The Decoupling of Voting and Economic Ownership (Instituut voor Ondernemingsrecht nr. 88), 4.7 Policy Implications and Conclusion

Policymakers and company managers appear increasingly concerned that institutional investors follow their proxy advisor 's voting recommendations blindly. In this paper, we have analyzed proprietary data from four large fund managers to test the hypothesis that institutional investors allocate the limited resources available for verifying the accuracy of voting recommendations to voting decisions that are likely to have the greatest impact on portfolio performance, using the funds' marginal propensity to deviate from voting recommendations as a proxy. Consistent with our hypothesis, we have found that the funds tend to deviate from voting recommendations relatively more often when they hold a large stake in the portfolio firm, when the firm performs poorly and when the proposal has potentially significant value implications.

We have also studied the effects of a potentially important channel through which the funds may communicate their govemance preferences to their proxy advisor before receiving voting recommendations: the customized voting policy. Our findings suggest that the effects of such voting policies vary per fund. Finally, we have looked at the human resources dedicated to corporate govemance. The data indicates that the average staff member needs to monitor hundreds of portfolio firms, which explains why the funds retain the services of proxy advisors.

From a policy perspective, the key question is what is the optimal level of reliance by institutional investors on proxy voting recommendations? Obviously, by exploiting economies of scale, proxy advisors can help to overcome some of the collective action problems that diversified shareholders are faced with. To the extent proxy advisors collect information that institutional investors would otherwise have to collect themselves, they help to minimize asset management costs incurred by the funds' beneficiaries. Schouten (2010) analyzes proxy advisors in terras of a trade-off between informed voting and independent voting, two distint mechanisms that each increase the probability that a majority of shareholders vote for the option that maximizes shareholder value. Considering the importance of independent voting, there is a case to be made for the statement made by the NYSE Commission on Corporate Governance that even if institutional investors obtain outside proxy advice, they remain responsible for ensuring that their votes are cast only after the exercise of thoughtful judgment. Yet the results presented in this paper caution against further-reaching measures. The funds in our sample appear to make deliberate choices about the resources they devote to different types of voting decisions, choices that presumably they believe to be in the interest of their beneficiaries. Unless this belief is misguided, for example because the funds systematically underestimate the benefits of active monitoring, requiring funds to devote more resources may effectively result in a wealth transfer from the funds' beneficiaries to free-riding minority shareholders in portfolio companies.

Table 1 Summary statistics

The table reports descriptive statistics for voting and proxy voting recommendations. Panel A shows the rate of deviation from proxy voting recommendations. Follow refers to votes cast in line with the proxy voting recommendation, Deviate refers to votes cast differently. Average fund refers to the arithmetic mean of the four fund mean deviation rates. Median fund refers to the median of the four fund mean deviation rates. Panel B reports the same statistics for the full sample. Panel C shows the rate of deviation from management's recommendations. FOR refers to votes cast in line with management's recommendation, AGAINST to votes cast differently. For funds in respect of which we do not have data on management's recommendation, we have assumed that management recommends voting FOR management proposals and AGAINST shareholder proposals, and we have counted ABSTAIN as a vote in deviation from management's recommendation (data on ABSTAIN is reported separately between square brackets). Panel D breaks down the results reported in Panel A by specifying whether the funds deviated from their proxy advisor's recommendation so as to vote with management or against management.

Panel A: Rate of deviation from proxy advisor (fund-by-fund)
	Fund 1		Fund 2		Fund 3		Fund 4		Average fund	Median fund
	Obs.	Mean	Obs.	Mean	Obs.	Mean	Obs.	Mean	Mean	Mean
Follow	17073	97.4%	9997	99.3%	14423	99.7%	34425	88.8%	96.3%	98.4%
Deviate	450	2.6%	72	0.7%	37	0.3%	4361	11.2%	3.7%	1.6%
Total	17523	100%	10069	100%	14460	100%	38786	100%	100%	100%

Panel B: Rate of deviation from proxy advisor (full sample)
	Obs.	Mean
Follow	75918	93.9%
Deviate	4920	6.1%
Total	80838	100%

