Use of active peer benchmarks in assessing UK mutual fund performance and performance persistence

ABSTRACT

The majority of UK style-specific mutual funds either report a broad market index as their prospectus benchmark or give no benchmark at all – a practice that may be (a) strategic, or (b) cultural and attributable to the lack of UK style-specific indices (e.g. mid-cap-growth, small-cap-value). The choice of a broad market index as a benchmark can bias the inferences of a fund’s performance and performance persistence. This study is the first to provide an alternative to style-specific indices in the UK, and suggests the use style-specific peer group benchmarks, following [Hunter, D., E. Kandel, S. Kandel, and R. Wermers. 2014. “Mutual Fund Performance Evaluation with Active Peer Benchmarks.” Journal of Financial Economics 112 (1): 1–29]. Our sample comprises of 817 active UK long-only equity mutual funds allocated to nine Morningstar style categories (peer groups) during the period 1992–2016. We show that the funds with the most significant positive peer-group-adjusted alphas continue to perform well one year ahead, in terms of both parametric and non-parametric measures of persistence in performance. Moreover, persistence in performance is driven by both winner and loser funds. The results within each peer group are by and large consistent with these findings.

KEYWORDS:

• UK equity mutual funds
• active peer benchmark
• performance ranking
• performance persistence

JEL:

• G11
• G12
• G23

Disclosure statement

No potential conflict of interest was reported by the authors.

ORCID

Irina B. Mateus http://orcid.org/0000-0003-2591-8080

Natasa Todorovic http://orcid.org/0000-0003-4875-623X

Notes

1. FCA Asset Management Market Study, Final Report (June 2017). Available at: https://www.fca.org.uk/publication/market-studies/ms15-2-3.pdf [Accessed May 2018].

2. https://www.msci.com/documents/10199/a9bfef62-3b73-4884-a84b-8239c9fc7463 [Accessed May 2018].

3. https://www.ft.com/content/33422714-6326-11e7-91a7-502f7ee26895 [Accessed June 2018].

4. There is no relevant change in style categories of our funds over the sample period, hence, for our peer-group classification we use the last available one.

5. Depending on the purpose of a study, thematic funds can be assigned to a separate peer group category that corresponds to their theme.

6. See, for example, Chan and Lakonishok (2004), Dimpson, Marsh, and Staunton (2004), Fama and French (1998), and Reinganum (1999), amongst others, for evidence on small cap and value outperformance.

7. Note that the last rolling period has 38 months, as the sample ends in February 2016. Also note that in Tables 4–6, the last rolling period used for the prediction of future performance is 2012–2014.

8. The UK market risk premium represents the return on the FTSE All-Share Index (R_M) minus a one-month UK Treasury bill (R_F), as per Gregory, Tharyan, and Christides (2013).

9. http://business-school.exeter.ac.uk/research/areas/centres/xfi/research/famafrench/files/.

10. Subscript APB represents the active peer group average of fund-specific parameters.

11. As in the study by Hunter et al. (2014). APB-adjusted alphas and their t-statistics in period t are estimated using previous t-36 months of data.

12. The Small-Cap Value category fails this requirement in all of the rolling periods except 2009–2011, when only eight funds were present. For this reason, the results for that style are not shown. They are available upon request.

13. Note that in the Mid-Growth style, due to an insufficient number of funds it was not possible to create quartiles in the first six rolling periods.

14. Results are available on request.

15. The results are available upon request.

16. In a similar manner, we test the predictive ability of the standard Carhart alpha and find that the overall marginal effect (beta from Equation 5) was reduced by 58% when historical Carhart alpha was used, from 15.53 to 9.8bps, the difference being significant at the 5% level.

17. Note that four quartiles are formed as before, according to the t-statistics associated with APB-adjusted alphas estimated for a fund for period t, using t-36 months of historical data.

18. Our results remain robust when the funds are split into winners and losers according to the values of their APB-adjusted alpha, not t-statistics. Note that four quartiles are formed as before, according to the t-statistics associated with APB-adjusted alphas estimated for a fund for period t, using t-36 months of historical data.

19. The results are available upon request.