4.3 Data and Methodology

Our sample consists of four funds, three managers of pension fund assets and one pure play asset manager. Three funds are located in the Netherlands, the fourth is located in the UK. The funds were selected non-randomly, based on two criteria. The first is size; we aim to study the voting behaviour of relatively large funds. The average amount of assets onder management (AUM) of the funds in our sample was USD 166bn (median USD 141bn) as at 1 January 2010, which places the funds among Europe's largest funds. The second criterion is governance style; we aim to study the voting behaviour of funds with a proclaimed active governance style. The funds in our sample each present themselves, through their website, as having an active governance style. Moreover, two of the funds perform corporate governance activities not only on behalf of asset owners whose assets they manage themselves, but also on behalf of asset owners whose assets are managed by a third party.

By selecting on the basis of these criteria, we expect to obtain results that represent the upper bounds of activism by European institutional investors (excluding activist hedge funds). We expect that the influence of proxy advisors on the voting behaviour of smaller funds and of funds with a more passive governance style will generally be larger than the influence of proxy advisors on the funds in our sample. Smaller but equally diversified funds will generally have fewer resources per portfolio firrn available to invest in monitoring, and therefore can be expected to be more inclined to rely on outside proxy advice. Similarly, for funds that do not present themselves as having an active governance style, governance of portfolio firrns will generally constitute a less important element of their investment strategy, and therefore they can be expected to rely more on outside proxy advice when voting shares in portfolio firrns.

We requested data from each fund with respect to voting decisions on management and shareholder proposals put to a vote at portfolio firms incorporated in Europe (EU27+EEA+Switzerland) and the US and taken between 1 January 2010 and 31 August 2010, enabling us to capture the peak of the 2010 proxy season. For purposes of our analysis, we are only interested in voting decisions FOR or AGAINST management, or ABSTAIN. We therefore exclude proposals where the fund decided to "take no action", which is usually the case when the portfolio firrn requires the fund's shares to be `blocked' prior to the meeting in order for the fund to be eligible to vote at the meeting. This leaves us with a dataset including 80,838 voting decisions. The median fund in our sample holds shares in 1,407 US and European portfolio firrns, and its median stake represents 0.07% of outstanding shares and has a market value of about USD 2.7m.1

Each of the funds in our sample retains proxy advice. Two funds retain proxy advice from ISS and two from Glass Lewis, the second largest proxy advisory firrn. For each voting decision, we requested the corresponding voting recommendation from the proxy advisor. Two of the foor funds receive voting recommendations on the basis of a customized voting policy. At the same time, each of these funds also receives the voting recommendations based on the proxy advisor's general voting policy, and we have requested both voting recommendations. This is one of the ways in which the present study adds value to existing studies, which only have access to general voting recommendations.

We matched the data on voting decisions with data on portfolio firms. Data on firm performance, firm size and number of outstanding shares was obtained from Datastream. Data on number of shares held, and on the funds' human resources devoted to corporate governance, was obtained directly from the funds.