This data – along with evidence demonstrating that the vast majority of actively managed funds persistently underperform and recent studies showing that only about 2 percent of them are generating statistically significant alpha even before taxes – leaves us with an anomaly. Why are so many investors engaged in a losing strategy?

There are several likely explanations. The first is that investors are unaware of the evidence, and the fund industry certainly isn’t about to educate them. Instead, it spends huge sums selectively advertising returns. Another likely candidate is that, while investors may be aware of the evidence, they are, on average, overconfident in their ability to reliably select the few future outperformers. Overconfidence is an all-too-human trait, and it has been found to exist even among those with little knowledge or skill in a given field.

Some studies have actually found that when it comes to choosing mutual funds, there is such a thing as “smart money.” For example, in his 1996 study, “Another Puzzle: The Growth in Actively Managed Mutual Funds,” which covered the period from 1985 through 1994, Martin J. Gruber found that the average performance of funds with net inflows was significantly positive, based on a risk-adjusted performance measure. Other studies found that funds with positive net flow perform better than those with negative net flow over the short term. However, subsequent research revealed that the smart money effect was the result of prior studies having overlooked the impact of momentum in stock returns.

Another interesting finding from the research is that the more mature the fund industry, and the more advanced the country, the more sophisticated investors are. Thus, perhaps money has gotten smarter over time – and prior studies, dominated by older data, could’ve missed a smart money effect.

Yeonjeong Ha, Hyeongseok Kang, Tong Suk Kim and Heewoo Park contribute to the literature with their March 2016 study, “Money Is Smart: New Evidence from Investors’ Buys and Sells.” While prior studies of U.S. mutual funds used quarterly data, the authors manually collected the actual monthly flow data of U.S. mutual funds, separated into inflow and outflow, to determine if there was smart money. Their sample included all actively managed domestic equity funds from January 1994 to June 2011. They found that there is a limited smart money effect after controlling for momentum and other common factors (market beta, size, value and liquidity). Interestingly, they also found that investors are not simply naive chasers of a fund’s past performance, which prior studies found them to be. After controlling for funds’ past performance, there is still a smart money effect.

Another interesting finding was that funds experiencing large outflows “underperform their peers with smaller outflows. This result means that investors withdraw their money from funds that perform worse in the next period. Investors therefore also have fund avoidance ability, which is also the smart money effect in another sense.” In other words, while cash inflows into larger funds do not predict superior future returns, cash outflows from them do predict inferior future returns. Unfortunately, the story is not one-sided.

The authors write: “In all cases of flows, differences in performances between highest and lowest flow portfolios no longer persist after 1 month.” The short-term horizon could explain why studies using quarterly flow data do not find the smart money effect. In addition, remember that in terms of inflows, the authors found smart money is only smart in the case of small funds (funds with limited asset size). On an equal-weighted basis, the four-factor (market beta, size, value and momentum) alpha difference between positive and negative net flow portfolios was 0.068 percent per month, and was statistically significant at the 5 percent confidence level. On a value-weighted basis, while the returns were still positive, there was no longer any statistically significant alpha.

While this latest study did find some evidence of a smart money effect with fund inflows, it’s important to understand that, just like theory predicts, as assets under management increase, the ability of the fund manager to generate alpha deteriorates. Thus, given the very short horizon of smart money benefits and limited fund capacity, there just doesn’t seem to be much there. And that’s certainly the case for taxable investors.