Thursday, June 18, 2015

Still angry about Sumner's analytical garbage

Even after writing it in pedantic logical symbols, I'm still upset by the analytical garbage that Sumner is foisting on the world. But I think you can see where Sumner does his slight of hand if you realize he means two different things by "independent central bank". One is "independent central bank" the other is "monetary policy is effective". In Sumner's monetarist view, these two statements are identical. An independent central bank has effective monetary policy. In the Keynesian view, they are not. A central bank can be independent, but the zero lower bound means monetary policy is ineffective.

Here's Sumner's characterizations in terms of monetary policy effectiveness: 
  1. Keynesian: Fiscal austerity is contractionary at the zero bound regardless of whether [monetary policy is effective].
  2. Market monetarist: Fiscal austerity is contractionary if you lack [effective monetary policy]. Fiscal austerity would not be expected to have much effect if you have [effective monetary policy], due to monetary offset.
Note that the monetarist view is the only one that continues to makes sense with this replacement. That's because "zero lower bound" means "monetary policy is not effective" in the Keynesian view:
  1. Keynesian: Fiscal austerity is contractionary [when monetary policy is ineffective] regardless of whether [monetary policy is effective].
Which means we can actually shorten this:
  1. Keynesian: Fiscal austerity is contractionary [when monetary policy is not effective].
  2. Market monetarist: Fiscal austerity is contractionary if you lack [effective monetary policy]. Fiscal austerity would not be expected to have much effect if you have [effective monetary policy], due to monetary offset.
Then there is fact that the modern Keynesian view actually believes in monetary offset when you're not at the zero lower bound ... that is: "away from the zero lower bound" and "monetary policy is effective" mean the same thing. So let's add that piece on ...
  1. Keynesian: Fiscal austerity is contractionary [when monetary policy is not effective]. [Fiscal austerity isn't contractionary when monetary policy is effective.]
  2. Market monetarist: Fiscal austerity is contractionary if you lack [effective monetary policy]. Fiscal austerity would not be expected to have much effect if you have [effective monetary policy], due to monetary offset.
These two statements effectively say the same thing! So how do Sumner and Sadowski view them as different? Because they don't believe you can make the replacement:

"zero lower bound" = "monetary policy is not effective"

But that's the entire liquidity trap model!

That's why they can keep in countries that aren't at the ZLB in the final graph -- zero lower bound has nothing to do with monetary policy effectiveness.

That's also happens to be the reason they think they can throw out the countries without independent central banks! They think they can take the "regardless" out of the Keynesian formulation and say that it must be true if the central bank is independent. But then they believe that monetary policy is effective if the central bank is independent. So they've constructed a contradiction:
  1. Keynesian: Fiscal austerity is contractionary [when monetary policy is ineffective]  [if][monetary policy is effective].
It's all because Sumner and Sadowski believe:

"zero lower bound" = "monetary policy is not effective" is FALSE
"independent central bank" = "monetary policy is effective" is TRUE

That is to say, they assume the market monetarist model! In the Keynesian model:

"zero lower bound" = "monetary policy is not effective" is TRUE
"independent central bank" = "monetary policy is effective" is ambiguous

In order to make this true or false, you need to add a bit about the ZLB.

"independent central bank" and "ZLB" = "monetary policy is effective" is FALSE
"independent central bank" and not "ZLB" = "monetary policy is effective" is TRUE

But then market monetarists don't think the ZLB is important, so they don't see the problem with leaving it off!

And round and round we go ...

The only way to make sense of this is that Sumner and Sadowski don't understand the liquidity trap model. Krugman understands the monetarist model -- it's the one he uses when you're away from the ZLB!


PS. You can replace independent with "awesome"
  1. Keynesian: Fiscal austerity is contractionary at the zero bound regardless of whether you have an [awesome] central bank.
  2. Market monetarist: Fiscal austerity is contractionary if you lack an [awesome] central bank. Fiscal austerity would not be expected to have much effect if you have an [awesome] central bank, due to monetary offset.


