Performance and characteristics of mutual fund starts

Abstract

We study the performance and portfolio characteristics of 828 newly launched US equity mutual funds over the period 1991–2005. These fund starts initially earn, on average, higher excess returns and higher abnormal returns. Their risk-adjusted performance is also superior to existing funds. Furthermore, we provide evidence for short-term persistence among top-performing fund starts, however, a substantial fraction of funds drop from the top to the bottom decile over two subsequent periods. Analyzing portfolio characteristics, we find that returns of fund starts exhibit higher ratios of unsystematic to total risk. Portfolios of new funds are typically also less diversified in terms of number of stocks and industry concentration and are invested in smaller and less liquid stocks.

Keywords:

• mutual funds
• fund starts
• performance evaluation
• performance persistence

JEL classification :

• G11
• G12
• G23

Acknowledgements

This paper has benefited from the comments of Bruno Rémillard, Chintal Desai, as well as participants at the 2007 Optimization Days at HEC Montréal, the 2008 Midwest Finance Association conference in San Antonio, the Desjardins Global Asset Management workshop, the 2008 Financial Management Association (Europe) meetings in Prague, and two anonymous referees. Aymen Karoui was supported by a grant from CREF and Iwan Meier acknowledges the research support from the Institut de Finance Mathématique de Montréal (IFM2).

Notes

ICI Mutual Fund Factbook 2008: www.ici.org, ‘Data Tables’; ‘Section 1: US Mutual Fund Totals’; Tables 3, 5, and 48.

Another strand of literature studies the impact of fund family affiliation. For example, Berzins (2006) analyzes institutional money managers and finds that the performance of a newly launched fund tends to fall into the same performance decile than the existing funds of the same family.

Peterson et al. (2001) and Prather, Bertin, and Henker (2004) report no significant difference in the performance of new and existing mutual funds.

The focus of Khorana and Servaes (1999), who study 1163 emerging funds over the period 1979–1992 (including other fund classes than equity funds), is on the determinants of mutual fund starts. They conclude that generating additional fees is the driving factor for fund openings. Fund families that outperformed their peers, larger fund families, and those who added new funds in the recent past are more likely to create a new fund. Fund families also more likely add a new fund when the largest fund families open new funds in a given investment style category.

http://mba.tuck.dartmouth.edu/pages/faculty/ken.french; Section ‘Data Library,’ ‘US Research Returns Data.’ The website also details the portfolio formation methodology to construct these factor returns.

http://pages.stern.nyu.edu/~jhasbrou; Section ‘Research and Working Papers.’

To check whether sample selection has an impact on our results on cumulative performance, we also tested with an alternative sample where we select funds from the CRSP Survivors-Bias-Free US Mutual Fund Database following the sample selection methodology of Pastor and Stambaugh (2002). The authors provide a list of ICDI (named ICDI_Obj in MFDB), Wiesenberger (Obj) and Strategic Insight (SI_obj) classifications to identify US equity mutual funds in the CRSP mutual fund database and eliminate balanced funds, bond funds, flexible funds, international funds, mortgage-backed funds, money market funds, multi-manager funds, and specialized funds. Multiple share classes have been introduced starting in the 1990s and we use a name algorithm to exclude share classes as in Nanda, Wang, and Zheng (2005). Qualitatively, the results are similar: The performance declines rather quickly after the initial 3 years and the difference is significant over two subsequent 3-year windows.

When we consider the trend starting in 1962, then average fees have increased over time.