Panel C: Rate of deviation from management
	Fund 1	Fund 2	Fund 3	Fund 4	Average fund	Median fund	Full sample
	Mean	Mean	Mean	Mean	Mean	Mean	Mean
FOR	86.1%	88.5%	73.6%	92%	85.1%	87.3%	87.0%
AGAINST	13.9%	11.5%	26.4%	8%	14.9%	12.7%	13.0%
	[3.8%]	[1.5%]	[7.5%]	[11.5%]	[6.1%]	[5.7%]	[3.3%]
Total	100%	100%	100%	100%	100%	100%	100%

Panel D: Deviate to vote FOR or AGAINST management
	Fund 1	Fund 2	Fund 3	Fund 4	Average fund	Median fund	Full sample
	Mean	Mean	Mean	Mean	Mean	Mean	Mean
FOR	1.8%	0.3%	0.1%	9.0%	2.8%	1.1%	4.8%
AGAINST	0.7%	0.4%	0.1%	2.2%	0.9%	0.6%	1.3%
Total	2.6%	0.7%	0.3%	11.2%	3.7%	1.6%	6.1%

Table 2 Propensity to deviate and stake size

The table reports descriptive statistics for voting and proxy voting recommendations. Panel A shows the funds' propensity to deviate from their proxy advisors' voting recommendations in relation to value of stake. Stake value refers to the market value of the stake in the portfolio firm as at the date of the shareholders meeting (share prices are obtained from Datastream), relative to the value of the fund's stakes in its other portfolio firms. Average fund refers to the arithmetic mean of the four fund mean deviation rates. Median fund refers to the mediae of the four fund mean deviation rates. Panel B shows the funds' propensity to deviate from their proxy advisors' voting recommendations in relation to size of stake. Stake size refers to the size of the stake in the portfolio firm (expressed as a percentage of outstanding shares, as obtained from Datastream), relative to the size of the fund's stakes in its other portfolio firms. Panel C reports descriptive statistics for the full sample, comparing the mean rate of deviation in respect of portfolio firms in which the size (value) of the stake is smalt relative to the size (value) of stake in the fund's other portfolio firms (the lower quartile) to the rate of deviation in respect of portfolio firms in which the size (value) of the stake is relatively large (the upper quartile). In Panel C, we also test the null hypothesis that the means are equal. *, **, and *** indicate whether we can reject the null hypothesis at a 10%, 5% and 1% significance level.

Panel A: Propensity to deviate and value of stake
	Fund 1	Fund 2	Fund 3	Fund 4	Average fund	Median fund
Stake value	Mean	Mean	Mean	Mean	Mean	Mean
25^th perc. <	2.4%	0.7%	0.3%	11.3%	3.7%	1.6%
25^th - 50^th perc.	2.4%	0.7%	0.3%	10.5%	3.5%	1.5%
50^th - 75^th perc.	2.9%	0.3%	0.2%	11.4%	3.7%	1.6%
> 75^th perc.	3.4%	1.0%	0.3%	11.5%	4.0%	2.2%

Panel B: Propensity to deviate and size of stake

Stake size	Stake size	Stake size	Stake size	Stake size	Stake size	Stake size
25^th perc. <	25^th perc. <	25^th perc. <	25^th perc. <	25^th perc. <	25^th perc. <	25^th perc. <
25^th - 50^th perc.	25^th - 50^th perc.	25^th - 50^th perc.	25^th - 50^th perc	25^th - 50^th perc.	25^th - 50^th perc.	25^th - 50^th perc
50^th - 75^th perc.	50^th - 75^th perc.	50^th - 75^th perc.	50^th - 75^th perc	50^th - 75^th perc.	50^th - 75^th perc.	50^th - 75^th perc
> 75^th perc.	> 75^th perc.	> 75^th perc.	> 75^th perc.	> 75^th perc.	> 75^th perc.	> 75^th perc.

