Hypothesis

Motivated by the mixed results of relationship between CSR and CFP, additionally no research has studied this relationship in M&A events, we test the relationship between CSR bidders and both their short-run and long-run stock abnormal return to investigate the impact of CSR’s implementation on their M&A activities in our study.

For bidders’ short-run stock performance, since signaling theory has been applied in the M&A activities for a long time, we assume that CSR can also function as a signal due to their better reputation(McGuire et al. (1998); Robinson et al. (2008)) and insurance effect(Godfrey et al. (2009)). Thus, we hypothesize the following:

Hypothesis 1 CSR bidders will experience more positive stock performance than

non-CSR bidders during announcement period.

For bidders’ long-run stock performance, based on Roll (1986)’s hubris hypothesis, Barnea and Rubin (2010)’s self-interested hypothesis and agency hypothesis, we expect the bidding firms’ managers will be overconfidence and overpay when buying targets and the market will overreact during announcement period owing to bidding firm managers’ overconfidence and investors’ over optimism, while in the long run, the managers’ self-interest behaviors and value-destroyed acquisitions will reflect on the poor performance. Thus we expect the following:

Hypothesis 2 CSR bidders will underperform non-CSR bidders in the long run.

8

III. Data

The data of mergers is collected from Securities Data Corporation’s (SDC) U.S.

Mergers and Acquisitions (M&A). We choose 40857 M&A events whose bidders are traded on the NYSE, AMEX and NASDAQ over the period of 1991/01/01 to 2010/12/31. Further we delimit the sample with the following conditions:

1. To avoid double counting the long-run stock returns, multiple takeovers within 3 years by the same acquirer should be excluded from the sample.

2. Acquirers and targets’ relevant stock market data should be covered by the Center for Research in Security Prices (CRSP).

3. CSR firms are covered by the Morgan Stanley Capital International Index (MSCI KLD 400 social index).

In order to get the sample as large as possible, we use the acquirer’s book value as a proxy for its firm size if we can’t find the market value, and use the deal value or the target’s book value as a proxy for its firm size. After these cut steps above, there are 4,527 acquisitions left in the whole sample, where 99 acquisitions are announced by CSR firms and 4,428 acquisitions by non-CSR firms.

Because our main concern is the effect of CSR on acquirer’s performance post M&A, we should retain the sample as large as possible. Since we sacrifice a lot of observations for the targets’ firm size (lots of targets are private and have no market values), we also present results of another sample (with total 6,050 observations, including 145 CSR firms and 5,905 non-CSR firms), in which the CSR firms are maintained as much as possible while targets’ size are missing. In this study, the results described in our text are dominated by the matching sample with total 4,527 observations, for the reason that all variables have their values.

9

Table 1 shows a breakdown of the sample of M&As by year. We report by year (1991-2010) the number of M&A events, aggregate deal value, and mean deal value.

The year where M&As happen most frequently is 1996 with 361 events, and the year where M&As happen least is 2010 with 129 events. Observation of CSR firms adopting M&A deals is much lower than non-CSR firms, though their mean deal value is greater than non-CSR firms, especially in 2005 with 2,111.38 million mean deal value for only 5 CSR bidding firms.

Table 2 reports the summary statistics of the firm characteristics for the full sample by both CSR firms and non-CSR firms. In this table, we define the free cash flow as OANCF-CAPX-DV, where OANCF is cash from operating activities, CAPX is capital expenditures, and DV is cash dividend. From the whole sample, we can see that CSR firms seem to be in a much larger scale than non-CSR firms. The mean total asset of CSR firms is 2,410.38 million while non-CSR firms’ is 585.25 million. The mean book value is 1,100.16 for CSR firms versus 161.14 for non-CSR firms. The mean market value is 4,155.62 for CSR firms versus 482.94 for non-CSR firms. Noting the deal value for CSR firms in Table 1 seems to have large fluctuations, we test the difference of deal value between CSR and non-CSR firms and find it significant (mean: 292.50;

t-value:11.08).

