07 Sep 2016

Sumner vs. Kling: Don’t Get Hit By Those Goalposts

Scott Sumner 8 Comments

This is quite strange. It shows the versatility of economists and how they can always make it sound like they’re winning the argument, no matter what happens.

Earlier this month, Arnold Kling wrote:

Wisdom from Erik Hurst

He says,

The facts are real wages moved very strongly with employment across regions. Nevada was hit very hard by the recession, for example, while Texas was hit much less hard. Wage growth, both nominal and real, was about 5 percent higher in Texas than it was in Nevada during the Great Recession.

Pointer from Tyler Cowen.

The point is that we do not have a single aggregate economy. If you think that every state faced identical demand conditions, then the state with the higher real wage growth (Texas) should have had the worse unemployment. And Hurst goes on to point out how the regional data make it difficult to defend the view that wage stickiness is the cause of unemployment. In fact, he refers to work, which I noticed earlier, that suggests a PSST story.

So to avoid confusion: In the above excerpt, Kling starts out with a quotation from Erik Hurst, and then Kling says, “The point is that we do not have a single aggregate economy…”

I’ve put in bold “The point is” because Scott Sumner then proceeds to change what Kling’s point (allegedly) was. Sumner writes in response:

I’ve done a lot of posts on wage stickiness, but misconceptions keep popping up. The sticky wage theory of the business cycle is based on the notion that only a fraction of wages get adjusted each period. This leads to some odd results. The bigger the fall in the equilibrium wage rate, the bigger the fall in the actual wage. That’s no big surprise.

But what does seem to surprise people is that the bigger the fall in the equilibrium wage, the more that actual wages exceed equilibrium wages. That means that, according to the sticky wage theory, during periods where wages fall most the rapidly, we should expect to see the highest rates of unemployment. The same is true regarding cross sectional data.

Thus suppose that Nevada was hit by a severe real estate collapse, and its equilibrium wage fell by 10%. Also suppose that at the same time Texas was buoyed by an oil boom and its equilibrium wage rose by 4%. Finally, assume that due to sticky wages, the actual wage only moves by 1/2 of the amount that the equilibrium wage moves, in a given period. In this example, Nevada’s actual wages would fall by 5% and therefore end up 5% above equilibrium, leading to mass unemployment. Texas wages would rise by 2% and end up 2% below equilibrium, leading to a tight labor market.

Now consider this example, when reading the following comment by Arnold Kling, who is discussing some research by Erik Hurst…

And then Sumner quotes the excerpt that I put up, originally. Sumner then concludes, “Actually, that’s exactly what you’d expect if sticky wages caused business cycles. And for similar reasons, interest rates tend to be at their lowest when they are the furthest above the Wicksellian equilibrium interest rate.”

Do you see the problem? Kling was saying “The point is” that the regional differences in unemployment and wage growth show that we don’t have one giant, homogeneous economy that faces a single Aggregate Demand curve. Yes, Kling says that Hurst “goes on” to talk about wage stickiness, but there’s more to the argument–seriously, click the link and search for “stick” to see how many times “stickiness” or “sticky” appear. The guy’s estimating model parameters etc.; the quick snippet from Kling’s quotation wasn’t about wage stickiness.

OK, so thus far you might think that all I’ve done is show that Sumner changed what Kling was arguing about, with that specific quotation. But it gets much odder.

You see, Sumner is making it sound like his own model–which he’s been tirelessly explaining, and quite frankly is frustrated that so many other economists still don’t get!–is perfectly able to handle an outcome like a real estate bust hitting Nevada harder, and thus causing higher unemployment there. Right? That’s clearly what Sumner is saying.

And yet, back in 2010, Kling was offering his same “Recalculation Story” about the sluggish recovery, and how the housing crash meant workers were in the wrong spots in the country, and that’s why unemployment was higher in states that had had bigger housing bubble peaks. At that time, Sumner certainly didn’t say, “Of course, that’s exactly what you’d expect from my sticky wage model.”

