1.4 Methodology 1 Mergers Sample
1.5.4 Cross-Sectional Analyses
We learn from section 7.2 that any single set of results based on a specific model should be interpreted within the context. This is especially true if further regressions are estimated to find out the factors that determine the CARs in response to bank merger announcement. In a regression, the estimated CARs are used as the dependent variable. Hence such regressions may become meaningless if the initial abnormal return estimates are not robust. Our model (1.1) is regressed using weighted ordinary least squares to find the coefficients based upon the sample collected. The only correction to be made is the use of White Heteroskedasticity-Consistent Standard Errors (1980) and Covariance in an effort to control for heteroskedasticity. All variables are linear in parameters except for size, which has been transformed using natural log in an effort to stay consistent with the work of Fama and French (1992).
We report the results of cross-sectional regression analyses in Table 1.11 (targets), Table 1.12 (bidders) and Table1.13 (combined). Following Houston and Ryngaert (1994), we run regressions with and without separate year dummies that are included to control for any changes over time that are not captured in the independent variables. We observe that the relative size seems to be statistically significant in these 3 tables. The coefficient for relative size is negative in Table 1.12. This means that the relative size does affect bidders’ returns negatively (i.e. relative size is associated with lower bidder returns). Also, it suggests the market reacts more unfavorably when the relative size increases. Accordingly, the bidding bank experiences lower abnormal returns, indicating low synergy or diseconomies of scale. Also, this could support the
overpayment hypothesis in sense that the bidder is willing to pay a higher premium in expecting of potential synergy resulting from merger.
Table 1.11
Cross-sectional regression results for target firms
Independent Variable
Regression 1 Regression 2
Intercept 0.246 0.361
(0.74) (2.78)***
Log (ME of target/ME of bidder) 0.011 0.0198
(2.02)** (2.34)**
Log (BM of target/BM of bidder) -0.042 -0.023
(-1.34) (-1.62)
Payment in Cash (dummy) 0.056 0.196
(3.12)*** (2.70)***
Payment in Stock (dummy) -0.0303 -0.210
(-1.61) (-1.04)
Interstate Merger (dummy) -0.135 -0.187
(-0.524) (-.68)
Number of Bidders 0.018 0.192
(1.36) (1.66)
Year Dummies No Yes
Adjusted R2 0.158 0.212
***, ** , and * denote significant at 1 %, 5% and 10% levels respectively.
As for the target and combined firms (Table 1.11 and 1.12), the positive coefficient on the relative size indicates that the cumulative returns for the target increase significantly as the target size increases relative to the bidder size. This result is consistent with the economics of scale hypothesis, suggesting the larger the target relative
to the bidder, the greater the abnormal returns to target and combined firms. Indeed, the economics of scale argument suggests that the coefficient should be positive. Put differently, this observation supports the hypothesis that one motivation of bank mergers is the economics of scale or economics of scope.
Table 1.12
Cross-sectional regression results for bidder firms
Independent Variable
Regression 1 Regression 2
Intercept -0.136 -0.155
(-0.53) (-0.74)
Log (ME of target/ME of bidder) -0.142 -0.119
(-2.44)** (-3.02)***
Log (BM of target/BM of bidder) -0.034 -0.007
(-0.225) (-0.55)
Payment in Cash (dummy) 0.024 0.028
(2.05)** (2.39)**
Payment in Stock (dummy) -0.161 -0.178
(-.284)** (-3.45)***
Interstate Merger (dummy) -0071 -0.066
(-5.09)*** (-5.38)**
Number of Bidders -0.126 -0.411
(-2.47)** (-2.54)**
Year Dummies No Yes
Adjusted R2 0.29 0.31
Another interesting interpretation to this result suggests that relatively large targets gain significantly larger merger premiums. For bidder returns, the negative coefficient is consistent with positive coefficient on relative size in the target regression if the higher returns to target in relatively large deals hurt bidder firms.
Table 1.13
Cross-sectional regression results for combined firms
Independent Variable
Regression 1 Regression 2
Intercept 0.059 0.086
(2.12)** (2.23)**
Log (ME of target/ME of bidder) 0.008 .039
(5.07)*** (5.42)***
Log (BM of target/BM of bidder) 0.025 0.033
(0.79) (1.42)
Payment in Cash (dummy) 0.008 0.029.
(2.02)** (2.35)**
Payment in Stock (dummy) -0.013 0.059
(-0.21) (1.16)
Interstate Merger (dummy) -0.024 0.013
(-0.83) (0.35)
Number of Bidders -0.032 0.015
(-0.59) (0.19)
Years Dummies No Yes
Adjusted R2 0.221 0.257
***, ** , and * denote significant at 1 %, 5% and 10% levels respectively.
We observe that payment in cash is positive and significant in all regressions. The coefficient on the cash dummy variable is significantly positive for targets, bidders and
combined firms. This result indicates that the market reacts positively to the announcement of mergers that are financed with cash. As for the payment in stock, the coefficient is only negatively significant for bidders. It seems that the rational market participants interpret a cash financed merger as a positive signal of the existing assets of the bidding bank and react accordingly. This implies that the bidder, target and combined receive a positive wealth gain when mergers financed with all cash. These results are consistent with Toyne and Tripp (1999) and Houston and Ryngaert (1994). They report that mergers financed with stock have lower abnormal returns than those financed with cash. However, these results contradict recent studies by Delong (2001) and Becher (2000) who report that the method of payment does not influence returns to combined firms.
It is worth noting that the number of bidders is only statistically significant in Table 1.12. This variable is negatively related to bidders returns. The negative sign of this variable is consistent with the overpayment hypothesis that predicts if more than one bidder bid on the same target, the winning bidder will overpay for the target to get the deal. As a result, the bidder return will be lower in this case. Surprisingly, this variable does not affect target and combined returns. This is inconsistent with our hypothesis that predicts target firm will experience higher returns where there are multiple bidders.
The coefficient for interstate variable is statistically different from zero only in bidder returns (Table 1.12). The results indicate that interstate (diversifying) mergers create negative abnormal returns to bidders. Relative book-to-market ratio variable is never statistically significant. These results indicate that the relative book-to-market ratios are not confirmed as relevant factor in our sample. Finally, the yearly dummies
appear to provide additional explanatory power when included in the regression specifications.