Do sovereign wealth funds herd in equity markets?

Click to increase image sizeClick to decrease image size

Acknowledgements

This paper was written during my stay at the CFAP—University of Cambridge. I would like to thank the CFAP itself and in particular John Eatwell for their hospitality during my visit. I would also like to thank Alberto Quadrio Curzio for his constant support; Bernardo Bortolotti for access to the FEEM/Monitor Group SWF database; Alessio Ciarlone, Francesco Corsello, Giancarlo Corsetti, Carlo Cottarelli, Michael Dempster, Lyudmyla Hvozdyk, Richard Sias, Vanessa Smith, Riccardo Trezzi, Geoff Whittington and participants at the Cambridge Finance Workshop for their helpful comments. This research has been financed by Università Cattolica through the project D.3.2—2009, ‘Stato e Mercato’. All errors are the sole responsibility of the author.

Notes

¹This worry is made worse considering that most SWFs come from countries where transparency standards are low and where there are little or no domestic authorities or regulations to comply with.

²For example, the two Russian funds (amongst the largest ones) are banned from equity trading.

³The Monitor-FEEM database uses multiple public sources including financial databases (Bloomberg, SDC Platinum and Zephyr M&A), disclosures from fund websites, information aggregators (Lexis Nexis and Factiva) and other internet sources (http://Zawya.com, Sovereign Wealth Fund Institute).

⁴For example, Temasek makes some investments through its subsidiaries, such as Vertex Venture Holdings or Aranda Investments.

⁵On the contrary those transactions that were announced, even if not realized, have been kept in the data-set since their announcement is considered suitable for determining herding behaviour.

⁶Only 1904 (as sum of acquisitions and dismissals) deals out of 2740 are considered for calculating the amounts because the remaining deals do not report any amount. For the transactions characterized by multiple investors/sellers, the amount related to the specific SWF was separated from the total amount of the deal. In case the amount attributable to the single SWF was not specified, the total figure of the deal was divided by the number of the participants.

⁷Cfr: http://mba.tuck.dartmouth.edu/pages/faculty/ken.french/Data_Library/det_12_ind_port.html.

⁸Only 849 (as sum of acquisitions and dismissals) deals out of 897 are considered for calculating the amounts because the remaining deals do not report the amount.

⁹Cfr: http://mba.tuck.dartmouth.edu/pages/faculty/ken.french/Data_Library/det_12_ind_port.html.

¹⁰Walter and Weber (2006) however find a slightly monotonically increasing result.

¹¹The only case where the LSV measure comes out positive is when all deals are considered regardless of whether they are carried out by the same SWF or not. In this case, the LSV measure is positive (significantly different from zero) and equal to 2.66%. However this is due to the fact that SWFs ‘herd on themselves’ and this makes their trades not independent. This outcome provides a proof that the sample can convey a positive result when trades are not independent.

¹² http://mba.tuck.dartmouth.edu/pages/faculty/ken.french/Data_Library/det_30_ind_port.html.