Tyler Cowen calls this, from Warren Buffet, the best argument for stimulus so far:
All you know is you throw everything at it and whether it's more effective if you're fighting a fire to be concentrating the water flow on this part or that part. You're going to use every weapon you have in fighting it. And people, they do not know exactly what the effects are. Economists like to talk about it, but in the end they've been very, very wrong and most of them in recent years on this. We don't know the perfect answers on it. What we do know is to stand by and do nothing is a terrible mistake or to follow Hoover-like policies would be a mistake and we don't know how effective in the short run we don't know how effective this will be and how quickly things will right themselves. We do know over time the American machine works wonderfully and it will work wonderfully again.
At least it's honest. And it may well be that merely being seen to be doing something is necessary--that aggregate demand will sag worse if the public perceives that there is no sheriff in town. But I keep coming back to one picture:
Will "stimulus" restore the economy to perfect health? Not unless you think government stimulus will sustain the massive private debt mountain we built up over the past 25 years.
Crashing asset values and shrinking GDP just make the debt ratios worse, even as households delever.
The idea behind stimulus is basically that the government will step in and take up the responsibility for the borrowing and spending that was being done by consumers, except instead of a Wii we'll get a high-speed rail line between LA and San Francisco, and hopefully the potholes filled in front of my house. At 0% interest rates, proponents argue, plausibly, that this borrowing is hardly going to crush the taxpayer under its onerous weight.
But that 0% is not on 30 year bonds; it's on shorter term debt that will eventually come due. What will our interest rate be when it's time to roll that debt over? It won't be pretty if the government is still having to fill in the output gap with heavy borrowing.
Stimulus is supposed to be, as Conor noted below, a short term and temporary strategy. But while it can ease the pain of a slowdown (at least in theory), as Tyler Cowen has been pointing out, the actual empirical evidence that massive government spending can shock an economy the size of ours into a permanently higher level of output is . . . well, it's sort of hard to find a wittily apt description of something that doesn't really exist.
There's a lot of solid Keynesian theory that says it will be so. But not that long ago we had a lot of pretty good theories from very smart economists about how this sort of financial crisis couldn't really happen again in the first place.