Moreover, since we use variables B/M ratio and relative size (target-to-bidder’s firm size) in our subsequent analysis, we wish to know these variables’ statistic descriptions deeply. The mean value of B/M ratio is 0.61, 0.42 and 0.61 in our full sample, CSR sample and non-CSR sample respectively; and the mean value of relative size is 0.60, 0.56 and 0.60, respectively. For CSR bidders, B/M ratio seem to be lower and the relative size seem to be larger than common bidders. As (Rau and Vermaelen (1997)) argue, glamour firms (those with low B/M ratio) underperform value acquirers

10

(those with high B/M ratio) in the long run. We control these two variables in our subsequent regression analysis.

Table 3 reports the summary statistics of the firm characteristics for the “matching sample” by both CSR firms and non-CSR firms. The matching sample includes 84 CSR bidders and 84 non-CSR peers, and each market value of non-CSR peers is between the range of 70% to 130% of the corresponding CSR firm’s market value. The total assets, book value, market value and free cash flow of the bidding firms reveal no significant difference between CSR and non-CSR firms, but the relative size is significantly larger for CSR bidders than for non-CSR bidders (mean: 0.16; t-value: 2.07). Although not reported in Table 3, we still test the difference of the deal value within CSR and non-CSR bidders and find no significance difference (mean: 99.31; t-value: 1.01). This finding rules out the possibility that there’re difference between the deal values between CSR bidders and their non-CSR peers.

IV. Methodology

1. Short-run Announcement Stock Performance

To test the market reaction to the merger announcement event, we define window (0,0) as the announcing date and follow the previous research’s event study methodology to use the Fama and French (1993) three-factor model for window (-1,-1) (0,0) (+1,+1) (-1,+1) (-3,+1), to capture the immediate reaction to the acquisition. The model is:

R

_jt

= 

_j

R

_mt

+ s

_j

SMB

_t

+ h

_j

HML

_t

+ 

_jt

,, (1)

11

where Rjt is the stock return of the j^th firm on day t; Rmt is the stock return of a market index on day t; SMB_t is the average small market-capitalization portfolios’ return minus the average three large market-capitalization portfolios’ return; HMLt is the average two high book-to-market equity portfolios’ return minus the average two low book-to-market equity portfolios’ return; jt is a random variable which is expected to be zero, and is assumed to be uncorrelated with R_mt, uncorrelated with R_kt for k ≠ j, not autocorrelated, and homoskedastic. j is a parameter that measures the sensitivity of Rjt

to the market index’s excess return; sj measures the sensitivity of Rjt to the difference between small and large capitalization’s stock returns; and hj measures the sensitivity of

R

jt to the difference between value and growth firm’s stock returns. The estimation period is from 20 days pre the announcement to 250 days post the announcement. We define the abnormal return for the stock return of the j^th firm on day t as:

𝐴𝑅_𝑗𝑡= 𝑅_𝑗𝑡− (𝛼̂ + 𝛽_j ̂ 𝑅_{j 𝑚𝑡}+ 𝑠̂𝑆𝑀𝐵_{j 𝑡}+ ℎ̂ 𝐻𝑀𝐿_{j 𝑡}), (2) where 𝛼̂ , 𝛽_j ̂, 𝑠_j ̂ and ℎ_j ̂ are estimations of ordinary least squares of _j j, j, s_j and h_j.

To test the market reaction to the merger announcement event, we follow the previous research’s event study methodology to use the cumulative average abnormal returns (CAAR) for the (-1,+1) (-3,+1) (-20,+1) announcement period, to capture the immediate reaction to the acquisition. The average abnormal return (AAR) in day t is:

𝐴𝐴𝑅_𝑡 = _𝑁¹ ∑^𝑁_𝑗=1𝐴𝑅_𝑗 (3) where N is the number of bidding firms. The cumulative abnormal return (CAR) is defined as following:

𝐶𝐴𝑅 = ∑^b_𝑡=𝑎𝐴𝐴𝑅_𝑡 (4) where t=0 is defined as the announcement day.

12

2. Long-run Stock Performance

We use cumulative abnormal returns (CAR) and buy-and-hold abnormal returns (BHAR) as our models to test the long-run (1, 2, 3-year) stock price of the bidding firms.

