And it's probably not only economics... does it even make sense to assess how "realistic" is the gp price of a castle, if then we totally ignore how much time it should take to build one?
I am afraid that the typical way to deal with purchases (and sales) in RPG is that of totally ignoring time. You have 50000gp and want a castle (or a +N sword of XYZ, for what matters)? POP! Here it is, take away. Want to sell your 100000gp shield of unbelievable power because you were so unlucky to find it in treasure when you're a TWFer? No problem, it takes as long as striking the entry on your PC sheet.
It's not that this is badwrongfun. It just that it really feels quite stupid to claim realism in prices and then assume perfect therefore unrealistic economy dynamics :/ If you enjoy realism or simulationism, this is a half-baked cake, and if you don't then why bother?
I reiterate that if a group doesn't want this high level of realism, then the only reason to have prices is to balance the relative power of equipment options and (if wanted) the overall power per level of each PC's equipment, but this latter should be a variable chosen by the gaming group, not a default like in 3ed!
Then the problem is that castles, non-adventuring hirelings, armies and any other stuff that cannot be used in adventures, simply cannot be measured in costs with the same scale as magic items, weapons & armor and other adventuring gear. How useful is a castle or an army is too much campaign-dependent. In a traditional dungeon crawl campaign, those things are useless, perhaps not even worth a single gp. OTOH in a campaign that features ruling over a kingdom, those things are important, but it doesn't make sense that they rely on you going personally into a dungeon to scavenge for treasure, when you probably can have money from your land and businesses.