Chapter 4. Empirical Result
4.3. Long-term Perspective
國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
34
4.3. Long-term Perspective
In section 4.2, we examine how stock market reacts to an acquisition for
innovation, and we have seen my hypothesis under short-term event windows is not
supported by the result. Instead, it once again confirms the existing argument that
acquirers tend to underperform. In this section, we focus on the long-term perspective
of wealth effect.
As we know, firms need further integration in every type of business
combination, including M&As; and it may be time-consuming to integrate two firms
with divergent cultures and capacities. Innovation capacities should not be exception.
For instance, after the ownership of target’s patents is transferred, it takes times for
acquirers to get familiar with them so that new designs could be accomplished.
Likewise, while taking over R&D team, it also takes time for acquirers to fit in the
new members of the teams. In this respect, the merits of post-merger integration
should be examined under a long-term horizon.
Table 5 shows the result of my long-term study on post-merger abnormal returns.
As mentioned in section 3.2, the methodology is pretty similar with that I employ in
my short-term study, except that I adopt monthly return data in this section. The
estimate period is (-36, -1), and I examine the abnormal returns under 4 innovation
‧
國立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
35
measures across 5 event windows: (+1, +6), (+1, +12), (+1, +24), (+1, +36), and (+1,
+60). We can see under all event windows, the acquirers bidding innovative targets
enjoy positive abnormal returns, their mean in each group reveal all positive numbers
and collectively outperform their respective peers. In contrast, the abnormal returns of
matching groups are proven to be insignificant; even some of the matching groups,
especially for those whose event periods are less than one year, still suffer from
negative performance (although not significant). Moreover, as I prolong my event
windows to more than two years, the abnormal returns become consistently
significant for various innovation metrics and across all of the event windows. This
finding is quite different from that of short-term study, showing that acquiring strong
targets behaves better than the overall M&A universe. My hypothesis of the long-run
perspective is therefore confirmed.
‧
Table 5 Abnormal returns for long-term event windowsThis table reports announcement and long-term abnormal returns for acquirers under different innovation measures. The strong-target group consists of deals with targets whose innovation metrics are greater than the median of overall sample, while the weak-target group contains the rest. Abnormal returns over (t3, t4) event window are defined in section 3.2.1. Mean value of abnormal returns is presented as Mean; standard deviation is presented as Stdev; p-values are in brackets.
Whole
‧
國立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
37
After studying the result of event study, I want to prove further if the superior
performance of strong-target groups could be attributed to the targets’ innovation
efforts. Therefore I conduct multiple regression analysis to see the direct relationship
between abnormal returns and all sorts of innovation effort. The full model is
discussed in section 3.2.2. I report only one innovation measure at one time in a
model. The regression coefficients and p-values are presented in Table 6.1, Table 6.2,
and Table 6.3 for the event windows of (+1, +24), (+1, +36), and (+1, +60),
respectively. These tables show that firms with better operation, as measured by
pre-acquisition ROA, perform positively; as to asset growth, another dimension of
operating result, the relation is positive, but insignificant. Also, past returns do not
explain future performance; this result coincides with weak-form market efficiency.
However, there is a surprising finding that the tender offer dummy show significant
negative effect on abnormal returns. Here I propose a possible reason for the
interesting finding. The matching of M&A list and patent data in my study acts as a
filter, automatically selecting firms periodically report their innovation efforts.
Though the extent varies, my sample universe consists of firms engaging in
innovation activities. Tender offers are deemed as hostile mean in M&A, more often
than not, the bidders replace the managements after taking over and restructure target
firms in a fierce manner. It inevitably does harm to post-merger integration, which is
‧
國立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
38
especially critical for an innovation-oriented firm, and therefore damage the future
stock price.
Now, go back to the main point of my study, innovation measures. The OLS
regression analysis shows that all innovation measures have positive relation with
abnormal returns, although the relations are not consistently significant. Only average
citation explains abnormal return consistently under the longest event window: (+1,
+60), suggesting average citation being a better measure in the long-run. This finding
is consistent with Zhao’s research (2009) that indicates citation count (Zhao define
citation counts in the same way I define average citation in my study) as the effective
measure for innovation quality, which plays a major role in M&A decision. On the
other hand, the innovation dummy, which stands for merger pair with weak acquirer
and strong target, yields an inconclusive result, with neither apparent pattern of sign
nor significance.
