On Persistence in Mutual Fund Performance

The author starts by summarizing the prior research of momentum, especially the 1-year momentum strategy of Jegadeesh and Titman; however, the author notes that his study finds those excess momentum returns are consumed by mutual fund fees and loads, and may even be just a consequence of portfolios having larger weights in momentum stocks as they rise.

The author used a database of 1,862 diversified equity mutual funds (excluding sector, international, and balanced funds) over the period 1962 - 1993; importantly, the database includes defunct funds so as to mitigate survivor bias. 

The author introduces his 4-factor model, which includes the volatility, size, and value characteristics of Fama/French's 3-factor model, and adds 1 factor for the momentum anomaly.  He then runs a regression on these factors, showing that the addition of momentum as a factor significantly improves the fit of the equations.  In additional tests, he shows that pricing errors based on the equation are significantly reduced when using the 4-factor model instead of the 3-factor model or the CAPM model.

The author then forms portfolios of mutual funds and ranks their performance into deciles and calculates the betas.  In all deciles, the 4 factors explain more than 89% of the variation in returns; size and momentum seem to be the deciding factor; and the factor proxy and value factors seem to have the same effect across all deciles (the factor proxy explaining the majority of the variation in returns, and the value factor explaining none).

The author then dives deeper into the mutual fund performance, noting that in most cases one year's losers become the next year's winners and vise versa.  Also, new funds may have returns significantly different than the mean, but they eventually fall in line with the averages, in a normal distribution. 

The author then ranked mutual funds by momentum, and found that momentum funds actually underperform contrarian funds and typically have higher turnover and fees which adds to the underperformance.  He contributes high performance of momentum portfolios to holding the prior year's winning stocks by chance.

The author then uses his 4-factor regression to rank estimated returns into deciles; as would be expected, top decile portfolios earned the highest excess returns and decreased with each decile.  The most significant contributions to return were the size and market proxy factor; value and momentum were negligible.  Also, the top decile portfolio returns seem to persist over the next 5 years.

The author ends with 3 suggestions for mutual fund investors: 1.  avoid funds with persistently poor performance, 2. past year winners typically have higher than average returns the next year, but not thereafter, and 3. expenses such as management fees, load fees, and turnover all have a significant negative effect on performance.


Carhart, M.M. (1997). On Persistence in Mutual Fund Performance. The Journal of Finance, 52(1), 57–82.