The BHAR model is:

𝐵𝐻𝐴𝑅_𝑇 =_𝑁¹∑^𝑁_𝑖=1[∏^𝑇_𝑡=0(1 + 𝑅_𝑖,𝑡) − ∏^𝑇_𝑡=0(1 + 𝑅𝑏𝑒𝑛𝑐ℎ𝑚𝑎𝑟𝑘,𝑡)] (5) where Ri,t is the monthly raw return for bidding firm i in month t, and Rbenchmark is the matching non-CSR firms’ monthly raw return for month t.

For the purpose of caution, we also provide long-run performance results using Fama and French three-factor model and Carhart four-factor model in the subsequent tables. The Fama and French three-factor model is:

R

p,t

- R

ft

= 

p

+ 

p

(R

mt

- R

ft

) + s

p

SMB

t

+ h

p

HML

t

+ 

p,t

, (6)

where Rp,t is the portfolio p’s return in month t; Rft is the one-month T-bills’s return in month t; Rmt is the market index’s return in month t; SMBt is the difference in the returns on a portfolio of small and big stocks in month t; and HMLt is the difference between portfolios of high to low book-to-market stocks in terms of the returns accrued in month

t and 

_p,t is the error term for portfolio p in month t.

The Carhart four-factor model is:

R

_p,t

- R

_ft

= 

_p

+ 

_p

(R

_mt

- R

_ft

) + s

_p

SMB

_t

+ h

_p

HML

_t

+ p

_p

PR1YR

_t

+ 

_p,t

, (7)

where Rp,t, Rft, Rmt, SMBt, HMLt and



p,t are defined as in the Fama and French three-factor model, PR1YRt is the difference in the returns of a portfolio of prior-year high return stocks and prior-year low return stocks in month t.

13

3. Cross-section Analysis

The regression model is set up to test the degree to which CSR will affect the bidder’s abnormal returns. We select firm characteristics and factors which have been proved affecting acquirer’s stock performance as control variables, and create the CSR dummy as our key dependent variable to form the regression model. Noting the variance between CSR and non-CSR firms’ characteristics, we use bidder’s market value as a proxy to control for firm size in our regression model. In addition, there are many factors which have been identified to influence bidder’s stock return (bidders’

B/M ratio, target to bidder’s relative size, diversifying acquisition, target’s public status and payment method) and we use these factors as our control variables. The regression model’s variables are defined as follows:

Dependent variable Acquirer return

We use the bidders’ cumulative abnormal return (CAR) during the (-1,+1) (-3,+1) (-20,+1) announcement period, as our dependent variable. The bidders’ CAR is calculated as in equation (1)-(4).

Independent variables CSR dummy

To test whether CSR bidders outperform non-CSR bidders, we create a dummy variable CSR dummy, as our main concern. The CSR dummy equals 1 if a bidder is covered by the MSCI KLD 400 social index, else equals 0.

14

Firm Size

We use log of a bidder’s market value as a proxy for its firm size.

B/M Ratio

Raghavendra Rau and Vermaelen (1998) claim that bidders with low B/M ratio lose significantly by 22.8% while high B/M ratio bidders win significantly by 14.45%.

Those low-B/M ratio-firms are often called ‘glamour firms’ and tend to overpay in M&As (Shleifer and Vishny (2003); Brau et al. (2012)). In this study, we define B/M

Ratio as log of the ratio of bidder’s book value to their market value.

Relative Size

Kooli, Kortas and L’Her (2003) state that greater relative size provides the target firm greater bargaining power. If such is the case, the target size relative to acquirer size is bigger, the acquirer will capture less profit. Kusewitt, JR (1985) also find similar results suggesting that relative size is negatively related to acquirer’s performance. We define Relative Size as the target’s market value divided by the acquirer’s market value.

Following Rosen (2006)’s research, for those targets whose market value are missing, we use the price paid in the acquisition as a proxy for it. If the price paid still can’t be found, we use the target’s book value to calculate its firm size.

