17 Oct 2015

What Evidence for Market Monetarism Sumner Giveth, Sumner Taketh Away

Scott Sumner 23 Comments

There’s no way I can condense the argument down to a few excerpts. In order for you to understand my point here, you’re going to have to roll up your sleeves and read Scott Sumner’s whole post, in which he presents two different lines of evidence to argue that “real” shocks don’t cause unemployment while slowdowns in NGDP growth do.

Then, after you’ve read his post, you can come back here and see my analysis. At that point, if you already had your doubts about Sumner, you will say, “Holy smokes, Bob is right. That’s hilarious!” On the other hand, if you’re a fan of Sumner, you will say, “I can’t believe I let Bob talk me into cooking a 15-minute nothing burger.”

So if you feel like taking this journey with me, go read Sumner’s post.

 *  *  *

OK, welcome back.  Let me now make a series of statements about Sumner’s post (and the fallout in the comments):

==> (1) Sumner’s whole purpose with this post is to argue that shocks in “real” factors can have huge impacts on welfare. However, they do not correspond to the business cycle. So long as the central bank exercises wise monetary policy, real shocks can be offset and full employment can be maintained. In contrast, we don’t need a real shock to get a recession and rising unemployment; all we need is the central bank to stupidly let NGDP growth fall below trend.

==> (2) The first line of evidence for this position Scott offers is Australia. Now Australia is a commodity exporter, so if real shocks mattered for the labor market, you would expect the commodity bust (i.e. steep fall in commodity prices over the last 14 months or so) to have caused unemployment to rise in Australia, But actually, Scott tells us, Australian unemployment has no discernible trend over the past year. Indeed, if you look at a longer history, you’ll see that Australian unemployment was gently rising before the commodity bust. That rise in Australian unemployment was (Scott tells us) probably due to below-trend NGDP growth.

==> (3) The next line of evidence is Texas. With the collapse in oil prices, you might have thought that unemployment in Texas would shoot way up–that’s what would happen if “real” things mattered. But nope, it has continued to gently fall. So again we see that real shocks don’t cause unemployment, it’s monetary policy, stupid.

==> (4) The funny thing is, Scott never believed that either of these items had anything to do with labor markets in the first place, so it’s weird that he’s invoking them as a “test” of the “real shocks” doctrine. In the case of Texas, in the main post itself Scott writes: “[R]eal shocks can have a modestly larger impact on Texas RGDP, as the drop in oil output affects RGDP more than employment.  It’s a capital-intensive industry.” He also earlier explained that “most of the Texas growth” of recent years isn’t due to oil, but other factors having to do with wise Texas government policies.

==> (5) Things are even weirder with Australia. Since unemployment was rising in the past, and Scott blamed below-trend NGDP growth, and in the last year was flat (give or take), you would assume that NGDP growth has returned to trend. But nope, that’s not what happened. As Justin D pointed out in the comments:

NGDP growth in Australia was extremely slow over the past year (2014Q2-2015Q2), up just 1.6%, slower than the 3.6% growth over the prior year, and yet unemployment was rising in the prior year and stabilized in the most recent year. How do we account for that in the market monetarist framework?

To this, Scott replied:

Justin, Good question. The decline was due to sharply falling prices of commodity exports, which has little impact on employment. That’s one reason I suggest that commodity exporters target total labor compensation, not NGDP.

You really have to stop and let that sink in. Rather than belabor the point, let me illustrate with an analogy:

Suppose I’m making the case that, contrary to conventional wisdom, moviegoers get more satisfaction from watching 3D movies with their special goggles in their laps, rather than wearing them. To illustrate my point, I cite the case of my cousin, who just watched the 3D version of “The Martian” and said he really loved it.

Someone on a lark decides to go ask my cousin if he wore the 3D goggles or left them in his lap. My cousin says, “I wore them, duh.” Puzzled, the questioner then asks me to explain this discrepancy: why did I cite my cousin in support of this rule, when in fact the opposite occurred? I answer, “Good question. My cousin is one of the rare people who has an oddly shaped head that really fits those goggles well. For people like him, my rule is that for maximum movie experience you want to wear the 3D goggles on the tip of your nose.”

