysts, and selection by investors conscious of what they invest in. The authors base this on Merton’s model. One part of which is that investors invest in things they know. Better CSR performance is a way to bring more awareness to their company, increasing the size of the investor base, and thus lowering the cost of capital. Furthermore, high CSR performing firms may face less litigation risk, due to the higher standards their products meet. They found that “firms with higher CSR scores enjoy significantly lower cost of equity capital.” (El Ghoul et al., 2011, p. 2401). Additionally, they looked at several specific factors comprising the CSR score companies received. Out of six factors they found that better employee relations, environmental performance, and product strategies lowered the cost of capital, pointing to specific areas companies could make an effort to work towards in order to improve their CSR performance, and lower their cost of capital.
Yen and André (2019) look at one of the more interesting areas of study regarding CSR and mergers, namely Emerging Markets. These markets do not yet have reached maturity, resulting in, among others, more information asymmetry, and a weaker legal system, providing more opportunities for companies to run rampant without consequence. About CSR investment for these companies the authors write: “we specifically acknowledge that EM acquirers have stronger incentives to undertake a significant SRIs before proposing a cross-border deal for their reputation enhancement…” (Yen and André, 2019, p. 115). As a downside, expected synergies must be higher in order to overcome the increased spending on CSR, based on a shareholder view. A strong corporate governance culture in addition to strong country legal institutions may help mitigate these negative effects.
5 Luffarelli, Markou, Stamatogiannakis, and Gonçalves (2018) researched whether there is an effect of Corporate Social Performance (CSP) on sales, and secondly whether or not there is a difference in the effect CSP has on sales between business-to-business firms, and business-to-consumer firms. Using a dataset of over 27000 observations, they find that for the most part, CSP decreases sales performance, with the exception of the diversity dimension. Furthermore, they found that the effect differs between business and business-to-consumer firms. For business-to-business firms the effect of CSP is negligible, while for business-to-consumer firms the effect is negative. When looking at the effect of CSP on financial performance per category, only the effect of employee diversity has a positive effect on financial performance, the other 6 categories (as defined by KLD Stats) have either an insignificant or negative effect. This paper provides a counterpoint to other papers read, which make the point that CSR has a positive effect on performance, showing that the debate around the effect of CSR is not done yet.
MERGERS
In mergers the effects of governance quality can help in solving issues and prevent value leakages that may occur. Meyer (2008) researches and theorises several leakages that can occur during mergers and how these can be prevented. Two areas are looked at: rent-seeking and implicit costs. Rent-rent-seeking is the process whereby “managers and employees seek their share of the gains by bargaining for private benefits at the expense of the shareholders (…).” (Meyer, 2008, p. 198). In other words; managers and employees preferring their own good over that of the company. Implicit costs are costs that “arise as a result of managers and employees reducing and/or reallocating their effort during the post-merger integration process.” (Meyer, 2008, p. 198).
Meyer looks at several different sources of leakages and solutions to those problems based on mergers with differing success in order to find out what went wrong/correct, why it happened that way, and what can be done to increase the likelihood of success. Problems (and solutions) that are included are diffuse governance structures (making predetermined decisions, setting up programmes to resolve governance disputes), excessive autonomy (change management, monitor the post-integration process, introducing second best solutions), rent-seeking employees (getting employees committed to the main goals early, settle disputes early, have an alternative agreement if possible), reduced effort (good communication, involvement, fair treatment of the other party), reallocation of effort (selective participation, digital communication to reduce travel time, extra capacity). Good governance can help companies be on top of these issues, resulting in less waste of resources increasing the ease at which the merge can be completed. This, in turn, can lead to synergies appearing sooner, increasing financial performance.
6 governance performance, as this strategy requires managers and personnel to always follow trends and spot opportunities to take advantage of and adjust their internal processes. Their research focusses on the flows of knowledge after an acquisition between the target and the acquirer. They found that “acquirer knowledge transfer contributed to performance following the acquisition” (Junni et al., 2015, p. 610), where acquirer knowledge is knowledge transferred from the acquirer to the target. This performance increase is attributed to efficiency increases and cost saving. This again gives an early indication of what the answer to the research question will be.
Deng, Kang, and Low (2013) combine the two themes of this thesis in their paper. They find that “compared with mergers by low CSR acquirers those by high CSR acquirers lead to higher announcement stock returns for acquirers (…)” (Deng et al., 2013, p. 108). The stakeholder theory and shareholder theory are once again put up against each other. Out of these two opposing theories Deng et al. (2013) find more support for the stakeholder theory. One of the arguments made earlier in support of the stakeholder view also holds in this paper, namely that “an EM acquirer with better pre-merger CSR performance engages and completes an international deal more easily.” (Yen and André, 2019, p. 125). Deng et al. (2013) find a similar conclusion for the US market that they focus on.
