4.1. Baseline regression
Results of the regressions are reported in Table III. Five models are run separately for total institutional ownership and each of the four identified investor types. Similarly to previous research, no significant relationship between total institutional ownership and
16
announcement period returns was revealed. The coefficient estimate for institutional ownership is 0.08, which is economically and statistically insignificant. Therefore no proof in support of Hypothesis 1 was found.
The coefficient estimate for investment advisor ownership is -1.63 and it’s statistically insignificant. Results are also insignificant in the regression estimating bank and insurance company influence on the M&A announcement period CAR, with coefficient equal to -10.12. On the other hand, coefficient estimates are significant for both hedge fund and pension fund ownerships; the coefficients equal to 6.53 for hedge funds and -23.48 for pension funds. This is a very remarkable difference economically, suggesting that hedge funds are better monitors in preventing managers from making bad acquiring decisions. The results for bank and insurance company, pension fund and hedge fund ownerships are very similar to those presented by Fung and Tsai (2012).
However, in case of investment advisors no consistent and significant relationship was found.
Extant literature (Wahal 1996; Karpoff, Malatesta and Walking 1996; Nesbitt 1994;
Smith 1996) shows mixed results regarding pension funds effectiveness as monitors.
Murphy and Van Nuys (1994) and Romano (1994) discuss that pension funds are not effective monitors because of the agency problems within funds themselves. However, there is little explanation of possible negative relationship between pension fund ownership and company performance. This paper suggests that pension funds, by investing in companies and resigning from monitoring them, strengthen company managers’ position. Presence of big investors who do not investigate into company
17
decisions might be an empowering factor, causing managers to be more likely to proceed with acquisitions that are destroying shareholding value. As prior research by Woidtke (2002) and Del Guercio and Hawkins (1999) has shown, pension funds are heterogeneous monitor agents, too. The negative effect might also be associated with activist public pension funds, which are thought to be motivated more by political or social influences than by firm performance (Woidtke, 2002).
The positive influence of hedge funds and negative influence of pension funds on CAR are consistent with previous papers by Chen, Harford, and Li (2007) and Ferreira and Matos (2008) who suggest that more independent investors can be associated with higher shareholder value. Thus, the results support Hypothesis 2; institutional investors are heterogeneous in influencing companies’ performance as measured through market reaction on M&A announcement.
As for control variables used, coefficients and significance do not vary notably between the five models. Results show no significant influence of Tobin’s Q on acquirer returns. There is a negative coefficient for indicator of stocks financed in other ways than cash only, however it’s not significant. Firm size is significantly and negatively correlated with acquirer returns, which might be a sign of managerial hubris. The coefficient for leverage is significant and positive, which means that higher debt levels are indeed positive incentive for managers to make better decisions. Coefficient for free cash flows is also significantly positive, which is opposite to what Jensen’s (1988) free cash hypothesis suggests. Additionally, acquirer returns are significantly influenced by the size of the transaction normalized by acquirer market value.
18
Results presented in this paper are consistent with those presented by Moeller, Schlingemann, and Stulz (2004) and Masulis, Wang, and Xie (2007), with the exception of leverage and free cash flows, which were previously reported insignificant. Results hold and remain analogous when year dummies are added to the models.
4.2. High Institutional Ownership Indicator
Additionally, to measure the impact the high level of institutional ownership has on acquirer returns, an indicator of high level institutional ownership is used. First, the high level institutional ownership indicator is created that equals 1 when institutional ownership of a given company is greater than median institutional ownership in the sample, and equals 0 otherwise. Regression includes the same independent variables as in baseline regression, including year dummies. Results, as presented in Table IV, differ from the baseline regression. Results for hedge funds and pension fund ownership became insignificant; coefficient remains negative for pension funds, and became negative for hedge funds. Notably, results for banks and insurance companies’
ownership became significant at the 10% level, with the negative coefficient of -0.83. A possible explanation for inconsistent results for hedge fund and pension fund ownership might be that the median, as a point of division of the sample into high and low ownership indicator, does not divide the sample into two groups, one in which institutional owners posses big enough holdings to be in position to effectively influence firm’s management, and one with holdings that are not big enough. Indeed, the sample distribution is positively skewed3. Thus high level institutional ownership indicator might
3 Skewness equals to .113 for total ownership, .505 for investment advisors, 2.305 for hedge funds, 1.569
19
include companies with relatively low institutional ownership levels if assigned by the median. Therefore in the second approach high level institutional ownership indicator equals to 1 when institutional ownership of a given company is greater than the 80th percentile of institutional ownership in the sample. In this approach, the coefficient for total institutional ownership becomes noticeably higher than in previous regressions, however still remains insignificant. Results also remain insignificant for investment advisors ownership. Results become significant at 5% level for both hedge fund and pension fund ownership, with coefficients equal to 1.20 and -1.15 respectively. The magnitude of coefficient for banks and insurance companies’ ownership becomes slightly lower and insignificant. Those results are consistent with baseline regression and support hypothesis 2. This outcome also suggests that the relationship between institutional ownership and acquirer returns become more significant in the subsample of acquirers with higher level of institutional holdings. The results of high institutional ownership indicator approach are consistent with the first model and with previous study by Fung and Tsai (2012).
4.3. Subsample
Consequently with outcomes from the previous models, regressions for subsamples, which consist of acquirers which institutional ownership was higher than median institutional ownership for the whole sample, are run. Each of the subsamples consists of 1107 M&A announcements. Results are presented in Table V. Most remarkably, a coefficient of -5.2 becomes significant for investment advisors ownership. It is consistent
for pension funds, and 3.705 banks and insurance companies.
20
with hypothesis 2, further documenting heterogeneity of institutional investors.
However negative coefficient is contrary to the expected results. It is also contrary to Fung and Tsai’s (2012) results which documented a positive relation between investment advisors ownership and firm performance through capital expenditures. The magnitude of coefficient and significance increases for hedge fund and pension fund ownership.
Coefficient remains insignificant for total institutional ownership and for banks and insurance companies.
In the subsample regressions, many control variables become insignificant, including firm size and leverage in all models, as well as free cash flows for total, hedge fund and pension fund ownership, and relative deal size for pension fund ownership.
Moreover consideration offered becomes significant for hedge funds and banks and insurance companies.
In general, results of the models support hypothesis 2, showing that institutional investors are diverse monitoring agents. Results indicate that hedge funds are positively associated with higher acquirer returns. Meanwhile investment advisor, pension fund and bank and insurance company ownership has a negative relationship with acquirer returns. No proof in support of Hypothesis 1 was found, thus there is no relation between total institutional ownership and acquirer returns exposed.
The results of the subsample regression are consistent with Fung and Tsai (2012) with the exception of category of investment advisors that showed opposite relationship than documented previously.