Monday, March 11, 2019

MMT cannot produce forecasts nor accurately fit data

FRBNY DSGE inflation forecast, January 2019. Objectively better than MMT, even if it's wrong.

Over the weekend, I wrote up a bit on my views of MMT including my critiques of it. I got some comments that it was too long or that it wasn't clear what my critique was. Let me try again with a different — and more practical — take.

Modern Monetary Theory (MMT) cannot produce forecasts nor accurately fit data and is therefore far inferior to even the most objectively inaccurate Dynamic Stochastic General Equilibrium (DSGE) model.

10 comments:

  1. This is what I call a succinct, devasrdevas crutique. Well done.

    ReplyDelete
    Replies
    1. Thanks, Anti.

      (Part of the inside joke is that even an AR process can do better than a DSGE model.)

      Delete
    2. Yes, I recall you demonstrating that, and it's not a joke!

      Delete
  2. I really like and appreciate your post.Thanks Again. Keep writing.

    ReplyDelete
  3. Clear and succinct post. Surprising that you're not getting more push back in the comments. How about on Twitter? You mention comments regarding your last MMT post. Were those on Twitter?

    ReplyDelete
    Replies
    1. Few people seem to comment on my blog anymore. Most of the interactions are on Twitter via tweets and DMs (or plain email).

      Delete
  4. Hi Jason, love your blog.
    I have been reading some of your old posts, and if I understand correctly, you are of the view that, economics hasn't had its Newtonian revolution which differentiates it from "hard" sciences like physics.
    Am I correct?
    If so, do you think that more focus should be on things understanding the workings of human brain, which might explain human actions in a better way.

    ReplyDelete
    Replies
    1. I'm going to guess Jason will say "no" to your question. That might be like studying quantum physics to determine the trajectory of a baseball: wrong scale.

      Delete
    2. Thanks CJ! Per your first question, I'd say yes — it's really about the appearance of a framework that organizes a bunch of observations. Newton gets a lot of hype because it's one of the first frameworks like that at all (besides something like religion). Darwin created such a framework. Supply and demand is a kind of framework, but is limited in the sense that it's not clear how it applies at the macro scale. And I don't think that a good framework will turn econ into a hard science — I imagine that there are parts of econ that will always be "soft science". But what a "Newtonian revolution" would do is help understand which part is which.

      Per your second question, in a sense Tom (By The Way) is right because there plausibly exists an effective macro theory that "integrates out" the individual human degrees of freedom. That said, there are failures of information equilibrium (non ideal information transfer) that seem to depend on how humans behave — panicking in financial crises, e.g. — that affect the outcomes at the larger (macro) scale. Human social & cultural factors and how people think will feed directly into things like why the IT index is different in the UK for the unemployment rate than it is in the US.

      It will also feed into why humans do things that differentiate among themselves by gender, class, race that lead to all kinds of inefficiencies since information entropy tends to be maximized for uniform distributions. Psychology and neuroscience will underlie sociology which, will in turn inform why we observe certain things and not others in economic systems.

      Delete
  5. Jason,

    I thought I would draw your attention to this, if you haven't seen it already:

    https://www.nber.org/papers/w25655#fromrss

    ReplyDelete

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.

Note: Only a member of this blog may post a comment.