Diversifying acquisition

Jensen (1986) has claimed that diversified mergers are less likely to succeed since acquiring firms are not familiar with the target industry. In Agrawal, Jeffe and Mandelker (1992)’s paper, they find that acquirer in non-diversifying acquisitions will suffer worse underperformance than diversifying acquisitions, which is contrast with popular belief. In this study, we compare the acquirer and target’s first two digit of the 4-digit SIC code reported by SDC to measure whether an acquisition is a diversifying acquisition. If bidder and target’s first two digit SIC codes are different, then we define

15

the event as a diversify acquisition since bidder and target are in different industries.

The dummy variable Diversify equals 1 if the acquirer firm is in different industry with the target firm, else equals 0.

Target Public Status

It has been proved that bidders have positive returns when buying private or subsidiary firms, and significantly negative returns when buying public firms, since the acquirer will argue more liquidity discount for the private firms’ lack of liquidity and difficulty to know the value (Fuller et al. (2002); Chang (1998)). In this study, we category the target’s public status into three dummy variables: Public Target, Private

Target and Subsidiary Target, where Public Target is the dummy variable that equals 1

if the target firm is public, otherwise equals 0; Private Target is the dummy variable that equals 1 if the target firm is private, otherwise equals 0; and Subsidiary Target is the dummy variable that equals 1 if the target firm is a subsidiary firm, otherwise equals 0.

Payment Method

In Fuller et al. (2002)’s study, they find evidence on the payment method affecting acquirer’s return. Their result demonstrates that for public targets, acquirer will have significantly negative returns when they use stock as payment method, while for private firms, acquirer will have higher returns for stock offers relative to cash offers. If a bidder use cash 100% to buy a target firm, then we define its payment method as Cash payment which equals 1, else equals 0. If a bidder acquirers a target firm all by stock, we define its payment method as Stock Payment which equals 1, else equals 0. If a bidder uses combination of the two, we regard its payment method as Mix Payment, which equals 1, else equals 0.

Finally we incorporate variables mentioned above, to build the final regression model. The model is:

16

Acquirer return

i,t = β0i,t + β1 CSRi,t

+ β

2 log of Firm Sizei,t

+ β

3 log of B/M Ratioi,t + β4

Relative Size

_i,t

+ β

₅ Diversifying_i,t +β₆

Public Target

_i,t

+β

₇

Subsidiary Target

_i,t

+β

₈

Stock Payment

i,t

+β

9

Mix Payment

i,t

+ ε

i,t, (8) where i is the bidding firm i and t is the announcement period t.

V. Results

1. Short-run Announcement Stock Performance

Table 4 reports the cumulative abnormal returns of the full sample of the acquiring firms around the announcing day. For all the windows (-1,-1) (0,0) (+1,+1) (-1,+1) (-3,+1) (-20,+1), CSR bidders outperform non-CSR firms. For CSR firms, the CARs are significantly positive for window (0,0) (+1,+1) (-1,+1) (-3,+1), insignificantly positive for window (-1,-1) (-20,+1). For non-CSR firms, the CARs are significantly negative for window (-1,-1), insignificantly negative for window (+1,+1) (-1,+1) (-3,+1) (-20,+1) and insignificantly positive for window (0,0). This result corresponds to Fuller et al.

(2002)’s statement that on average, bidders in merger and acquisition’s announcement gain a zero abnormal return, and acquirers do not lose necessarily (Andrade et al.

(2001)). It is noteworthy that CSR firms outperform non-CSR firms significantly by 1.97% for window (+1,+1) , 3.67% for window (-1,+1), 3.69% for window (-3,+1) and 4.79% for window (-20,+1). Fuller et al. (2002) have mentioned that “while bidder returns are on average small, there is a tremendous variation in returns and many bidders are trying to be one of the winning firms.” Given the significant difference between CSR and non-CSR bidders, we claim that CSR firms win from the numerous

17

bidders.

Table 5 reports the cumulative abnormal returns for bidding firms at the announcement period (-1,+1), grouped by 4 panels. Panel A contains the results for the matching sample by payment methods. For common bidders, takeovers by cash payment usually outperform those using stock (Travlos (1987)). However, in our matching sample, we find bidders using cash payment underperform bidders using stock or mix payment, both in the CSR group and non-CSR group. For non-CSR firms, there is a significant 2.53% abnormal return using cash compared to a significant 3.46% using stock, and for CSR firms, the abnormal return using cash is 2.75%. In addition, CSR bidders using mix payment underperform their non-CSR peers by 0.56% insignificantly.

