Saturday, June 05, 2010

Scott Sumner meet Tyler Cowen

update: Cowen did not mix up supply and demand. I apologise.

Sumner identifies "expectations about monetary policy in the indefinite future" with "monetary policy" *and* assumes that monetary policy is not currently limited by the zero lower bound on nominal interest rates. Thoma has a reasoned criticism of Sumner whose assertions are plainly nonsense. Attempting to make sense of his claim, I assumed he identified investment with aggregate demand.

In fact, he was ignoring the current federal funds rate when discussing monetary policy and asserting that, if the Fed can't control expectations about monetary policy in the future, then it must pursue the most damaging possible monetary policy in the present. That is, he feels that an assertion that the Fed can't do it alone, since safe short term interest rates are currently approximately zero is equivalent to the assertion that the Fed can't undo the effects of a fiscal stimulus. These totally different assertions are asserted by Sumner to be just the same.

I'd say I was much too gentle with Sumner. Also, Prof. Sumner thanks for quoting me at length, linking, and pointing out a spelling error. Your insinuation that I suggested that the national accounts identity tells us something about causation has nothing to do with anything which I wrote. Readers please check. I think you will see that Sumners criticism has nothing to do with anything which I wrote.

end update:

Between the two of you, you are three fourths of the way to understanding the main national accounts identity.

Let me refresh your memories Y = C+I+G+NX

Tyler Cowen mixed up the concepts of demand and supply when he argued that uncertainty causes low investment and that this is low aggregate supply (via Brad DeLong). In the medium term, but not the short term, low investment causes lower aggregate supply. Low investment, right now, means low aggregate demand, right now. This is the strange case considered by uh Keynes in the General Theory. Cowen argues that, if uncertainty is causing low investment, then fiscal stimulus will have a "marginal" effect. This is an unfortunate choice of words, as Cowen* is also confusing levels and derivatives (that is level and marginal effect). Uncertainty implies low I and low Y for given G. It does not imply a low effect of G on Y.

Not to be outdone, Scott Sumner forgets that C, G, and NX contribute to aggregate demand and argues that a fiscal stimulus works only if people believe it will work (Via Paul Krugman and Mark Thoma). Evidently, he equates stimulating the economy with stimulating investment. It is possible he assumes that G crowds out C completely unless people believe the stimulus will work. This is, of course, nonsense. If people assume that they will pay the full increase in G with higher taxes (ignoring how the increase can partly pay for itself by increasing Y and tax revenues) then the stimulus will work unless the increase in G is permanent or G and C are perfect substitutes.

Behind the apparently diametrically opposite errors, there is a common perspective -- the only thing that matters is private investment and therefore the confidence of businessmen.

The Economist is determined not to deprive its readers of this perspective inviting both of them as guest contributors.

I think they are both really working from the sole methodological a priori that further fiscal stimulus would be bad policy so they consider any argument valid if the conclusion is that fiscal stimulus is a bad idea. I do stress the word "sole." If the distinctions between supply and demand, levels and derivatives and investment and aggregate demand are optional, then clearly nothing at all matters but getting the right conclusion. I mean right as opposed to left.

*spelling error corrected.

Update: Oh my this post has gotten more attention than I expected with links from Thoma and Sumner. I posted here rather than at www.angrybearblog.com, because I thought the post was venting and not ready for prime time. I was right.

pulled back from comments

Comments:
Robert:

I thought that Tyler had mixed up AD and AS too, when I first read his post. The usual argument is that decline in investment immediately reduces AD, and only later (because the capital stock is growing more slowly than it otherwise would be) causes a decline in AS (or AS to grow at a slower rate).

Then I read the Pindyck(?) paper Tyler linked to. In that model, an increase in uncertainty really does cause an immediate decline in AS, and not through the normal channel of slower growing K.

Every period, some sectors have a positive, and some a negative, productivity shock. If firms reallocate labour from negative to positive sectors every period, aggregate productivity stays the same over time. But if firms stop reallocating labour, aggregate productivity immediately falls in that same period.

Because there are costs to reallocating labour, it's an investment decision. An increase in uncertainty causes that investment in reallocating labour to fall (because it increases the chance you would want to reverse it next period). With less reallocation of labour, aggregate productivity falls immediately. So does AS.

# posted by Nick Rowe : 10:

2 comments:

Nick Rowe said...

Robert:

I thought that Tyler had mixed up AD and AS too, when I first read his post. The usual argument is that decline in investment immediately reduces AD, and only later (because the capital stock is growing more slowly than it otherwise would be) causes a decline in AS (or AS to grow at a slower rate).

Then I read the Pindyck(?) paper Tyler linked to. In that model, an increase in uncertainty really does cause an immediate decline in AS, and not through the normal channel of slower growing K.

Every period, some sectors have a positive, and some a negative, productivity shock. If firms reallocate labour from negative to positive sectors every period, aggregate productivity stays the same over time. But if firms stop reallocating labour, aggregate productivity immediately falls in that same period.

Because there are costs to reallocating labour, it's an investment decision. An increase in uncertainty causes that investment in reallocating labour to fall (because it increases the chance you would want to reverse it next period). With less reallocation of labour, aggregate productivity falls immediately. So does AS.

Robert said...

Thanks Nick. I put posts here rather than AngryBear because they are not ready for prime time. I use this blog to vent (and I have been in need of venting like the top hat in the gulf -- don't watch the video -- it's frightening).

OK so Cowen is not mixing up aggregate demand and aggregate supply.

However, I still suspect that he is using a model in which uncertainty doesn't affect aggregate demand, because investment in fixed capital isn't important. This is not a realistic model.

One can tell aggregate demand from aggregate supply by looking at the price level. If the problem is that firms aren't willing to hire and train workers, then firms have very high marginal cost (they making current workers work overtime). So they should raise their prices.

The argument that output is limited by aggregates supply is the argument that increased nominal aggregate demand will cause higher prices not higher real output. With core inflation 1% and falling, there is no evidence to support Cowen's argument.