Fund Selection: Sense and Sensibility

Abstract

Studying a comprehensive universe of European-domiciled cross-border UCITS equity and fixed income funds we find that (i) active equity funds on average outperform passive alternatives before fees by about the level of the fees, (ii) active fixed income funds underperform on a net basis, (iii) fees consume a significant part of the value added of active management, (iv) simple fund selection rules help to identify the better active managers. Finally, we develop the concept of minimum required selection skill and find that investors in actively managed funds could meet this hurdle provided fees are limited.

Keywords:

• active funds
• fund performance
• fund selection
• mutual funds
• passive funds

PL Credits: 2.0:

Disclosure statement

No potential conflict of interest was reported by the author(s).

Acknowledgments

We would like to thank two anonymous referees as well as the co-editors of the journal for their helpful comments and suggestions. All remaining errors are ours.

Notes

1 Luxembourg and Ireland are the world's second and third-largest fund domiciles after the US. With assets under management of EUR 8.3 trillion (excluding fund of funds) at the end of 2020, Luxembourg and Ireland jointly represent 16% of global and 47% of European mutual fund assets (EFAMA International Statistical Release (Q4 2020), available at https://www.efama.org/sites/default/files/files/International%20Statistical%20Release%20Q4%202020.pdf).

2 Malkiel (1995) finds a slight gross underperformance of actively managed US equity funds relative to the index, while Gruber (1996) showed a slight outperformance.

3 More than half the countries in Ferreira et al. (2013) are European but they exclude offshore fund domiciles, which are the focus of this paper.

4 This sample includes 11 European countries but excludes Luxembourg because of its offshore status.

5 Like the US 1940 Act, one of the main goals of the UCITS (Undertakings for Collective Investment in Transferable Securities) regulatory framework is to enhance investor protection, for example by requiring portfolio diversification and reducing possible conflicts of interest between investment manager and fund investor. The UCITS “passport” allows for distribution to investors in the European Economic Area, Switzerland, the UK and across several Asian and Latin American countries. Luxemburg and Ireland are the largest European fund domiciles for cross-border distribution (Lipper 2010). According to the Association of the Luxembourg Fund Industry (ALFI), Luxembourg-domiciled funds are distributed in more than 70 countries. It is no exception for individual Luxembourg and Irish UCITS funds to be registered for sales in 25 countries or more. These funds are typically the opportunity set for professional fund selectors with private banks, wealth managers, fund-of-funds, sovereign wealth funds and fiduciary managers across the globe.

6 The impact of dividend tax differs by domicile because of varying bilateral tax treaties. Blitz, Huij, and Swinkels (2012) show the importance of dividend tax in explaining return differences for index funds. Their results imply that the appropriate benchmark for actively managed funds is the performance of passive funds in the same domicile.

7 Institutional share classes are strictly restricted to institutional investors or retail investors who have given an institutional manager a discretionary mandate. In addition, most fund providers impose large minimum investment criteria for institutional share classes to further prevent the higher-fee-paying retail investors from accessing the cheaper institutional alternatives.

8 Studies analyzing Luxembourg funds are Hazenberg et al. (2015) focusing on fund flows in relation to brand and performance, and Hazenberg and Terink (2016) focusing on performance and costs in relation to fund governance.

9 The Morningstar data base is free of survivorship bias in that funds never disappear once added. As of December 2020, 50% of the equity and 37% of the fixed income funds in our universe are “dead.”

10 We set the minimum threshold at 8 to allow for a reasonable choice for fund selection within a given category.

11 We apply a track record extension over the period 2008 up to the moment the first passive fund is available by using the benchmark performance, adjusted for the average cost of the passive fund in later years.

12 We infrequently observe gaps in TNA, which we fill in by multiplying the previous period TNA times that month’s fund gross return, assuming zero flows.

13 While there are therefore on average about 4 share classes per equity fund, it is not necessarily the case that each fund category has both institutional and retail share class alternatives available. Indeed, not all fund managers who offer retail share classes also provide an institutional alternative while fund managers with only an institutional version are an exception.

14 This is the benchmark that Morningstar has assigned to funds based on their Morningstar category.

15 We use a block size of three (months) and 10,000 bootstrap draws as our main setting. However, we have verified that other approaches, including traditional asymptotic t-statistics and block sizes of one or six, yield comparable results.

16 We construct the passive combination separately for institutional and retail share classes by value weighting all funds by their previous month-end TNA. As ETFs do not have separate institutional and retail share-classes we assume a 75% institutional/ 25% retail TNA split to construct their value-weighted passive performance (see ECB, Financial Stability Review, May 2017, p. 108, available at https://www.ecb.europa.eu/pub/financial-stability/fsr/focus/2017/pdf/ecb∼316313375b.fsrbox201705_08.pdf for evidence supporting this split). Moreover, in case no index fund or ETF is available for a particular category early in the sample, we apply a track record extension as described in Section 2.

17 Note also that, consistent with Leippold and Rueegg (2020), especially the US Large Cap equity segments had—in our case often significant—negative gross and net excess returns.

18 Note that the inferred fees are negative for the institutional share classes of India and Natural Resources as the passive alternative for these categories had higher fees than the active ones.

19 These are roughly the categories in Table 2 where the average net return over passive is positive, except that the results of Table 2 are based on historic costs whereas Table 5 assumes December 2020 fees.

20 Available from the authors on request.

Additional information

Notes on contributors

Guido Baltussen

Guido Baltussen is a professor of finance at Erasmus University Rotterdam in the Netherlands, and head of Factor Investing at Robeco Institutional Asset Management.

Stan Beckers

Stan Beckers is Chair of the AQR Asset Management Institute at London Business School.

Jan Jaap Hazenberg

Jan Jaap Hazenberg is head of product strategy at NN Investment Partners.

Willem Van Der Scheer

Willem Van Der Scheer, CFA, is a senior investment manager at Aegon Bank.