‧
Table 6.1 Relation between Innovation and Future Stock Performance under the event window of (+1, +24), estimated by OLS
In_Measure Patcount Patcount Patcount Citotal Citotal Citotal Citavg Citavg Citavg Citwgtpat Citwgtpat Citwgtpat
Model 1 2 3 4 5 6 7 8 9 10 11 12
‧
Table 6.2 Relation between Innovation and Future Stock Performance under the event window of (+1, +36), estimated by OLS
In_Measure Patcount Patcount Patcount Citotal Citotal Citotal Citavg Citavg Citavg Citwgtpat Citwgtpat Citwgtpat
Model 1 2 3 4 5 6 7 8 9 10 11 12
‧
Table 6.3 Relation between Innovation and Future Stock Performance under the event window of (+1, +60), estimated by OLS
In_Measure Patcount Patcount Patcount Citotal Citotal Citotal Citavg Citavg Citavg Citwgtpat Citwgtpat Citwgtpat
Model 1 2 3 4 5 6 7 8 9 10 11 12
‧
國立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
42
In addition to OLS, I employ Weighted Least Squares (WLS) in my analysis with
the acquirers’ inflation-adjusted size as the weights. As I mentioned in Section 1.1,
R&D activities generally burn money. Firms with bigger size possess relatively
fruitful resources. For a large acquirer, it is reasonable to suppose it enjoys abundant
funds, integrated supply chain or even well-recognized brand name that enable the
acquirer to take better advantages of the target’s innovation capacity. With regard to
this, I use acquirers’ size before announcement (adjusted for inflation) as the weights
in WLS procedure, granting larger weights for deals with large acquirers. In this part,
I only present the coefficients and p-values of innovation variables in Table 7, results
of the controlling variables are alike to those of OLS. The positive relation between
innovation measures and abnormal return is once again confirmed by WLS procedure,
which reveals a set of all-positive coefficients. Besides, the outcome is similar to OLS
procedure in that average citation has significant influence on abnormal return under
the longest event window: (+1, +60). Further, the positive relation in other measures
is proven to be evident. The positive coefficients of patent counts, total citations and
citation-weighted patent counts are statistically significant. This provides an
intriguing insight that innovation does interact positively with post-merger stock
performance, especially for large acquirers.
‧
Table 7 Relation between Innovation and Future Stock Performance, estimated by WLSTable 7 reports the cross-sectional regression estimates from regressing abnormal return (AR) on selected independent variables for 605 announcements of completed M&A deals. ARs are from the (+1, +24) event window. The period for this analysis runs from 1980 to 2006. Year fixed effects are included in all specifications. The p-values are reported in brackets. See section 3.1.2 and section 3.1.3 for variable definitions. *** denotes significance at 1% level; ** denotes significance at 5% level; *denotes significance at 10% level.
Panel A, Event Window of (+1, +24)
In_Measure Patcount Patcount Patcount Citotal Citotal Citotal Citavg Citavg Citavg Citwgtpa Citwgtpa Citwgtpa
Model 1 2 3 4 5 6 7 8 9 10 11 12
‧
國立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
44
Table 7 Relation between Innovation and Future Stock Performance, estimated by WLS(continued)
Panel B, Event Window of (+1, +36)
In_Measure Patcount Patcount Patcount Citotal Citotal Citotal Citavg Citavg Citavg Citwgtpa Citwgtpa Citwgtpa
Model 1 2 3 4 5 6 7 8 9 10 11 12
Patcount_T 0.0010 0.0010
(0.028) ** (0.024) **
Citotal_T 0.0000 0.0000
(0.017) ** (0.014) **
Citavg_T 0.0019 0.0026
(0.316) (0.184)
Citwgtpat_T 0.0004 0.0004
(0.017) ** (0.015) **
Dat -0.0941 -0.0133 -0.1676 -0.1189 -0.1758 -0.1114 -0.1554 -0.1171
(0.601) (0.940) (0.403) (0.551) (0.241) (0.432) (0.477) (0.591)
Controlling Variable Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes
Year Dummies Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes
‧
國立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
45 Table 7 Relation between Innovation and Future Stock Performance, estimated by WLS(continued)
Panel C, Event window of (+1, +60)
In_Measure Patcount Patcount Patcount Citotal Citotal Citotal Citavg Citavg Citavg Citwgtpa Citwgtpa Citwgtpa
Model 1 2 3 4 5 6 7 8 9 10 11 12
Patcount_T 0.0005 0.0006
(0.147) (0.100) *
Citotal_T 0.0000 0.0000
(0.125) (0.140)
Citavg_T 0.0062 0.0069
(0.000) *** (0.000) ***
Citwgtpat_T 0.0002 0.0002
(0.125) (0.142)
Dat -0.1587 -0.1094 0.0709 0.0962 -0.1853 -0.0102 0.1379 0.1575
(0.272) (0.439) (0.660) (0.549) (0.119) (0.927) (0.437) (0.373)
Controlling Variable Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes
Year Dummies Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes
‧
國立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
46
4.4. Calendar Time Portfolio
In the previous section, I show the strong long-term performance of acquirers
bidding innovative target and further prove the significant relation between
innovation and abnormal return. This section aims to provide practical insights in
investment for the public investors. I want to study whether or not investing in
acquirers buying external innovation yields abnormal returns via employing
calendar time portfolio. My investment horizon starts from January 1980, ends at
December 2006. I create a portfolio composed of acquirers making M&A
announcements during the past 24 months and re-form the portfolio month after
month. The portfolio can be constructed with either 2 types of weights:
equally-weighted (EW) and value-weighted (VW). By comparing the time-series
portfolio returns with expected returns estimated by Fama-French 3-factor model,
the significance of abnormal returns are examined. The result is reported in Table 8.
