Volatility-Managed Portfolios of Mutual Funds

Performance evaluation of mutual funds using factor pricing models is usually distorted by the existence of a volatility anomaly and correlated residuals. By augmenting the Fama-French five-factor model with an active peer benchmark, we eliminate the measurement errors caused by these distortions. We find that the APB-augmented five-factor model can select mutual fund portfolios with superior out of sample alphas. However, the excess portfolio returns are volatile and subject to crash risk. A risk management strategy which uses the past volatility of the mutual fund portfolio returns to scale the investment positions can help reduce the volatility of the portfolio returns and avoid extreme losses. Previous studies on stock and currency portfolios find that such a risk management technique also increases the Sharpe ratio of the portfolios. However, we find that the net impact of the risk management on the Sharpe ratio of mutual fund portfolios depends on the investment style of the mutual funds in the fund portfolios. Similar to the stock and currency portfolios, the Sharpe ratios of the mutual fund portfolios can be increased by the risk management if the portfolios are formed in mutual funds which specialize in large cap stocks. By contrast, risk management reduces the Sharpe ratios of portfolios which include small stocks-focused funds. The reason is that the mutual funds which specialize in small stocks relies more on their abilities to take on more factor risks when the factor returns are high to generate returns. Volatility timing reduces the premium of those funds’ market timing skills. Compared to investors in mutual funds which specialize in value stocks, investors in growth stocks-focused mutual funds can better enhance their market timing skills by the risk management technique. More specifically, after volatility timing, the growth stocks-focused mutual fund investors are able to take on more factor risks when the factor returns are high and gain higher average returns.