This paper, too, suggests a positive outcome for the hypothesis made; “unlike combined firms with high CSR acquirers that experience no significant change in post-merger operating performance, combined firms with low CSR acquirers experience deterioration in their post-merger operating performance.” (Deng et al., 2013, p. 98). Furthermore, they find that stock performance of acquirers with high CSR outperforms that of firms with a lower CSR rating.
DATA & METHODOLOGY DATA
The data for this thesis is gathered from three sources. Stock data is obtained from the CRSP database, accounting data from Compustat, and data about governance quality from the Thomson Reuters Asset 4 database. The data was selected in the following manner; first all monthly stock data from January 2005 to January 2015 was chosen in CRSP along with the delisting code and acquirer variables. Then the data was restricted to those companies that are involved in mergers as either target or acquirer. The resulting Tickers were then put in the aforementioned databases to collect the data needed for the same time period. In order to merge the different datasets, the BoardEx CRSP Compustat Link tool is used to provide a common indicator variable, Compustat’s gvkey, to be used for linking, as well as following a company throughout time. Furthermore, the Fama-French 3 factor data is used for market returns, the High-minus-Low, Small-minus-Big factors, and the risk-free rate.
7 CSR strategy rating, management scores, and regulations around shareholder rights. In order to use the Governance score from the Asset 4 database, the letter grades were converted to a numeric scale in the following way, see table 1.
The ESG score used is an overall measure of how well a company is performing in the three typical fields of CSR: Environmental, Social, and Governance. In addition to the three factors used for governance rating, this score incorporates resource use, emissions, innovation, workforce, human rights, community, and product responsibility factors in the rating.
The shares considered in the sample have a Share Code of either 10 or 11 in the CRSP database; in other words, these are Ordinary Common Shares, that do not need further defining. These are the shares typically considered in studies.
TABLE 1
Conversion of Letter Grades to Numeric Grades
Letter Grade Numeric Grade
A+ 9 A 8 A- 7.5 B+ 7 B 6.5 B- 6.3 C+ 6.25 C 6 C- 5.75 D+ 5.5 D 5 D- 4.75
Note: Adapted from and based on https://students.uu.nl/sites/default/files/geo-grading-systems-holland-vs-us-uk.pdf
TABLE 2 Summary Statistics
(1) (2) (3) (4) (5)
VARIABLES N mean sd min max
Price 157,082 23.34 37.58 -252.5 1,349
Return 156,448 0.0110 0.156 -0.935 7.007
Market return 160,475 0.466 4.433 -18.41 11.56
SMB factor 160,475 0.0700 2.164 -4.180 5.810
HML factor 160,475 0.0300 2.102 -7.260 5.490
Overall ESG Score 427 51.04 18.25 13.18 92.16
Governance Score 415 6.499 0.840 5 9
8 SUMMARY STATISTICS
Table 2 shows the summary statistics for some of the more important factors used in this thesis.
METHODOLOGY
As mentioned, the Compustat company identifier is used to get the different datasets to work together. This identifier along with the date variable is used to set the dataset up as panel data. The CRSP Permanent Company Number is used to match targets and acquirers, as well as the date the merger or acquisition takes place. Using these dates, estimation windows can be set up to allow for estimation around the date of the merger event to set up event studies.
The regressions used to answer the research question are the following: the Capital Asset Pricing Model (CAPM), the Fama-French 3 factor model, as well as two variants of this 3 factor model. These variants have an additional factor that measures either the effect of the Governance rating or the ESG rating of the firm on returns. These factors are constructed in the following way:
𝐺𝑜𝑣𝑒𝑟𝑛𝑎𝑛𝑐𝑒 & 𝐸𝑆𝐺 𝐹𝑎𝑐𝑡𝑜𝑟 =10% 𝑏𝑒𝑠𝑡 𝑠𝑐𝑜𝑟𝑒𝑠−10% 𝑤𝑜𝑟𝑠𝑡 𝑠𝑐𝑜𝑟𝑒𝑠
𝑛𝑢𝑚𝑏𝑒𝑟 𝑜𝑓 𝑜𝑏𝑠𝑒𝑟𝑣𝑎𝑡𝑖𝑜𝑛𝑠 (1)
These factors are made on an annual basis, as companies are rated yearly in the Asset4 programme.
The returns of firms are calculated based on continuous pricing:
𝐸𝑥𝑐𝑒𝑠𝑠 𝑅𝑒𝑡𝑢𝑟𝑛𝑖,𝑡 = ln(𝑃𝑟𝑖𝑐𝑒𝑖,𝑡) − ln(𝑃𝑟𝑖𝑐𝑒𝑖,𝑡−1) − 𝑟𝑓,𝑡 (2) Where rf,t is the risk free rate at time t as reported on the Fama-French website, and Pricei,t the price of the stock of firm i at time t.