No, at that time Sumner totally rejected Kling’s misallocation of resources story, and said his sticky wages / Aggregate Demand story fit the facts better:

There’s no question that Arnold Kling’s recalculation view is more intellectually appealing than the messy arguments about wage stickiness used by us “GDP factory” proponents…

Yes, macroeconomics should be all about specialization and trade.  Except business cycle theory, which needs a special ad hoc sticky wage/price model.  Why?  Because the evidence simply doesn’t fit any other approach.  Here’s Kling on the construction bust:

I want to suggest that the output that is “lost” is output that people do not want. In 2008 and 2009, Americans do not want 2 million houses to be built. So I do not think that it is right to speak of a shortfall in output. Instead, we should say that the people who were building houses have not found a pattern of trade in which they can produce something that people want.

Yes, housing output was low in 2009 and unemployment was high.  But is there a causal relationship?  I say no.  Housing starts peaked in January 2006, and then fell steadily for years:

January 2006 — housing starts = 2.303 million, unemployment = 4.7%

April 2008 — housing starts = 1.008 million, unemployment = 4.9%

October 2009 — housing starts = 527,000, unemployment = 10.1%

So housing starts fall by 1.3 million over 27 months, and unemployment hardly changes.  Looks like those construction workers found other jobs, which is what is supposed to happen if the Fed keeps NGDP growing at a slow but steady rate.  Then NGDP plummeted, and housing fell another 480,000…The reason housing fell far below normal is because the severe fall in NGDP created a deep recession.  Unemployed factory and service workers aren’t going to buy new houses.

Most importantly, the huge run-up in unemployment did not occur when the big fall in housing construction occurred, but much later, when output in manufacturing and services also plummeted.

I realize I threw a lot of stuff at you, and I am 99.4% sure that Scott Sumner would read all of his quotations above and respond, “Yes, I was dead right in 2010 and I’m dead right today. What’s your point?”

My point is that Sumner back in 2010 was saying Kling had to be wrong, and only the sticky wages story made sense, because (he argued) unemployment had nothing to do with housing.

Now, in 2016, Kling is pointing to unemployment being tied to regional housing markets, and Sumner is saying, “Yes, that’s exactly what the sticky wage model predicts.”

Both of those stances can’t be right.

P.S. If you want to see my reaction to the Kling/Sumner battle last time, read this. I think Kling threw in the towel prematurely.

8 Responses to “Sumner vs. Kling: Don’t Get Hit By Those Goalposts”

  1. Patrick Szar says:

    For an Austrian, you sure do give up a lot on method to your non-austrian friends. I get that you’re trying to show them how they contradict themselves. But you’re playing by their rules, which means, you’re just as arbitrary as they are. Either the method works, and can be discussed as such, or it doesnt, and should be thrown out.

    • Major.Freedom says:

      You can tell a crazy person that they are being logically inconsistent, and yet not “play by their rules”. You are playing by your own rules, you’re just also speaking to them violating their own rules.

  2. E. Harding says:

    Very sophisticated response, Bob. In any case, this is Sumner:

    Harding, You said:

    “It [NGDP in the EZ] never fell after 2009”

    If you keep wasting my tame, I will stop responded. You’ve been warned.

    Of course, I’m right. If Sumner thinks I’m pedantic, I’m no more pedantic than Bob or him at times.

    • Bob Murphy says:

      I vaguely remember him saying that to you E Harding. Does that relate to my post here or you’re just saying, “Sumner can’t handle the truth”?

      • E. Harding says:

        It relates to one of Sumner’s replies to Kling; doesn’t it?

  3. Major.Freedom says:


    By NGDP, Sumner meant Nevada Gross Domestic Product.

  4. guest says:

    “Unemployed factory and service workers aren’t going to buy new houses.”

    Unless you print money and use ACORN to agitate for the non-existent right of unemployed factory and service workers to own homes in the late 90s.

    I mean in 2008.

    *Takes a bow*

Leave a Reply