They do actually measure those numbers. Send an burst signal to the speakers and push the fader until the reading won't go up any more. Doesn't matter if the LF driver was limiting 10dB ago, they'll carry on. If you were to replace the test signal with anything else, the speakers would likely burn up very quickly as the test relies on a very short signal that won't give the drivers any thermal stress.
If you ask me, they might as well be dropping the speakers off buildings and seeing how loud it is when they hit the ground. The data is about as useful.
Chris
For those of us who've been around since mid-last century... there were all kinds of finger pointing "liars" back when home HiFi/stereo was a huge business. The "Institute of High Fidelity" had their standards for distortion, power and bandwidth measurements, IEC had another, there were others (Peak Power, burst, etc) and finally the US Federal Trade Commission weighed in with some regulations regarding power output but that was it.
As Ivan says, the great thing about standards is having so many to choose from.
From a manufacturer's POV consumers are largely stupid, the vast majority of them making decisions based on easily promoted distinctions that appeal to the 'more, louder, bigger' based emotions. This human condition of wanting simple & obvious distinctions gave way to reliance on single-number specs and/or use of criteria that are, at the end of the wire, not terribly important. So long as the number is bigger or perhaps smaller than the competition, it's fair game.
It would make me happy to see specifications, even single numbers, from multiple measurement criteria - real RMS output like Crown used to brag about; burst power, bandwidth-limited output, power at a target distortion, etc. Won't happen, as it makes it easier for competitors to target each other. Same applies for loudspeakers.
Because we get feed marketing specs that are based in maybe 20% reality it's more important than ever to audition gear before committing to large purchases. The cost of a trip to see, hear and touch a prospective system or item can be worth it, either to confirm the choice or especially to find out you'd have regretted buying that system or item.