This leads to the following regressions used to estimate the results: Jensen’s CAPM:
𝐸𝑥𝑐𝑒𝑠𝑠 𝑅𝑒𝑡𝑢𝑟𝑛𝑖,𝑡 = 𝛼𝑖 + 𝛽1(𝑟𝑚,𝑡− 𝑟𝑓,𝑡) + 𝜀 (3)
Fama-French 3-factor model:
𝐸𝑥𝑐𝑒𝑠𝑠 𝑅𝑒𝑡𝑢𝑟𝑛𝑖,𝑡 = 𝛼𝑖 + 𝛽1(𝑟𝑚,𝑡− 𝑟𝑓,𝑡) + 𝛽2𝑆𝑀𝐵𝑡+ 𝛽3𝐻𝑀𝐿𝑡+ 𝜀 (4) Fama-French 3 factor model with additional Governance factor:
𝐸𝑥𝑐𝑒𝑠𝑠 𝑅𝑒𝑡𝑢𝑟𝑛𝑖,𝑡 = 𝛼𝑖 + 𝛽1(𝑟𝑚,𝑡− 𝑟𝑓,𝑡) + 𝛽2𝑆𝑀𝐵𝑡+ 𝛽3𝐻𝑀𝐿𝑡+ 𝛽4𝐺𝑜𝑣𝑡+ 𝜀 (5) Fama-French 3 factor model with additional ESG performance factor:
9 Where ri,t is the return of firm i at time t, rf,t is the risk-free rate at time t. This combines to form the excess return mentioned earlier, in formula 2. α is the abnormal return of stock i in excess of what would be expected based on the Capital Market Line, β1 measures the effect of the market return on the return of the stock, β2 measures the influence of the High-minus-Low effect (value effect) on the return of firm i, β3 measures the influence of the Small-minus-Big effect (size effect) on the return of firm i, β4 measures the effect of the governance performance factor of firm i on the return, β5 measures the effect of the overall ESG factor on the return of firm i, and ε is the error term, capturing the residuals.
These models are used for all the different smaller samples made from the whole sample.
RESULTS
Firstly, several fixed effects panel regressions are run on the complete sample in order to identify the effect, if any, the added factors have on the performance of the model. Here a fixed effect means that the factors in the model appropriately explain the changes within firms. These fixed effects regressions provide the results as shown in table 3.
These early results provide some interesting points. The first being that over this period (2005-2015), abnormal returns are negative, showing that it was not a good period to be investing in the stock market. If an investor were to hold a portfolio of the companies in the sample, they would have experienced a loss on their portfolio. This shows that the recession of 2008 left great marks on the economy of the United States, that later years have not compensated for. There may be a bias in the sample due to the focus of this study. As will be shown further on, there are 1204 targets and 521 acquirers. These targets might be a target due to weaker financial performance as a consequence of this crisis, while acquirers may be the stronger companies that survived the financial crisis of 2008 and performed well after that. Additionally, only firms involved in either a merger or acquisition are in the sample, so there is no comparison possible with firms that either went bankrupt, has become privatized, or went through this time period without being acquired by or acquiring themselves another company in the stock market.
10
TABLE 3
Results from Fixed Effects Regression on Entire Sample
(1) (2) (3) (4)
VARIABLES CAPM FF3 FF3 + Governance FF3 + ESG
β1(rm,t− rf,t) 1.276*** 1.078*** 1.077*** 1.061*** (0.0173) (0.0165) (0.0164) (0.0165) β2SMBt 0.712*** 0.707*** 0.713*** (0.0236) (0.0236) (0.0236) β3HMLt 0.268*** 0.275*** 0.288*** (0.0288) (0.0290) (0.0290) β4Govt 0.0229*** (0.00746) β5ESGt 0.00237*** (0.000213) α -0.00713*** -0.00678*** -0.0387*** -0.0718*** (8.21e-05) (7.90e-05) (0.0104) (0.00586) Observations 152,213 152,213 152,213 152,213 R-squared (within) 0.124 0.131 0.131 0.132 Number of firms 2,533 2,533 2,533 2,533 R-squared (overall) 0.123 0.131 0.131 0.131 Type of model FE FE FE FE
Robust standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1
Note: This table shows the results of a fixed effects panel regression on the entire sample for the
period of 2005 to 2015, using the four models of interest for this thesis.
EVENT STUDY
These regressions lead to the next part of this thesis; to see whether better governance performance leads to better financial performance after mergers. This is done through running regressions over a window of a year before and after the merger event, based on the perspective of the acquirer. The choice for a year after the merger event is made based on the time it takes to integrate companies. It takes time for a merger to have its effects on the people that work in a company, whether they are a target or acquirer. In choosing a year, it gives time for a company to (start to) reorganise their structure to accommodate the new parts. The year before the merger event is taken in order to increase data availability to increase the accuracy of the regression and cancel out (some) investor speculation driving up prices of different companies.
