The claim is made here ->
https://assets.ctfassets.net/4zjnzn...iers_for_Active_Loudspeakers_Applications.pdf, although it is really more of a mathematical fact than a listening based study.
Michael
I don't have enough technical knowledge in this field to authoritatively refute the claim. But I will take a stab at interpreting it, which is that the paper is looking for cost effective active speaker performance, rather than the best performance under more general circumstances. There is little doubt that well-made active monitors offer the most cost effective hifi performance. The cost effective market has little overlap with the market for flagship products such as the la90.
In addition, the paper you quoted handpicked parameters such as the "typical listening volume" parameter without justification. I guess the author is lacking the tools that Amir has, or doesn't want to use them for some reason, but I would expect a 3d graph for each amplifier of thd+n vs frequency vs wattage if the author was trying to be thorough in making his point, or at least 2d graphs of each pair.
Finally, before the author's "cost compensation," significant differences do appear in the measurements.
TLDR: The paper does not prove that the loosely defined heuristic is a "mathematical fact" to my mathematically trained eyes.