Based on prior section, I suppose that abnormal returns of strong-target groups
being positive and outperform their respective matching groups. In the calendar time
portfolio analysis, however, strong-target group do not indicate consistently positive
abnormal returns under patent count, total citation count and citation-weighted
patent count, nor do they reveal greater coefficient in magnitude.
‧
國立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
47
On the other hand, there is one innovation measure strongly confirming my
hypothesis: average citation, which presents positive and significant abnormal
returns for the acquires bidding innovative targets at 1% level of significance. In
contrast, the weak-target matching group commonly shows lower alpha in
magnitude. Moreover, the strong-target EW portfolio yields superior performance
relative to the weak-target peer at 10% level of significance, measured by the
difference between abnormal returns; the difference is also positive for VW portfolio
(though not significant). This phenomenon responds to the finding in previous
section that average citation positively relates to acquirers’ post-merger performance
under long-term horizon, and again confirms that average citations has close relation
with M&A performance.
‧
國立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
48 Table 8 Abnormal returns estimated via calendar time portfolio approach
Table 8 reports the time-series regression estimates from employing calendar time portfolio from 1980 to 2006. The p-values are reported in brackets. See section 3.1.2 and section 3.1.3 for variable definitions. *** denotes significance at 1% level; ** denotes significance at 5% level; *denotes significance at 10% level.
Patcount
Citotal
Weak (1) Strong (2) (2)-(1)
Weak (1) Strong (2) (2)-(1)
EW VW EW VW EW VW EW VW EW VW EW VW
Alpha 0.1402 0.4171 0.1084 0.4842 -0.2346 -0.1582 Alpha 0.0421 0.4424 0.2012 0.4884 -0.3209 -0.3123
(0.428) (0.034) ** (0.554) (0.015) ** (0.443) (0.605) (0.796) (0.026) ** (0.312) (0.024) ** (0.242) (0.340)
RmRf 1.1677 0.8995 1.0931 0.7512 RmRf 1.1202 0.8533 1.1513 0.8098
(0.000) *** (0.000) *** (0.000) *** (0.000) *** (0.000) *** (0.000) *** (0.000) *** (0.000) ***
SMB 0.3623 -0.1319 0.3933 -0.2567 SMB 0.4028 -0.0779 0.3413 -0.2364
(0.000) *** (0.015) ** (0.000) *** (0.000) *** (0.000) *** (0.159) (0.000) *** (0.000) ***
HML 0.1774 -0.3170 -0.0787 -0.4294 HML 0.1839 -0.1676 -0.0742 -0.4701
(0.005) *** (0.000) *** (0.210) (0.000) *** (0.001) *** (0.016) ** (0.277) (0.000) ***
Adj. R2 0.7552 0.6526 0.7557 0.5936 Adj. R2 0.7752 0.5908 0.7463 0.6070
‧
國立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
49 Table 8 Abnormal returns estimated via calendar time portfolio approach (continued)
Citavg Citwgtpat
Weak (1) Strong (2) (2)-(1)
Weak (1) Strong (2) (2)-(1)
EW VW EW VW EW VW EW VW EW VW EW VW
Alpha -0.3172 0.0931 0.5684 0.5686 0.5173 0.0516 Alpha 0.0058 0.4277 0.1736 0.5031 -0.3209 -0.3123
(0.053) * (0.651) (0.006) *** (0.008) *** (0.097) * (0.890) (0.973) (0.030) ** (0.373) (0.018) ** (0.242) (0.340)
RmRf 1.0783 0.8680 1.1888 0.8415 RmRf 1.1342 0.8552 1.1573 0.8175
(0.000) *** (0.000) *** (0.000) *** (0.000) *** (0.000) *** (0.000) *** (0.000) *** (0.000) ***
SMB 0.3777 -0.0782 0.3695 -0.2381 SMB 0.4180 -0.0719 0.3458 -0.2373
(0.000) *** (0.176) (0.000) *** (0.000) *** (0.000) *** (0.194) (0.000) *** (0.000) ***
HML 0.3376 -0.1328 -0.2111 -0.4557 HML 0.2312 -0.1648 -0.1208 -0.4686
(0.000) *** (0.067) * (0.003) *** (0.000) *** (0.000) *** (0.018) ** (0.071) * (0.000) ***
Adj. R2 0.7340 0.5683 0.7671 0.6237 Adj. R2 0.7561 0.5904 0.7489 0.5989
‧
國立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
50