Again, we come back to the question of reasonable error. In order for the 1e modules to be played up to full potential, a lot more behaviors and variables must be assumed than for the 3e modules. For the 3e modules, you just have to assume that all encounters are satisfactorally dealt with. Making a comparable assumption in 1e may not even get you the kill XPs because of the potential for loss. To net the other XPs for treasure, even more assumptions of timing, thoroughness, and transport capacity must be made.
Well, the problem with worrying about this element is multifold. First off, the amount of treasure "hidden" in modules has been shown to be a fairly small portion of the total. The amount of trasure shown to be incredibly cumbersone and difficult has been shown to be almost trivial (if you look back at the horde used as an example from G3, the 15,000 pounds of copper and silver amount to only about 3,320 gp value out of a horde with total worth of 206,000 gp - less than 1.5% of the total). So the only thing you are left with that might have a significant impact is the assumption that the PCs will lose out on XP due to training, but since Gygax has asserted he never played using those rules, it is unlikely that TSR assumed others would either.
And the issue here is that the modules aren't being assumed to be played optimally, at least insofar as the XP calculations go. The calculations explicitly exclude XP for magic items, which even if not sold, amounts to a huge chunk. The PCs could literally miss a quarter of the monsters and treasure in an adventure and still come up with as much (if not more) experience as listed in the examples due to the gain from magic item XP. If the examples assumed that the PCs would play through the adventure optimally, then the XP calculations would be much higher, and the 1e PCs would advance even faster than they are shown to be doing.