Optimal diversification across mutual funds

Abstract

We evaluate a strategy that minimizes the specific risk of investing in a reasonable number of mutual funds. Our results are consistent with the previous studies, which suggest that actively managed mutual funds are not totally diversified. Our strategy behaves well in terms of diversification, not only in-sample but also out-of-sample. Using different benchmarks, minimizing idiosyncratic risk is also the best strategy for investors seeking alpha.

Keywords:

• mutual funds
• idiosyncratic risk
• diversification
• volatility

JEL Classification::

• G11
• G10
• G12

Notes

¹ We repeat the study for two subsamples (1990–2000 and 2000–2009) to get more robust results. The conclusions are identical.

² For the time period 1999–2000, the percentage that the idiosyncratic risk represents in the total volatility is much higher for both categories, being 29% for EI and 18% for GR funds.

³ For the optimization, we only consider those funds with nonmissing data.

⁴ Given that both the MIR and the MV are optimizers, when we solve them, we obtain the weights of an optimal portfolio, but some of those weights are very low and could be negligible. Then, we assume that the portfolio is only formed by using those weights higher than 5%. We also compute by using a minimum weight of 3% and the results are almost identical.

⁵ Given that each quintile collects a reduced number of funds, we only repeat the experiments for 50 times and select the optimal portfolio within groups of 25 mutual funds.

⁶ Results are identical using a size of 24 observations for the out-of-sample window.