Paul Krugman writes:
A couple of weeks ago I went out to Lincoln, Nebraska (the AP graders - an institution you don't learn about until you're hawking a principles book) and they asked me to give my views on crowding out. I think my formulation is close to yours.
Here's how I explained it. Think of a standard IS-LM picture. Does that match current reality? Obviously not: the Fed doesn't target the money supply, so holding M constant is not a useful thought experiment, and actually confuses students. In fact, since the Fed actually targets the Fed funds rate rather than the money supply, you might think that the LM curve should be replaced with a horizontal FF curve. This would seem to suggest no crowding out at all.
But except in the very short run the Fed doesn't set the interest rate passively; instead, it tries to stabilize output around potential. A reasonable way to represent a Taylor rule or something like that in a simple diagram is to draw a vertical line, the BB curve (for Ben Bernanke). This gives us 100% crowding out.
And I think that's right. Except in liquidity-trap conditions or in the very short run, before the Fed has a chance to catch up, fiscal policy doesn't change aggregate demand, only the mix. The exceptions are important: we had a near-liquidity trap experience in 2003, and it was a good thing that we had some fiscal stimulus (and a bad thing that the stimulus was so poorly designed). But the normal rule is that fiscal policy is fully crowded out.
I think this is important. Fiscal prudence--budget balance--is a politically unpopular but economically important cause. The last thing we need is people claiming that budget deficits have, in our current economy, virtues they do not possess.