Panel C: Full sample
	Stake size				Stake value
	Obs.	Mean	Std. dev.	Std. err.	Obs	Mean	Std. dev.	Std. err.
25^th perc. <	18570	5.5%	0.227	0.002	16825	6.1%	0.238	0.002
> 75^th perc.	18931	7.9%	0.270	0.002	23099	6.8%	0.252	0.002
Difference of means		2.5%***				0.8%***
t-value		-9.64				-3.13

Table 3 Propensity to deviate and proposal type

The table reports descriptive statistics for votes cast in deviation from the proxy advisor's recommendation. Panel A reports statistics for the mediae fund. We follow the classification of proposals by ISS. For funds in our sample using Glass Lewis as a proxy advisor, we have converted Glass Lewis's classification to ISS's classification. Deviate refers to votes cast differently than recommended by the fund's proxy advisor. Follow refers to votes cast in line with the fund's proxy advisor's recommendation. Panel B compares the meao rate of deviation in respect of routine matters to the rate of deviation in respect of non-routine matters. Routine refers to management proposals classified by the proxy advisor as a routine matter. Non-routine refers to all other management proposals. Panel C reports statistics for the full sample and tests the null hypothesis that the means are equal. *, **, and *** indicate whether we can reject the null hypothesis at a 10%, 5% and 1% significance level.

Panel A: Medium fund
Proposal type	Deviate	Follow
Management proposals	Mean	Mean
Directors Related	1.9%	98.1%
Routine/Business	0.6%	99.4%
Capitalization	1.5%	98.5%
Non-Salary Comp.	1.8%	98.2%
Antitakeover Related	0.3%	99.7%
Reorg. and Mergers	1.7%	98.3%
Preffered/Bondholder	0.0%	100.0%

Shareholders proposals
Directors Related	4.4%	95.6%
Compensations	1.1%	98.9%
Routine/Buseniss	0.8%	99.2%
Health/Environment	15.3%	84.7%
Other/Misc.	11.3%	88.7%
Corporate Governance	1.7%	98.3%
Soc./Human Rights	8.7%	91.3%
Gen Econ Issues	0.0%	100.0%
Social Proposal	9.8%	90.2%

Panel B: Routine vs. non-routine mgmt. proposals (fund-by-fund)
	Fund 1	Fund 2	Fund 3	Fund 4	Average fund	Median fund
Proposal Type	Mean	Mean	Mean	Mean	Mean	Mean
Routine	0.8%	0.4%	0.0%	9.2%	2.6%	0.6%
Non-routine	2.7%	0.8%	0.4%	12.0%	4.0%	1.7%

Panel C: Routine vs. non-routine mgmt. proposals (full sample)
Proposal Type	Obs.	Mean	Std. dev.	Std. err.
Routine	22539	4.8%	0.215	0.001
Non-routine	57096	6.5%	0.246	0.001
Difference of means		1.6%***
t-value		8.64

Table 4 Propensity to deviate and portfolio firm characteristics

The table reports descriptive statistics for votes cast in deviation from the proxy advisor's recommendation. In Panel A, Performance refers to market adjusted return in the 12 months prior to the date of the shareholder meeting, compared to that of other portfolio firms. In Panel B, Firm size refers to market capitalization as at such date (data from Datastream), compared to that of other portfolio firms. Panel C reports statistics for the full sample, and compares the mean rate of deviation in respect of portfolio firms that performed badly relative to the fund's other portfolio firms (lower quartile) to the deviation rate in respect of firms that performed relatively good (upper quartile); the Panel also compares the mean deviation rate in respect of firms that are relatively small (lower quartile) to the rate of deviation in respect of portfolio firms that are relatively large (upper quartile). Panel D looks at the funds' propensity to deviate in relation to firm nationality. Domestic refers to votes cast in respect of domestic portfolio firms; Foreign to votes cast in respect of firms from other European countries, and US to votes cast in respect of US portfolio firms. Panel E reports statistics for the full sample, and compares the mean rate of deviation in respect of proposals at domestic firms, foreign firms and US firms. In Panels C and E we also test the null hypothesis that the means are equal. *, **, and *** indicate whether we can reject the null hypothesis at a 10%, 5% and 1% significance level.