11
TABLE 4
Results from Fixed Effects Regression Before and After Merger Events (Acquirer Perspective)
(1) (2) (3) (4)
VARIABLES CAPM FF3 FF3 + Governance FF3 + ESG
β1(rm,t− rf,t) 1.295*** 1.111*** 1.110*** 1.103*** (0.0524) (0.0512) (0.0507) (0.0489) β2SMBt 0.594*** 0.594*** 0.592*** (0.0673) (0.0674) (0.0673) β3HMLt 0.392*** 0.405*** 0.398*** (0.0920) (0.0925) (0.0920) β4Govt 0.0769 (0.0590) β5ESGt 0.00347 (0.00312) α -0.00946*** -0.00877*** -0.116 -0.105 (0.000293) (0.000284) (0.0823) (0.0865) Observations 12,428 12,428 12,428 12,428 R-squared (within) 0.148 0.155 0.156 0.156 Number of firms 611 611 611 611 R-squared (overall) 0.153 0.160 0.160 0.160 Type of model FE FE FE FE
Robust standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1
Note: This table shows the results from a fixed effects panel regression run on the acquirers, with a
timespan of one year before and after the merger or acquisition event.
These regressions provide similar results to the first set or regressions. Abnormal returns are again negative, and greater in magnitude compared to the results of the fixed effects regressions earlier. These returns are significantly different from zero for the two traditional models, while for the models with the additional factors the return is negative, but not statistically significant. This is, for the traditional models, in line with the common reasoning that companies overpay to secure the buyout of their target. This overpaying is extra expenses that lower the performance of the acquirer, which is reflected in lower stock returns.
12 EFFECT OF BETTER ESG
The main goal of this thesis is to find out whether better governance performance leads to better financial performance. That will be looked at in this section. Due to data constraints, ESG score is taken rather than governance rating of the company. The same models are used as previously, with the sample divided into two parts, a lower and a higher ESG score. The results are provided in tables 5 (low ESG score) and 6 (high ESG score).
These results show that having a high ESG score is rewarded. The different risk factors are found to be lower for the acquirers with high ESG scores. The abnormal returns, while still negative are somewhat higher for the firms with higher ESG scores in the traditional models. Testing whether this difference is significant returns p-values of 0.01 for the CAPM and 0.000 for the Fama French 3 factor model. This may point towards the stakeholder argument in which CSR spending is a rewarded expense. In line with this, in the sample with low ESG scores, the Fama-French 3 factor model with Governance factor returns a not significant, but negative result, compared to a not significant and positive result in the sample with high ESG scores. Testing, however, shows that this difference is not significant. For the model with the ESG factor, the return for the sample with higher ESG scores is less negative in the high ESG score sample. This difference between the abnormal returns is significant, at the 5% level.
TABLE 5
Results from Random Effects Regression on a Sample with Low ESG Scores
(1) (2) (3) (4)
VARIABLES CAPM FF3 FF3 + Governance FF3 + ESG
β1(rm,t− rf,t) 2.230*** 2.158*** 2.152*** 2.027*** (0.221) (0.310) (0.290) (0.355) β2SMBt 1.155 1.159 1.531** (0.759) (0.736) (0.700) β3HMLt -0.715 -0.706 -0.918** (0.438) (0.535) (0.396) β4Govt 0.0137 (0.220) β5ESGt 0.0154* (0.00912) α -0.0351*** -0.0477*** -0.0674 -0.496* (0.0115) (0.00926) (0.320) (0.268) Observations 41 41 41 41 Number of firms 19 19 19 19 R-squared (overall) 0.395 0.427 0.427 0.443
Robust standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1
Note: This table shows the results of a random effects panel regression on the acquirers. The time
13
TABLE 6
Results from Fixed Effects Regression on a Sample with High ESG Scores
(1) (2) (3) (4)
VARIABLES CAPM FF3 FF3 + Governance FF3 + ESG
β1(rm,t− rf,t) 0.815*** 0.398 0.404 0.291 (0.143) (0.284) (0.279) (0.310) β2SMBt 0.932** 0.961** 1.028** (0.423) (0.466) (0.422) β3HMLt -0.756 -0.767 -0.718 (1.128) (1.098) (1.124) β4Govt -0.0572 (0.273) β5ESGt 0.00444 (0.00547) α -0.00267 0.00114 0.0792 -0.122 (0.0126) (0.00883) (0.380) (0.157) Observations 68 68 68 68 Number of firms 16 16 16 16 R-squared (overall) 0.0469 0.0868 0.0874 0.0957
Robust standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1
Note: This table shows the results of a random effects panel regression on the acquirers. The time
period considered is one year before and after the merger or acquisition. The sample is split into two parts based on their ESG rating. Table 6 shows the results for companies with an ESG scores greater than 50/100.