Panel B contains the results for the matching sample by diversify dummy.

Non-CSR firms seem to make diversify takeovers more frequently than CSR firms. In the non-CSR group, the cross-industry acquisitions gain a significant 5.07% versus to the a significant 2.42% abnormal return by acquisitions in the same industry. This result contradicts to (Jensen (1986))’s argue that diversified mergers are less likely to succeed since acquiring firms are not familiar with the target industry.

Panel C reports results by target public status. For both CSR and non-CSR groups, takeovers with publicly traded targets outperform private targets and contradicts to (Faccio, McConnell, and Stolin (2006))’s result that acquisitions of unlisted targets will win while acquisitions of listed targets will lose. We repute that for these takeovers with larger bidder size, investors treat them different from common takeovers, thus give different results.

Finally, results on Panel D also suggest some distinctions between CSR and non-CSR bidders. Takeovers by CSR firms with higher relative size will underperform non-CSR firms, but for those takeovers with lower relative size, the result is opposite.

18

Since greater relative size provides the target firm greater bargaining power (Kooli et al.

(2003)), it is more difficult for bidder to buy a relatively larger target. However, in our matching sample, both CSR and non-CSR bidders give the opposite answer. Our result show that CSR bidders for relative size greater than medium gain a significant abnormal return by 3.12%, greater than those for relative size less than medium with a significant 2.99% abnormal return.

2. Long-run Stock Performance

Table 6 reports the long-run stock return post M&A for both CSR bidders and non-CSR bidders using the BHAR measure. We find that CSR bidders underperform their non-CSR matching firms insignificantly for all the three holding periods. The CSR bidder’s BHAR is an insignificant negative 12.98% than their non-CSR matching firms for holding 1 year, insignificant negative 10.24% for holding 2 years, and insignificant negative 0.81% for holding 3 years.

However, in Table 7, we can see CSR bidders’ long-run stock returns insignificantly positive for all the three holding periods under Fama and French three-factor model and Carhart four-factor model using the OLS and WLS methods. The weight is defined as 1 divided by the volatility square. Only when holding for 3 years using WLS method can we find CSR firms perform significantly positive (0.0050% and 0.0063%). We suspect there’re two reasons for these confused results using different models. First is the number of samples are limited. The data period is from 1991/01 to 2010/12. In order to calculate the 3-year stock return, the event month period is from 1992/01/ to 2007/12 and the sample number is down to 77. The second is that CSR firms may have specific characters these two models can’t detect. This unresolved problem provides direction for future research as well.

19

In Table 7, we also test the robust t-value. The robust t-value is 0.28, 0.37 and 1.14 for CSR bidders and 1.90, 2.15 and 2.46 for non-CSR bidders under the Fama and French three-factor model; and is 0.37, 0.56 and 1.54 for CSR bidders and 1.87, 2.38 and 2.89 for non-CSR bidders under the Carhart four-factor model.

3. Cross-section Analysis

Table 8 presents regression results of factors influencing the acquirers’ announcing returns on announcement period (-1,+1). After controlling for log of firm size, log of B/M ratio, relative size, target public status dummy, payment method dummy and diversify dummy, we find a significant positive relation between CSR dummy and the bidders’ announcing returns. In the matching sample (totally 168 observations with 84 CSR firms and 84 non-CSR peers), we find CSR dummy a significant positive relation (0.0388) with bidder’s announcement period abnormal return. The CSR dummy also shows a significant positive influence (0.0230) on bidder’s stock price when we use the full sample (4,527 observations bidding firms with 112 firms missing). Our results indicate that investors really favor for CSR firms in the M&A announcement.

Moreover, given the control variables not significant in our matching sample, we offer the regression result for the full sample with 4,336 observations and find all control variables be significant except the Diversify dummy. The Private Target dummy

Moreover, given the control variables not significant in our matching sample, we offer the regression result for the full sample with 4,336 observations and find all control variables be significant except the Diversify dummy. The Private Target dummy

在文檔中 投資者是否對併購案中的企業社會責任公司區別對待？ (頁 12-0)