Peter Urbani
Dec 15, 2009
Hi John you raise some very valid points which occurred to me also. However to control for the small sample bias I also generated random samples of 'male returns' of a similar sized group as that of the 'female' only sample. Although there was more variability in these smaller groupings even the returns of the top decile of these did not approach those of the 'female' only sample. Nothing that can be done about survivor bias which would be inherent to the entire database.
john
Dec 02, 2009
It seems to me that the results may be flawed due to the large imbalance between male and female managers. In our industry, the ratio of male to female may be in the magnitude of 1000:1. This therefore means that the average of male return will be near the mean of the entire universe, while one outlier would shift the average female return tremendously. Given the inherent survivor bias in industry results the overall impact of an outlier would be to shift the female average to the upside, hence your "out performance". The way to design a truly representative study would be to take a random sample of equal number of observations and compare those results.
Peter Urbani
Nov 27, 2009
I had a brief look replicating these results and in general found the findings to be accurate. However, the available pool of clearly identifiable female only managers remains very small. The HFR research is based on the performance of their Diversity Index which includes all 'minorities' as well as woman. An equally weighted Index of funds where the portfolio manager was clearly identifiable as a woman produced a CAGR of +10.73% from Jan 03 to Sep 09 versus the +9.35% for the HFRX Diversity Index and the +1.30% to +1.94% for the HFRX Equally Weighted and Global Hedge Fund Indices over the same period.
James J
Nov 17, 2009
Hi David. You make a good point. Here is a link to a PDF version of the report, Women in Fund Management: http://www.ncrw.org/hedgefund/?page_id=7
You may find the answers to your questions within the report's findings. I also included the link in the story.
David J
Nov 17, 2009
To give this finding some weight and credibility, it would be helpful to link to the paper. Were the results statistically significant? Was there any attempt to control for bias such as investment style or AuM?