Another interesting point to point out is how the factor weightings change between the two variants. For the sample with low ESG scores, the market factor (β1) is lower than in the sample with high ESG scores, while the size effect (β2) is greater for the low ESG scoring firms. Furthermore, the growth effect is significant for the models with the high ESG scores, but not for the firms with lower ESG scores.
DISCUSSION MODEL SUITABILITY TEST
For each different section of regressions, Hausman tests are ran to determine the suitability of a random effects model. If based on the Hausman test different models are recommended, the outcome for the Fama French 3 factor model recommendation is chosen for all 4 variants. This choice is made to make the results comparable and consistent within a section.
TESTING THE EFFECT OF THE ADDED FACTORS
14 The estimation results of these OLS regressions are stored and the average excess return is calculated. This is then ran as the following cross-sectional model:
𝐸[𝐸𝑥𝑐𝑒𝑠𝑠 𝑟𝑒𝑡𝑢𝑟𝑛𝑖] = 𝜆1𝑖,𝑡+ 𝜆2𝑖,𝑡𝛽̂1𝑖,𝑡+ 𝜆3𝑖,𝑡𝛽̂2𝑖,𝑡+ 𝜆4𝑖,𝑡𝛽̂3𝑖,𝑡 (+𝜆5𝑖,𝑡𝛽̂4𝑖,𝑡+
𝜆6𝑖,𝑡𝛽̂5𝑖,𝑡) (7)
Where 𝛽̂ are the stored estimators from the models; 𝛽̂1 is for the market return, 𝛽̂2 for the exposure to the small firm effect, 𝛽̂3 for the exposure to the value effect, 𝛽̂4 for the governance factor, and 𝛽̂5 for the ESG factor. The E[Excess returni] is the average excess return for firm i calculated in the OLS regressions that are run for each company. The subscripts indicate the different companies at different times. This yields the results shown in table 7.
As can be seen from table 7, the added factors do not seem to add to the precision of the model since the constant is further away from zero compared to the Fama French model. If these additional factors are to help in the prediction, the expectation is that the constants would be closer to 0. On the other hand, these additional factors help in explaining more of the variance as seen by the increased R-squared.
TABLE 7
Cross-sectional OLS Model Results
(1) (2) (3) (4)
VARIABLES CAPM FF3 FF3 + Governance FF3 + ESG
𝜆2 -0.00101*** -0.000547*** -0.00118*** -0.00121***
(7.99e-05) (7.28e-05) (7.75e-05) (7.78e-05)
𝜆3 -0.000367*** -0.000378*** -0.000449***
(4.96e-05) (5.63e-05) (5.60e-05)
𝜆4 -0.000537*** -0.00111*** -0.00114***
(5.97e-05) (4.82e-05) (4.83e-05)
𝜆5 5.78e-05***
(7.03e-06)
𝜆6 -0.000644
(0.00123)
𝜆1 0.000183* -0.000251** 0.000379*** 0.000475***
(0.000105) (9.84e-05) (9.54e-05) (9.58e-05)
Observations 158,990 158,158 132,039 132,051
R-squared 0.004 0.008 0.022 0.022
Robust standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1
Note: This table shows the results of applying a cross-section of the results on the abnormal return
15
ENTIRE SAMPLE
As mentioned above, the return over the entire sample is negative. The influence of the financial crisis in 2008 plays a significant role in this. Running the two classic models on the sample divided split over 2005-2009 and 2010-2015 yields the results shown in table 8.
TABLE 8
Comparing Two Time Periods Using the Traditional Models
(1) (2) (3) (4) CAPM CAPM FF3 FF3 VARIABLES 2005-2009 2010-2015 2005-2009 2010-2015 β1(rm,t− rf,t) 1.312*** 1.181*** 1.105*** 0.994*** (0.0211) (0.0205) (0.0207) (0.0207) β2SMBt 0.732*** 0.655*** (0.0279) (0.0385) β3HMLt 0.279*** 0.281*** (0.0345) (0.0397) α -0.00897*** -0.00291*** -0.00946*** -0.00117*** (1.92e-05) (0.000234) (4.44e-05) (0.000237) Observations 96,671 55,542 96,671 55,542 R-squared (within) 0.128 0.110 0.137 0.116 Number of firms 2,344 1,503 2,344 1,503 R-squared (overall) 0.128 0.107 0.136 0.113 Type of model FE FE FE FE
Robust standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1
Note: This table compares the sample over two time periods, 2005 – 2009 and 2010 – 2015, using the
two traditional models (CAPM and Fama French 3 factor model) in a fixed effects model. The goal is to discern if there is a difference between the two time periods.