In this (ridiculous) scenario, how would people feel about me citing my cousin to prove the rule?

23 Responses to “What Evidence for Market Monetarism Sumner Giveth, Sumner Taketh Away”

  1. Khodge says:

    Oh, that was painful.
    First, three charts with three different times? Not just poor optics but outright misleading; the most obvious sign that there are problems with the data.
    Second, as you pointed out, it doesn’t give me much confidence when the premises is modified with statements like: there are other things going on but I am going to cram my answer into the box I built.
    As an aside or an addendum to (1) above: There is a discernable pattern in the first chart: the first month of the quarter is the largest with the next two months declining. Not, I’m guessing, what he was trying to show us.

    • TravisV says:

      TravisV here (Scott Sumner fan).

      Re: Justin D’s question about slow NGDP growth in 2014Q2-2015Q2, I suspect the “natural rate hypothesis” might explain why unemployment didn’t surge. See here:


      “This is the natural rate hypothesis in action. Even though NGDP growth remains slow, and indeed is getting slower, wage moderation does eventually allow the labor market to heal.”

    • Tel says:

      Oh, that was painful.

      After making me read that, Bob better take back his comment about Tom Woods being a slave driver.

  2. E. Harding says:

    I guess Sumner thinks non-commodity related NGDP matters more than NGDP as a whole. But if so, Russia’s non-commodity related NGDP in Q4-Q1 2014-2015 must have skyrocketed: do you see any slowdown in Q4 here?
    Yet, unemployment there still rose by 1.1 points, August-April, and everybody admits a recession took place between those months.

  3. E. Harding says:

    This, BTW, is Australia’s unemployment and NGDP, up to Q4 2014.


  4. E. Harding says:

    I’m a fan of Sumner, and I don’t think this is a nothing-burger.

  5. Transformer says:

    ” So long as the central bank exercises wise monetary policy, real shocks can be offset and full employment can be maintained”

    First I think is a mis-statement. I don’t think Sumner believes that real shock can be offset, but that nominal shocks are more significant than real shocks in the short-term and can be offset by good policy.

    Secondly, I think that “wise monetary” policy is policy that adjusts for these nominal shocks. NGDP targeting is an obvious example of such ‘wise policy” but it is certainly not the only possible one and is not a “one-size fits-all policy” for all countries. It certainly is not a policy practiced by Australia, and that clearly does not stop Sumner from thinking that Australia does have “wise policy”, at least compared to others countries.

    So: I agree that Sumner has left himself open to your critique by his references to Australia’s rates of NGDP growth, which is inconsistent with the facts and with his comment that “That’s one reason I suggest that commodity exporters target total labor compensation, not NGDP”. But overall I think we are looking here at a bit of terminological and analytical sloppiness in this particular post rather than a hole in Market Monetarists theory.

    I hope he responds to your post and clarifies.

    • Bob Murphy says:

      Transformer, when I say that Sumner believes wise monetary policy can offset a real shock, I don’t mean that it can cause cars to spring forth from printing presses. I mean it can offset the impact on the labor market.

      So if there’s an earthquake that destroys half the factories, no monetary policy can prevent real GDP from collapsing. But we don’t need to get 50% unemployment unless the central bank does something stupid like try to keep CPI inflation at 2%.

      (I am open to correction from Scott, but I’m pretty sure this is his view, at least stated loosely.)

      • Transformer says:

        I agree that bad monetary policy could make a real shock worse and that targeting inflation after a negative supply shock would be an example of bad policy.

        This doesn’t seem quite the same as saying that “real shocks can be offset” by good policy which implied to me that RGDP effects can be prevented , but its probably not worth quibbling over .

  6. Andrew_FL says:

    “Australia has no had a recession since 1991”

    “But for the business cycle, and especially for fluctuations in the unemployment rate . . . well, it’s all about the musical chairs model.”

    “Aussie unemployment had a mild upward trend before the commodity bust[…] That was probably due to the slower than trend rate of NGDP growth.”