Panel A: Propensity to deviate and firm performance (fund-by-fund)
	Fund 1	Fund 2	Fund 3	Fund 4	Average fund	Median fund
Performance	Mean	Mean	Mean	Mean	Mean	Mean
25^th perc. <	2.5%	1.4%	0.1%	13.5%	4.4%	1.9%
25^th - 50^th perc.	3.6%	0.9%	0.3%	10.3%	3.8%	22%
50^th - 75^th perc.	2.3%	0.4%	0.4%	10.5%	3.4%	1.3%
> 75^th perc.	2.1%	0.2%	0.1%	10.4%	3.2%	1.1%

Panel B: Propensity to deviate and firm size (fund-by-fund)
	Fund 1	Fund 2	Fund 3	Fund 4	Average fund	Median fund
Firm size	Mean	Mean	Mean	Mean	Mean	Mean
25^th perc. <	1.8%	0.8%	0.2%	12.7%	3.8%	1.3%
25^th - 50^th perc.	1.8%	0.3%	0.4%	11.2%	3.4%	1.1%
50^th - 75^th perc.	2.1%	0.8%	0.3%	10.9%	3.5%	1.4%
> 75^th perc.	4.2%	0.8%	0.2%	10.2%	3.8%	2.5%

Panel C: Propensity to deviate and firm size and performance (full sample
	Performance				Firm size
	Obs.	Mean	Std. dev.	Std.err.	Obs	Mean	Std. dev.	Std. err.
25^th perc. <	19124	7.1%	0.257	0.002	16370	6.7%	0.251	0.002
> 75^th perc.	18591	5.6%	0.229	0.002	24559	5.9%	0.236	0.002
Difference of means		1.5%***				0.8%***
t-value		6.07				3.27

Panel D: Portfolio firm nationality (fund-by-fund)
	Fund 1	Fund 2	Fund 3	Fund 4	Average fund	Median fund
Stake size	Mean	Mean	Mean	Mean	Mean	Mean
Domestic	4.7%	3.3%	2.2%	17.5%	6.9%	4.0%
Foreign	3.0%	0.7%	0.3%	9.4%	3.3%	1.8%
US	2.1%	0.5%	0.1%	9.9%	3.2%	1.3%

Panel E: Portfolio firm nationality (full sample)
Portfolio firm nationality	Obs.	Mean	Std. dev.	Std. err.
Domestic	8533	15.5%	0.362	0.004
Foreign	37213	4.5%	0.206	0.001
US	35093	5.5%	0.228	0.001
Method		Value
Anova F-test		770.826***
Welch F-test*		370.976***

Table 5 Propensity to deviate and customization of voting policy

The table reports descriptive statistics for voting and proxy voting recommendations for the two funds in our sample that receive voting recommendations based on a customized voting policy. In Panel A, we measure the degree of overlap between both types of voting recommendations. Difference refers to difference between voting recommendations based on the customized voting policy and voting recommendations based on the general voting policy. Overlap refers to voting recommendations based on the customized voting policy and voting recommendations based on the general voting policy being identical. In Panel B, we compare the funds' propensity to deviate from customized voting recommendations to their propensity to deviate from general voting recommendations. Deviate refers to votes cast differently than recommended by the fund's proxy advisor. Follow refers to votes cast in line with the fund's proxy advisor's recommendation.

Panel A: Customized vs. general voting recommendations
	Difference	Overlap
Fund 1	0.8%	99.2%
Fund 3	15.0%	85.0%

Panel B: Propensity to deviate
Fund 1	Deviate	Follow
Customized voting recommendations	2.6%	97.4%
General voting recommendations	2.7%	97.3%
Fund 3
Customized voting recommendations	0.3%	99.7%
General voting recommendations	14.9%	85.1%

Table 6 Human resources devoted to voting

The table reports descriptive statistics for human resources devoted to govemance of portfolio firms by the funds in the sample. Staff on corporate govemance team refers to the number of staff employed by the fund charged with govemance of portfolio firms (through voting, engagement or otherwise). Portfolio firms refers to the number of firms in the fund's portfolio that are incorporated in Europe (EU27+European Economic Area countries+Switzerland) or in the US.

	Staff on corporate governance team	Portfolio firms (Europe + US)
Fund 1	8	1571
Fund 2	7	762
Fund 3	3	1242
Fund 4	29	3897
Mean	12	1868
Median	8	1407