  1. is these kind of scatter plots that Paul Krugman is so fond of that made me start giving failing grades to most bloggers. But really I dont think you should get so upset about the countries not at the zero New Keynesian models the same effect would happen anytime the interest rate is incorrectly, set too ultimately the graphs are ok in that regard....although Krugman doesn't believe in that aspect of new Keynesianism...I presume because that gives enough data points to reject most new Keynesianism models...

    But more importantly, divide the y intercepts by 5..what average annual growth do they imply? That should be enough to tell us that these graphs are garbage...

    1. I agree that these plots are somewhat ... naive. You really need a full model with a counterfactual.

      I am ceding the discussion of whether the data sets show anything (taking the data as a given) -- what upsets me the sheer bloody-mindedness of assuming market monetarism is right and Keynesianism is wrong and they using those facts to cull the data to show that indeed monetarism is right and Keynesianism is wrong.

      And yes, Sadowski just takes the difference between the current account balance in 2009 in percent and the current account balance in 2014 in percent. If you had a deficit of -10% of GDP in 2009 and a surplus of 2% in 2014, you had 12% of GDP in "austerity".

      And the dude apparently just got a phd!

    2. The importance of the ZLB is that the central bank can not return the nominal interest rate to its natural value. The only two monetary "policies" that New Keynesian models (or just about any microfounded representative agent macro model that allows monetary policy to impact the economy) say can work at the ZLB have to do with expectations management: either increase the rate of expected inflation or increase expected consumption (by holding interest rates low for longer than normal). I have a more in depth explanation of my reasoning here but basically the only theoretical argument market monetarists have when it comes to conventional models is that central banks have complete control over expectations.

  2. In the FWIW department:

    I don't read Sumner, despite the fact that he seems to have some good ideas, mainly because of the next statement.

    Logic does not seem to be his long suit.

  3. My understanding of your view is now:

    The simplest statement of the 2 views would be:

    1. because of monetary offset fiscal policy is never effective. This is the MM view
    2. at the zlb monetary offset does not work. (this is common to all Keynesian views)

    Note: it is assumed you need an independent central banks to carryout monetary offset.

    Sumner claims to have found evidence in the data that 2 is false as countries with independent central banks have less contractionary outcomes than those that do. It seems he has included countries not at the ZLB in his data, therefoer the results are irrelevant to statement 2 and he has not really dented the Keysian view in the way he claims

    I have 2 questions:

    1. Have you asked Sumner for clarification
    2. What happens if you throw out countries not at the ZLB for his data set ? Do the results chnage ?

    1. Hi Rob,

      Sumner has actually said the same thing on several occasions (that I referenced in the earlier post). I'm pretty sure his mind is made up. Besides the way to win arguments is to convince people listening to them, not people making the arguments -- the latter never change their positions. Max Planck has a good quote:

      "A new scientific truth does not triumph by convincing its opponents and making them see the light, but rather because its opponents eventually die, and a new generation grows up that is familiar with it."

      If you just throw out the non-ZLB countries in the last graph (even just Iceland), the remaining data shows a downward trend again -- but I would say that it is inconclusive (it doesn't support either model) at that point: not enough data points are left!

      But that isn't the entire point -- you can't throw out the EU countries (or consolidate them into a single country, essentially weighting them by a factor of 1/19 ~ 0.05) in order to disprove the Keynesian view because in the Keynesian view, there is no reason to throw them out.

      There are two steps here: eliminate the data that is relevant to the Keynesian view (countries without independent central banks) and then include data that is irrelevant to the Keynesian view (countries not at ZLB).

      Nothing about Monetarism and Keynesian liquidity trap can't be shown on the same graph: they make different assumptions about the relevant data. Keynesian view looks at ZLB countries and monetarist looks at independent central banks (ICB). ZLB and ICB are not equal sets and their intersection doesn't include enough data to say much of anything.

    2. Jason, regarding a plot with only ZLB and ICB countries: I asked Mark about that and here's his response
      Essentially he says you'd only have seven points left: US, UK, Sweden, Switzerland, Euro Area, Czech Republic and Japan, resulting in a p-value of 0.6705.

