A difference in maintenance costs is not necessarily a consequence of disinvestment... it could simply be that lower performing schools are more likely to be urban, and in urban areas everything related to the construction and maintenance of buildings, whether old or new, is more expensive (not to mention admin salaries, etc.). Or it could be that plus what you say for a double wammy at some schools. I'm just brainstorming reasonable seeming hypotheses that you could imagine trying to rule out with the right dataset.