The pattern shows up in long-run data across many markets, which is more than most calendar effects can claim, and it is one of the more studied anomalies in finance.
Explanations offered include holiday-period liquidity, institutional flows around fiscal years, and seasonal variation in risk appetite. None is well established, and the absence of a solid mechanism is the main reason for scepticism.
Practically, the summer period has generally still produced positive average returns, just smaller ones, so acting on the rule means sitting out of a rising market half the time. The dispersion around the average is also enormous. It is an observation about averages, not a rule for any single year.
Related: seasonality, january-effect, santa-claus-rally, sample-size, overfitting