He has made that claim in numerous blog posts, but this one is typical, and it’s entitled “For the 247th time, the fiscal multiplier is roughly zero.” The 247 is presumably there to patronise dimwits like me and others who don’t agree with him. Anyway, let’s examine Sumner’s argument.
His reason for saying that the fiscal multiplier is zero is set out in the first two sentences of his post: “Keynesian economists have never been able to accept my assertion that the fiscal multiplier is roughly zero because the Fed steers the (nominal) economy. There’s a mental block on their part…”
Well now a very good analogy to the latter argument can be made with those cars with two steering wheels and designed for learner drivers. The learner controls the car most of the time, but the instructor can take over by using the other wheel (e.g. in an emergency). Now you could say that the learner is not in control in that the instructor has the power to negate what the learner does. But does that mean that what the learner does is pointless or has no effect? Does it mean that really it’s the instructor who is “steering the nominal path” of the car (to use Sumner’s phraseology)?
Well in the case of a learner who is clued up, the instructor wouldn’t interfere. And in that case it’s the learner that is driving the car. It’s amazing that I need to spell this stuff out, isn’t it?
Same applies to monetary and fiscal policy. If fiscal stimulus is implemented, and the central bank thinks that stimulus is desirable, then the central bank won’t negate it. So in that case that particular bout of fiscal stimulus will work. Indeed, during the recent crisis, central banks (much to the surprise of Sumner, perhaps) AUGMENTED fiscal stimulus by cutting interest rates.
Which is better: monetary or fiscal stimulus?
That of course all leaves the question as to which is better: monetary or fiscal stimulus. My answer to that is that PURE monetary policy (i.e. relying just on interest rate adjustments or QE) is not a good idea. My reasons are here, hereand here.