This shows that in the earlier period, markets were riskier when looking at the market component of the formulas. Looking at the α, the abnormal return, shows that the return in the second time period, while still negative, is higher. Using a χ2 test these differences are found to be significant (p-value of 0.0000) for both the CAPM and the Fama French 3 factor model. What this shows is that the financial markets have been recovering after the financial crisis, however, not enough to even out the big hit the markets took in 2008. It is possible that there may be a sample effect in this due to a large part of the sample being companies that are taken over, 1595 out of 2486 to be exact. One of the reasons a company could be targeted is that they are struggling financially, resulting in lower financial performance, which of course makes the target more affordable. This could affect overall returns.
16 other factors are 0.00466, 0.000700, and 0.000300 for the market return, SMB, and HML factors respectively. However, this does not explain why the abnormal return goes down compared to the other models, as scaling down the factor to be more in line with the other factors only multiplies the coefficient by the same amount.
The difference might be explained by the ESG factor being based on an annual rating compared to the other factors that are on a monthly basis. This may make the model more inaccurate on a monthly basis, increasing the inaccuracy of the model. This might also explain the small increase in the explanatory power of the model. If this is the case, the increase may only be attributable to the increase in factors, rather than a factor that increases the explanatory power of the model through patterns that help explain the return.
RETURN AROUND THE EVENT
Table 9 shows the four regressions run on the targets, with samples starting a year before the date of the event. The two original models, in addition to those in table 4, show that the conventional adage of targets return going up in the run up to a takeover, while that of acquirers goes down holds for this sample. When comparing the same amount of time (i.e. a year before the merger/acquisition event, shown in table 10) for both samples the difference is 2.07 percentage points for the CAPM and 2.11 percentage points for the Fama French 3 factor model. Another point of note to support this phenomenon is that the models from the acquirer perspective explain a larger part of the variance in returns (R2 of 0.14for acquirers versus 0.06 for targets) showing that there are more factors at play for targets that cannot be captured by the model. It is likely that as the merger comes closer, news starts to leak, and the markets take note of the upcoming event. Running the models on a smaller time period of 100 days before the event shows that the gap is larger for the smaller time period, as shown below in tables 12 and 13. Of note here, in addition to the α are the weightings of the factors. Especially the market exposure is much lower for targets compared to the results in table 9.
17
TABLE 9
Results from Random Effects Regression around Merger Events (Target Perspective)
(1) (2) (3) (4)
VARIABLES CAPM FF3 FF3 + Governance FF3 + ESG
β1(rm,t− rf,t) 0.995*** 0.873*** 0.871*** 0.845*** (0.0459) (0.0500) (0.0500) (0.0506) β2SMBt 0.535*** 0.506*** 0.535*** (0.0788) (0.0795) (0.0787) β3HMLt -0.0349 -0.00321 -0.00898 (0.0942) (0.0945) (0.0939) β4Govt 0.0930*** (0.0213) β5ESGt 0.00321*** (0.000493) α 0.0179*** 0.0186*** -0.111*** -0.0699*** (0.00114) (0.00115) (0.0294) (0.0135) Observations 13,381 13,381 13,381 13,381 Number of GVKEY 1,204 1,204 1,204 1,204 R-squared (overall) 0.0609 0.0644 0.0654 0.0661
Robust standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1
Note: This table shows the results from random effects regressions run on the target companies in the
sample, running from one year before the date of the takeover.
TABLE 10
Results from Random Effects Regression 1 Year Before Merger Event (Acquirer Perspective)
(1) (2) (3) (4)
VARIABLES CAPM FF3 FF3 + Governance FF3 + ESG
β1(rm,t− rf,t) 1.181*** 1.013*** 1.011*** 1.009*** (0.0635) (0.0627) (0.0624) (0.0625) β2SMBt 0.533*** 0.535*** 0.534*** (0.0920) (0.0922) (0.0921) β3HMLt 0.291*** 0.299*** 0.295*** (0.104) (0.105) (0.105) β4Govt 0.0789 (0.0796) β5ESGt 0.00140 (0.00182) α -0.00279 -0.00250 -0.112 -0.0409 (0.00287) (0.00287) (0.111) (0.0514) Observations 5,897 5,897 5,897 5,897 Number of GVKEY 521 521 521 521 R-squared (overall) 0.140 0.146 0.145 0.146
Robust standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1
Note: This table shows the results from random effects regressions run on the acquirer sample for a
18
TABLE 11
Results from Random Effects Regression 1 Year After Merger Events (Acquirer Perspective)
(1) (2) (3) (4)
VARIABLES CAPM FF3 FF3 + Governance FF3 + ESG
β1(rm,t− rf,t) 1.360*** 1.166*** 1.164*** 1.154*** (0.0728) (0.0749) (0.0745) (0.0735) β2SMBt 0.631*** 0.625*** 0.637*** (0.0992) (0.0993) (0.0993) β3HMLt 0.439*** 0.466*** 0.455*** (0.127) (0.128) (0.127) β4Govt 0.102* (0.0542) β5ESGt 0.00249** (0.00119) α -0.0111*** -0.00993*** -0.152** -0.0792** (0.00195) (0.00195) (0.0751) (0.0328) Observations 7,026 7,026 7,026 7,026 Number of firms 598 598 598 598 R2 overall 0.150 0.157 0.158 0.158
Robust standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1
Note: This table shows the results from random effects regressions run on the acquirer sample, for a
period of 1 year starting at the date of the merger or acquisition event.
