4.4.1 Summary Statistics

Consistent with prior studies, we find that as a general matter, the funds in our sample rarely deviate from proxy voting recommendations. Table 1 (Panel A) shows that the average fund deviates from its proxy advisors' voting recommendation 3.7% of the time. However, deviation rates vary significantly across the funds (range 0.7 - 11.2%), and the median deviation rate is much lower (1.6%). Panel B reports statistics for the full sample, and shows that 6.1% of all votes included in the sample deviated from the proxy advisor's recommendation.

By way of comparison, Panel C looks at voting in relation to management's recommendation. The median fund votes in line with its proxy advisor's voting recommendation more often than with management's recommendation (98.4% vs. 87.3%). Panel D shows that when the median fund does deviate from its proxy advisor 's voting recommendation, it does so in order to vote FOR management most of the time. This is interesting, because proxy advisors recommend to vote FOR management most of the time, so if funds would invest an equal amount of resources in verifying the accuracy of proxy voting recommendations, deviations should be randomly distributed and we would expect funds to mostly deviate to vote AGAINST management, rather than to vote FOR management.

One explanation for this finding may be that funds sometimes have firmspecific information indicating that it is not necessary to vote AGAINST management even if the proposal technically is not in line with the `one size fits all' voting policy on which the proxy voting recommendation is based. Indeed, some funds in our sample explicitly state in their voting policy that in deciding how to vote, they take into account the relevant circumstances at the portfolio firm. Management obviously is an important source of information about these circumstances, if only because it tends to include explanations for its proposals in notes to the meeting agenda. Funds' propensity to collect and analyze this information may be partly endogenous: that is, a proxy voting recommendation to vote AGAINST management may act as a trigger for the fund to verify the accuracy of such recommendation. A less benign explanation may be that funds are reluctant to vote AGAINST management of portfolio firms for fear of jeopardizing their commercial relationship with these firms. The studies referred to earlier suggest that business ties between US mutual funds and portfolio firms affect the voting behaviour of those funds. We cannot exclude the possibility that business ties also affect the voting behaviour of the funds in our sample, particularly since a majority specializes in managing pension assets.