The MA only need approximately 8 watts to achieve 95db at the driver, this speaker would need about 16watts (due to whatever additional parts it has). I have no idea where on earth you are getting 100w.
Thermal compression they might be about equal per volume, maybe less on the MA (or even a good bit less as SPL climbs) for having a ventilated former that's also heatsinked, and has more surface area with no spider to limit circulation.
The MA is at 100db at 17.5khz, and down to 92db at 20khz. Even if you were to say that the Fostex has less because of the dust cap shape, fact is that it only accounts for a few db (just think of it like a baffle), not 7-15db.
Fostex's get shouty because the bandwidth collapses as they try to play real loud. It's just like EQing some of the annoying frequencies higher than they should be. (This is BL curve, which you can easily know what it is if you explore driver motor design more study up on underhung high sensitivity then look at XBL2 for contrast)
Compression drivers inherently need less because they have super limited limit travel. If they could make up for the fact magically, they wouldn't need all the mechanical advantages. It's easy for them to exert control because they don't have to move much at all - you hear this part because they're highly damped, very low distortion. Acceleration cannot magically be faster for any given frequency because either it means more SPL, which we aren't talking about even though they play loud, or that the time window is destroyed and you have massive distortion since you're changing the frequency. When a diaphragm moves it does so within the envelope of the frequency's time window, the only way for it to change volume is to displace more air within that time window. That means it must accelerate faster to close the distance in the same given amount of time. SPL is a function of acceleration, and velocity loosely only represents a null before deceleration. If velocity played another function you'd be making square waves. This is notable for the fact that you can't accelerate faster to a velocity to maintain, to make the same frequency (you'd be making DC, 0hz). That matters because it proves that you can't have any driver accelerating faster than another if they play the same SPL and are of a relative same size. If you say smaller drivers move faster, well, yes, they do, but it's also a bandwidth restriction & respective of SPL (that may include assisted loading).
The truth is that an awful lot of compression horn drivers have significantly lower acceleration than a lot of dome tweeters. The dome tweeter for the same SPL must accelerate faster because it has no loading advantages at all, it must displace more raw air, spending a lot more wattage on lost energy going other directions. Because it must have more capable travel, it will inherently have a less damped structure - you hear this. This is compounded by the larger sized diaphragms often found in compression tweeters. When it comes to mid/midwoofers the size being equal the dome must travel faster for a given SPL, but if you've got say 8-10" driver the requirements for acceleration go down a lot and the compression driver at 2" or whatever is probably moving faster for a shared frequency they play.
A lot of the ability related to SPL low output for single drivers of high sensitivity but low xmax is assistance like the loading you talk about, but there's another factor. They often need less feedback so that the amp will be like "ok whatever" when it sends out a bass note that ends up looking distorted to the amp due to the loading assistance (this is more complicated and involves weird stuff like reversal in phase below FS etc).
Anyways... Mark Fenlon knows this stuff, and designed his drivers very intelligently to make a more useful fullrange driver. It doesn't beam so easily because of the motor design - it's an advantage not disadvantage and the cost was some sensitivity. When using a 15-25w amp or such, you get more out of it because of the smarter design, and it doesn't require complicated transmission lines with feedbackless amps to play music (although it probably sounds pretty darn good with an SET). They have high enough dispersion that off-axis is preferred as well! They're fascinating drivers.