TABLE 12
Results from Random Effects Regression 100 Days Before Merger Events (Acquirer Perspective)
(1) (2) (3) (4)
VARIABLES CAPM FF3 FF3 + Governance FF3 + ESG
β1(rm,t− rf,t) 1.260*** 1.107*** 1.106*** 1.104*** (0.118) (0.113) (0.113) (0.115) β2SMBt 0.465*** 0.466*** 0.467*** (0.146) (0.146) (0.145) β3HMLt 0.403* 0.414* 0.406* (0.242) (0.242) (0.243) β4Govt 0.0823 (0.123) β5ESGt 0.000601 (0.00211) α -0.00633 -0.00605 -0.121 -0.0226 (0.00403) (0.00402) (0.171) (0.0599) Observations 1,967 1,967 1,967 1,967 Number of GVKEY 517 517 517 517 R-squared (overall) 0.128 0.133 0.133 0.134
Robust standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1
Note: This table shows the results from random effects regressions run on the acquirer sample, starting
19
TABLE 13
Results from Random Effects Regression 100 Days Before Merger Events (Target Perspective)
(1) (2) (3) (4)
VARIABLES CAPM FF3 FF3 + Governance FF3 + ESG
β1(rm,t− rf,t) 0.576*** 0.472*** 0.483*** 0.435*** (0.0976) (0.102) (0.102) (0.103) β2SMBt 0.486*** 0.394** 0.478*** (0.171) (0.171) (0.171) β3HMLt -0.0937 -0.0196 -0.0611 (0.169) (0.167) (0.168) β4Govt 0.266*** (0.0495) β5ESGt 0.00492*** (0.00115) α 0.0584*** 0.0587*** -0.312*** -0.0776** (0.00263) (0.00259) (0.0685) (0.0315) Observations 3,530 3,530 3,530 3,530 Number of GVKEY 1,195 1,195 1,195 1,195 R-squared (overall) 0.0191 0.0218 0.0279 0.0252
Robust standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1
Note: This table shows the results from random effects regressions run on the target sample, starting at
100 days before the merger or acquisition event. EFFECT OF BETTER ESG
As mentioned in the results chapter, rather than the governance scores, general ESG scores were used due to data constraints. Not for all companies involved is data available for the ESG score, and generally even less so for governance scores. There are 59 observations for governance score in the time period around the event date for the acquirer, while there are 109 observations for ESG scores out of a total of 12,689 observations in the event window before the merger or acquisition event. This, while still not very many observations compared to the total number of observations, should at least provide slightly more reliable inferencing. The suspected reason for not so many observations is that not all companies are being rated, especially in the earlier years, as it was developing itself after being founded in 2003. Furthermore, because no filters such as minimum market value of the companies or value of the deal are applied to the sample collection, smaller companies can be part of the sample that may not be selected by the rating agency in their annual rating. Additionally, it can be argued that the increased focus of investors on CSR is a more recent phenomenon resulting in more interest for these services. Since most of the sample is concentrated in the earlier years, coverage is lower still.
20 ESG scores are not significant, the differences between them are significant for the CAPM, Fama French 3 factor model and the EGS model, at the 5%, 1%, and 5% levels respectively. Additionally, market risk and size effect risk are greater for companies with lower ESG scores, while the value effect is greater compared to companies with a lower ESG score. This may mean that investors see more CSR investment as reducing risk to a company. The increased information given out can help investors to more accurately value a company, reducing the risk that they face.
Although the differences between the SMB factors are not significant, they may provide some indication of what is going on. The lower exposure to the SMB factor, can indicate that the companies with higher CSR scores are larger than those with lower CSR scores. This, in turn, could indicate that smaller companies do not consider investing in CSR a priority. Running a correlation on their overall ESG Score and Governance Score with the market value of a company results in positive, but weak, correlations of 0.389 for ESG and 0.372 for the Governance Score. Larger companies may have more disposable income available to invest in CSR and could use this to distinguish themselves from their competitors to receive more attention from investors who also look at the performance of companies in the social responsibility area, going along with the stakeholder theory.