    These three statements can’t all be true! If slowdowns in NGDP cause recessions, and Scott wants to attribute increasing Australian unemployment to slowing NGDP growth, then Australia has indeed had a recession since 1991.

    • E. Harding says:

      Nah, the rise in Aussie unemployment was too small and too long-term to be considered a recession.

      • Andrew_FL says:

        But it’s short term enough to be caused by NGDP?

        One of the above statements must be wrong. If it’s not statement 1 then it’s either 2 or 3. I don’t care which of the statements is wrong, but at least one of them is.

        • Tel says:

          The statement about the commodity bust is wrong, spot prices have been sliding down for several years, it started much earlier than Scott claims.

          • Andrew_FL says:

            A decrease in price is not the same thing as a bust. If you said we’d been having an oil bust for the past few years people would look at you like you were crazy.

            • Tel says:

              Oil is a lot more political than coal, but from a producer perspective (like Saudi Arabia) it sure is a bust. Those guys are just watching their bank accounts burn, while also trying to run a few proxy wars that are very soon turning unproxy … not a nice situation.

              One might argue the Saudis brought it on deliberately, because they could feel the OPEC system wasn’t delivering the value that it used to do and they wanted to kill off the weak producers. I guess the word “bust” kind of implies that it happened by accident, amongst the oil producers you could almost call it a “trade war” … not an accident in the least.

              Getting back to Australia, I don’t believe there was ever a dumping of either coal or iron onto the market, with the intention of price hammering and running the weaklings out of business. Nor for that matter was there a tech shock like when the alternative technogies (fracking, etc) hit the oil industry. What happened in coal and iron was a genuine demand slump.

              That said, this topic was supposed to be about employment and I doubt the guys turning spanners on rigs have a whole lot of control over the geopolitical goings on. So yeah, from the working guy’s perspective it’s gonna feel like a bust, regardless of the detailed cause.

  7. E. Harding says:

    Scott Sumner has responded (sorta unconvincingly).

  8. E. Harding says:

    Also, what happened to make the NGDP shock of the early 1980s in Australia have such a huge effect on unemployment and the equivalent NGDP shock of 2008-9 in Australia have such a mild effect on unemployment? It’s not something in the NGDP itself.

    • Bob Murphy says:

      E Harding, I’m assuming Scott would say something like, “The NGDP slowdowns that caused unemployment to go up in Australia weren’t due to energy price swings, so that’s why that particular NGDP trend had an effect on the labor market.”

  9. Tel says:

    The spot price of coal peaked in 2011 and has been falling since, which does seem to correlate pretty closely with the low point for Australian unemployment during 2011 and rising somewhat after that.


    I do think there are some dodgy aspects to the way that unemployment is measured. Anyone who finds it depressing being out of work for example, gets counted as disabled, therefore keeping unemployment low.

    • Tel says:


      You can poke around some of the suburbs around Perth (where most of the mining money circulates), there’s a trend graph available partway down the page, set it to “Annual” and you can see a dip around 2011 to 2013 in this example. The above is near the water so houses average around a million, but depending on where you look that dip might be a bit later. The middle of

      Mind you interest rates also shifted around, the recent uptick in house prices has been driven by falling rates (IMHO).

    • Tel says:


      There’s a lower class suburb of Perth, away from the water, houses only half a million, same dip around 2011.

  10. RPLong says:

    To prevent us all from reasoning from a price change, does anyone know why commodity prices were “sharply decreasing?” It wouldn’t happen to have been any sort of a real shock, would it?

    • Major.Freedom says:

      And to prevent us from reasoning from a spending change, does anyone know whyAHAHA just kidding, you can reason from a spending change and violate economic logic, because it encourages people to start one’s frame of reference from a controlled total spending perspective, which then encourages one to believe central banking “letting” NGDP change is a positive driver for why NGDP changes.

      Then we can ignore the real factors in the present, and monetary conditions in the past that affected those real factors.

      Then tralalala baddabing baddaboom, market monetarism is proven.

Leave a Reply