I have posted on this previously, but want to provide a link and more detail. The link is old, to the 2011 Credit Suisse Yearbook, out a year ago, an article entitled "The quest for yield," by Elroy Dimson, Paul Marsh, and Mike Staunton, on pp. 15-23 of credit_suisse_global_investment_yearbook_2011.pdf . I note that so far there has not been a single academic publication on the gist of this paper, that over the last 20 years in 19 out of 21 countries, buying annually reconfigured high yield stock portfolios provided both higher returns and lower risk than alternative investment strategies. Needless to say, this is an anomaly that violates CAPM, the Efficient Market Hypothesis, and several other sacred cows of conventional financial economics theory.
Of the 21 countries studied, the only ones where the "yield effect" was negative, higher yield portfolios underperformed lower yield ones on returns over 20 years, were New Zealand and Ireland, not exactly major markets. The yield effect was highest in Austria, France, and Japan.
The more crucial matter is risk, given that standard portfolio theory dating from Markowitz, if not much earlier, is that risk and return are positively related, not negatively (although another violation has been the matter of home asset bias, where most would do better on both risk and return by internationally diversifying portfolios more than they do, although this has been known and much studied in the academic lit for a long time, unlike this matter). So in Figures 8 and 9 one finds the crucial findings, both across all countries and then broken down for them individually, comparing high yield, low yield, zero yield, and country index funds. For the varying yields, it is monotonic, with risk rising as yield declines using both standard deviation and beta, and also with the Sharpe ratio declining (return per extra unit of volatility) across the yields from high to zero. It is a closer call in comparing with the country index funds, which do better than the lower yield strategies on all of these. However, it is only on standard deviation that index beats high yield, but just barely, 21.4 to 22.6. High yield beats index on beta, 0.89 to 1.0, and simply tromps it on the Sharpe ratio, 0.42 to 0.30.
So, how is this explained? Well, it is not likely it is chance, and while there might be some tax effect, the authors provide arguments why this is probably not the case or only minimally so at most. This leaves a behavioral explanation: people overbuy low yield growth stocks in bubbles that end up being more volatile because they crash so much harder. Looks pretty reasonable to me, but nobody in academic economics or finance is talking about this at all.