Smaller companies, on the other hand, may not have the cash flow necessary to invest in CSR programmes in addition to their normal operations. For these companies, the wiser choice may be to search for projects with high returns in order to distinguish themselves from the more well-known stocks, which has been shown to play a role in consumer stock selection in the field of behavioural finance. The higher returns these smaller companies may be looking for, come with higher risks, which is reflected in their increased exposure to market risks. For these companies the shareholder argument may be more fitting than the stakeholder argument as presented earlier and in, among others, Yen and André (2019).
Another factor that may be of interest is the governance factor. For the high ESG score sample the sign on the governance beta is negative, while in the low ESG sample the sign is positive. This difference is not significant, but it may indicate about a change in how investors feel about companies with higher ESG. One of the factors that is used to determine the governance score is independence of managers. An increased independence of managers may let investors feel like they do not have enough control on the company. On the other hand, for companies with lower governance score, another factor of the governance score may increase attractiveness for investors, namely shareholder protections. Increased shareholder protections may help investors feel more secure about where they are investing their money.
21 LIMITATIONS
There are areas in this thesis that further research can improve on. Probably the largest is that for the main research question, only 109 observations are available. This can be remedied by having access to a different database for CSR ratings such as MSCI’s KLD Stats, which the author does not have access to. If there are more observations available, the accuracy of the inferences made increases, making a more impactful effect on the existing literature.
Another area where this thesis may fall short is in the data for which companies are involved in mergers. In this thesis, the “Acquiring PERMNO” and “Acquiring PERMCO” variables in the Wharton CRSP database are used to determine which companies are target or acquirer. Other databases may have more accurate data, such as the Zephyr database by Bureau van Dijk, which is not accessible to the author. This database specialises in mergers and acquisitions. This may have resulted in a larger selection of companies in the sample. The result of that would be more observations, which can lead to more matches with the ESG data. This, in turn, would lead to more reliable inferencing.
CONCLUSION
Using the CRSP and Asset4 databases this thesis aims to answer whether better ESG performance leads to better financial performance post-merger compared to firms with lower ESG performance. To answer this research question, Jensen’s Capital Asset Pricing Model, the Fama French 3 Factor model, and two models with an additional Governance or ESG factor are used.
Based on these regressions, the answer to this research question is that if the acquirer has a higher ESG rating this indeed leads to better financial performance. However, the common rationale about stock market performance for the acquirer still holds. On average, the return following a merger or acquisition as the acquirer results in negative returns. In this thesis it is shown that this holds over a period of 1 year after the event.
The literature discussed in the literature review further supports this answer. For example, Deng et al. (2013) find a similar conclusion that better CSR scores leads to better financial performance. Yen and André (2019) find similar results in their sample for emerging markets: “acquiring firms in the sample are performing better than their industry counterparts in the pre- and post-acquisition periods (…)” (p. 126).
This conclusion aligns with the stakeholder view of CSR investment whereby companies that consider their impact on their direct and indirect relations are rewarded for doing so.
22 REFERENCES
Papers
Aktas, N., Bodt, E. D., & Cousin, J.-G., 2011. Do financial markets care about SRI? Evidence from mergers and acquisitions. Journal of Banking & Finance, 35(7), 1753–1761. Beck, C., Frost, G., & Jones, S., 2018. CSR disclosure and financial performance revisited: A
cross-country analysis. Australian Journal of Management, 43(4), 517–537.
Deng, X., Kang, J.-K., & Low, B. S., 2013. Corporate social responsibility and stakeholder value maximization: Evidence from mergers. Journal of Financial Economics, 110(1), 87–109.
Ghoul, S. E., Guedhami, O., Kwok, C. C., & Mishra, D. R., 2011. Does corporate social responsibility affect the cost of capital? Journal of Banking & Finance, 35(9), 2388– 2406.
Junni, P., Sarala, R. M., Tarba, S. Y., & Weber, Y., 2015. The Role of Strategic Agility in Acquisitions. British Journal of Management, 26(4), 596–616.
Luffarelli, J., Markou, P., Stamatogiannikis, A., & Gonçalves, D., 2019. The effect of corporate social performance on the financial performance of business‐to‐business and business‐to‐consumer firms. Corporate Social Responsibility and Environmental Management, 26(6), 1333–1350.
McGuire, J. B., Sundgren, A., & Schneeweis, T., 1988. Corporate Social Responsibility and Firm Financial Performance. Academy of Management Journal, 31(4), 854–872. Meyer, C. B., 2008. Value Leakages in Mergers and Acquisitions. Long Range Planning,
41(2), 197–224.
Yen, T.-Y., & André, P., 2019. Market reaction to the effect of corporate social responsibility on mergers and acquisitions: Evidence on emerging markets. The Quarterly Review of Economics and Finance, 71, 114–131.
Online Article
23 Data
CRSP data: Wharton Research Data Services. "CRSP Monthly Stock" wrds.wharton.upenn.edu
Compustat data: Wharton Research Data Services. "WRDS" wrds.wharton.upenn.edu ESG data: Thomson Reuters Asset 4 database
