Karl Smith is talking about it again. This is an idea that Nick Rowe brought up a little while ago.
The logic is that a downgrade would shift the IS curve in an old Keynesian model out to the right, which should increase output. The problem that we're dealing with, though, is that government bonds and cash are interchangable. Downgrading bonds makes them less interchangable with cash. Shouldn't this increase the demand for cash? Nick and Karl point us to basic macro, but let's not forget basic micro. When demand for cash is strong and a cash substitute becomes less like cash, money demand is going to increase, right? The LM curve is going to shift to the left as the IS curve is shifting to the right. Where is the expansion in this downgrade?