On top of my mind, there are two recent papers:

(1) Berk and Green has a JPE paper last year, attempting to "rationalize" flow and return. The basic arguement is decreasing return to scale and investor's learning of manager's skills.

(2) On this year's WFA meeting, there is one paper by a guy from Oregan discussing who is monitoring the manager - the guys with money flowing in or withdrawing the money. It is a kind of interseting but as he only gets data from one fund, it is hard to generalize the results.

I share with your view that flow and cross sectional return are quite interseting.