    3. Now if I knew what a p-value was I'd be in good shape. I take it from his response there and in the above responses and his statement that the slope (which is positive for those) doesn't matter in that case that a p-value of 0.6705 is too big to be able to draw any meaningful conclusions from the data set.

      And I just (reread?) this to (re)educate myself:
      So now I know.

    4. Was the p-value lower with these counties included?

      I just realized that the current account balance actually just means size of recession in the monetarist view when you have independent central banks. There should only be a loss in revenue due to the recession if the central bank is "doing its job".

    5. Reading through some of the other comments, I think the p-value was higher before eliminating the non-ZLB countries. Looks like it was 0.9816.

    6. That is pretty funny.

      And the thing is that the y-axis conflates economic growth and changes from trend. High growth trend countries are near the top and low growth are near the bottom.

      The question is not whether austerity turns you into a low growth country (or low growth counties engage in austerity). The question is whether austerity impacts growth relative to your trend.

      The graphs don't show what the purport to show at all.

      Sumner has proven that austerity doesn't turn you into a low growth country. As most low growth countries are advanced countries, he is basically saying austerity doesn't turn you into an advanced country.

      That's something I'll believe!

    7. Jason, I hope you don't mind that I've been incorporating concepts I've encountered here into questions on Sumner's blog w/o crediting you. At my level of understanding, I don't really have a right to an opinion... so I'm basically just asking questions. I was worried that your fun and interesting titles here might negatively affect their incentive to respond to me and/or color their responses. Also, it looks like commentator "OhMy" has gone ahead and posted a couple of links to your posts here anyway.

    8. Ha! -- it turned out to have some impact!

  4. Jason, I don't know if you've ever read my blog but I've been trying to warn about Sumner's analytical sophism for ages. Glad you're angry now about what he foists on an unsuspecting world. Join the club.

    1. I remember you commenting here long ago. I hadn't seen your blog, but I will check it out.

      I'm not sure when it was, but at one point I figured out that there is really no content to the theory that is "market monetarism". There is content to ordinary monetarism (it's a falsifiable theory). They even used to think V was stable in MV = PY. But V turned out to be unstable, so they said it could vary. Finally Sumner comes along and says its not M but expectations of "monetary policy" and that P doesn't matter so you end up with:

      "monetary policy" = NGDP

      It's quite the trick.

    2. Well I appreciate someone else finally being annoyed by Sumner's sophism. For a long time I felt like Krugman should even answer him just once a month-just to make sure his disinformation campaign isn't wholly unchecked.

      I appreciate someone like yourself who seems to have pretty imposing theoretical underpinnings joining the fray.I'm just a layperson who speaks only English-which makes it easy for some in the mainstream to dismiss what I say as 'mere literature.'

      I think of it as a serious social benefit to correct the record here.

  5. Anyway what I'm trying to say is it's nice to have someone who's pretty sharp joining the fray as I've felt like a voice in the wilderness on Sumner. I've also felt like many Keynesians/liberals like Krugman don't take him seriously enough-I think he has made a real impression on a lot of people who knew nothing about economics before reading him.

    If when you get a chance you could specifically read the link I posted above I'd be very appreciative. I'd be curious if you think my specific criticisms in that post hit the mark.

  6. "zero lower bound" = "monetary policy is not effective"

    The thing to remember here is the MM argument is that it's wrong to say 'monetary policy is not effective at the ZLB' but 'conventional monetary policy is not effective at the ZLB'-by conventional policy they mean interest rate policy by the Fed.

    The premise is that if you expand monetary policy to include nonconventional monetary policy then they claim the statement remains true-monetary policy works at the ZLB as well if the Fed do QE or even better NGDPLT.

    To move from an inflation target to a NGDP target in theory might seem an upgrade-as monetary policy can respond then differentiate between supply and demand side shocks.

    However, the trouble is with NGDP is how do you know if the CB has hit its target? Sumner's answer is an NGDP futures market-and idea that hasn't been able to get off the ground despite the financial assistance of Ken Duda.


Comments are welcome. Please see the Moderation and comment policy.

Also, try to avoid the use of dollar signs as they interfere with my setup of mathjax. I left it set up that way because I think this is funny for an economics blog. You can use € or £ instead.