Overall, turnover proved to have a smidgen of predictive power, but it was not as strong or consistent as the best predictors. While fees, stars, and manager tenure produced a nice stair-step down, the turnover results were lumpier. So while the cheapest quintile beats the next-cheapest quintile and so on, turnover didn't work that way.
Instead, we had rather flat results up until the highest-turnover
quintile, where results really plummeted. So, put turnover behind stars,
expenses, and manager returns, but put it ahead of active share. Along
asset-class lines, turnover had some predictive power for equity funds.
Shopping for low-turnover balanced funds would have actually hurt
performance, and turnover had no effect--positive or negative--in bond
funds. No big surprise there.
The results are pretty close to what other studies have found. For
example, Roger Edelen, Richard Evans, and Greg Kadlec found a small
impact for turnover in their paper, "Scale Effects in Mutual Fund
Performance: The Role of Trading Costs." However, in their study it
wasn't as powerful as expense ratios or their estimate of trading costs.