The Government Pension Fund Global: How are the bets distributed, and where is the active return created?: A holdings-based analysis of The Government Pension Fund Global’s investments in equities from 2014 to 2021.

Abstract

The Management of the Government Pension Fund Global (hereafter the “Fund”) is a heated topic. The Fund is managed actively to outperform the benchmark and generate a positive active return. The literature is skeptical of this possibility, but the Fund advocates that the active management has yielded a positive active return. This thesis studies the active management of the equities portfolio of the Fund. Therefore, a holdings-based analysis is conducted for the time period from 2014 to 2021. First, we investigate how actively the Fund has been managed. The Fund’s active management has been low, and we find indications of a decreasing degree of active management during the time period. Second, we identify how the Fund’s bets can be divided into overweights and underweights, regions, and industries. We find that the Fund’s overweights have a larger variation in size than the underweights. Addi tionally, the distribution of bets between regions and industries is largely determined by their market capitalization. This distribution is somewhat affected by different degrees of active management in the regions and industries. Third, we investigate the contribution to the active return from the different groups of overweights and underweights, regions, and industries. We did not find statistically significant con tributions to the active return from any regions. This was the case for industries as well, without the exception of a significantly positive contribution to the active return in Utilities. This positive return was mainly caused by good performance in security selection. We found a significantly positive contribution from the 20 largest underweights. We also found indications of a positive contribution to the active re turn from all underweights and a negative contribution from all overweights. These results were not statistically significant, and we cannot conclude this was caused by poor or well performance and not a coincidence.