On Smith On Cowen

Update: Incredibly embarrassing and rude misspelling corrected all credit due to Robert in comments who sure is different from Robert who is now typing.

Noah Smith writes better than I can and in particular that

Nobel-winning economist Paul Krugman recently claimed that John Maynard Keynes — or the general idea of Keynesianism — is winning the battle of opinion in the public sphere. George Mason University economist and blogger Tyler Cowen responded, playing devil’s advocate, and listed a bunch of points that he thinks indicate that Keynesian ideas (liquidity traps are important, austerity during recessions is destructive, fiscal stimulus is useful) might be losing the substantive battle.

Econ blog arguments are fun, so I’ll play devil’s advocate squared — angel’s advocate? — and go through some of Cowen’s points.

Sure sounds like fun. Also Tyler Cowen is on my very short list of reasonable and reasonably honest conservatives (in fact, as not very far as my memory serves me, it is entirely possible that “Tyler Cowen” *is* my list of reasonable and reasonably honest conservatives).

Acting as devil^2 advocate, Smith concedes two points to Cowen

Here is Cowen:

Keynesians predicted disaster following the American fiscal sequester, and the pace of the recovery accelerated.

This is intriguing especially in light of the fact that everyone blames a 3 percent sales-tax hike in April for crashing the Japanese economy. If that small tax increase sent the Japanese economy into a tailspin, why didn’t the across-the-board spending cuts of the sequester inflict similar damage on the U.S. economy This does make me doubt the standard Keynesian story.

I wonder which Keynesians predicted disaster. I certainly didn’t and I don’t recall Paul Krugman doing so. I thought the policy was obviously bad and the opposite of what should be done, but I certainly didn’t expect it to cause a recession. Here (and elsewhere) Cowen appears to use constant growth as a baseline. But the recovery has consistently disappointed (except for the “morning in America” surpassingly rapid turnaround when ARRA spending started).

Doing better than amazingly bad isn’t enough to make me doubt the standard Keynesian story. To be intrigued I would have to see economic growth higher than that predicted by a Keynesian model.

Cowen seems to consistently use constant growth as a benchmark. This is odd. To argue by analogy, I note that Krugman predicted a u shaped recovery and a u curves up, that is, accelerates. More seriously, there are a number of factors which should imply accelerating growth. One is the declining stock of housing per adult. The very low rates of house construction mean this falls over time. That should cause pent up demand for housing and accelerating construction. Another is deleveraging — debtors who are no longer considered credit worthy and can’t roll over their debt have been gradually paying it back (or having it cancelled by foreclosure or bankruptcy). This means that the demand depressing effect of their very binding liquidity constraint is weakening. The recovery improves state and local government budget balances and this affects their spending with a lag due to the annual budget cycle.

Now sequestration might not have had a dramatic effect on US real GDP. It also didn’t have a dramatic effect on US real government consumption plus investment (G). If you didn’t know when sequestration occured, could you figure it out from this graph ?


What I see is declining G during the long consistently disappointing recovery (until 2014q2 see below). I mean I also see the ARRA which started just before the recovery started.

Oh and finally, growth didn’t accelerate after sequestration which started March 2013, 2013q2 growth was lower than 2013q1 growth. Growth since sequestration has averaged 2.7%/yr which is very low given the output gap. Cowen is using the whatever lag I please operator.

On Smith “small” ??? 3% of consumption is 2% of GDP even in Japan. The ARRA was less than 2% of US GDP over the 2 years it lasted. The consumption tax increase was a huge shift in fiscal policy which dwarfs sequestration. That might explain the difference in the effects.


Japan is in a (supposed) liquidity trap, but negative real shocks have not in fact helped their economy, contra to the predictions of that model (start with here and here). Nor does anyone think that the bad weather in the first quarter of U.S. 2014 was good for us, although a basic liquidity trap model implies it will boost inflation (beneficially) because the supply restrictions lead to price hikes which tax currency holdings and thus boost aggregate demand. Come on, people, that is weak.

This is a good point (and I also like the use of the rhetorical device “Come on people, that is weak”). The main academic models that support the Keynesian story have tons of holes in them, as all academic macroeconomic models do.

Wait now Cowen appeals to “Nor does anyone think” not data. In fact, US GDP growth accelerated markedly after the bad weather with annualized quarterly growth of 4.4% and 4% in the second and third quarters. The second quarter is bounce back following the thaw, but third quarter growth is actual good news for a change.
Appeals to what everyone thinks have no place in any social science. It is not OK to ignore all of the data relevant to a claim when making it.

I am not using the any lag I please operator — the story for stimulative effects of snow works through real interest rates (as monetary policy does). There is a robust consensus that monetary policy works with a lag of about 6 months. 2014q3 is the point to look at to see the effect of a shift in real interest rates in 2014q1. I don’t believe in the snow as stimulus story, but it fits the data vaguely invoked by Cowen.

Also recall real government consumption plus investment. A casual look at the data show quarter after quarter of declining real G and disappointing growth followed by two quarters of increaseing real G with extremely rapid GDP growth. I suppose it is possible to imagine a better fit between the data and Keynes story, for someone who has a much more powerful imagination that I have.

I’d say that Cowen is playing heads I win, tails you lose, and losing.

On Smith, I agree that New Keynesian models are no good. But I ask is the problem the Keynesian or the new ? Old Keynesian models are not academic any more — they are not allowed in peer reviewed journals. They are also used by Krugman (along with new Keynesian models and he only believes either when they agree). What empirical failures do they have ? As far as I can tell, they fit the data fine.