This sounds like something resembles multiplier effect of simple short-run model in macroeconomics, so it is supposed to be easy to accomodate this problem by manipulating numbers.Real problem here is :
we have 2 effects and each feed each other
prosperity -> construction (construction is calculated from prosperity directly)
construction -> prosperity (by default project housing)
So this creates a loop. prosperity -> construction -> more prosperity -> more construction
But I still believe the root of this prospersity/food-related issues lies elsewhere.
If by design the development team expects a normal Bannerlord playthrough run for at most 20-30 years(in-game), this approch is ok, but family/clan system looks redundant if this is the case. If the development team expects a normal Bannerlord playthrough can run for generations, then it's not good idea to put a soft cap that can be reached in one generation on prospersity of fief. (Though tbh current content of Bannerlord can't support such a long playthrough) Well, surely it's possible to make the soft cap only reachable after generations by doing nothing but numerical manipulation, but then this part of the game will be boring. Player cannot intereact much with this part of game, simply sit there and watch. I'm not major in game design during university, but my friend recommends a textbook for game design major called 'player make decisions', I really like this idea, player should be allow to make decision so they can intereact with the game world.
So my point is that player should be given more space to manage their fief, like food production.