Wednesday, March 16, 2016

The ITE guide to schools of economics

Modern Monetary Theory = Standard macroeconomics + politics

Post-Keynesian economics = Standard macroeconomics + politics

New Keynesian economics = Standard macroeconomics

Market monetarism = Standard macroeconomics – liquidity trap

Real business cycle theory = Standard macroeconomics – sticky prices

**Information transfer economics = standard macroeconomics + empirical data


This was inspired by Simon Wren-Lewis on MMT. Consider that seconded. I did a series on Post-Keynesian economics (and the drudgery that is accounting approaches to economics) here, here and here. My last word on market monetarism was here.


PS None of these schools can do this with two equations and five parameters:


Update: This is a choice quote from Palley [pdf]:
The criticism of MMT is not that it has produced nothing new. The criticism is that MMT is a mix of old and new, the old is correct and well understood, while the new is substantially wrong.
I just don't understand why anyone would want to try to fix macroeconomics but be completely uninterested fixing the lack of comparing quantitative models to numerical data. It is baffling. To that end, here is the ITE guide to schools of economics view of empirical data:

Modern Monetary Theory = Ignore empirical data

Post-Keynesian economics = Ignore empirical data

New Keynesian economics = Ignore empirical data

Market monetarism = Consistent with any possible empirical data

Real business cycle theory = Ad hoc trend removal and reproduce moments of empirical data

**Information transfer economics = Use empirical data


  1. I love it! Poke those other schools with a stick and see what happens... :D

    1. I'm not just poking with a stick here. Most of these "schools" of economics are houses of cards that fall apart with the slightest breeze of empirical data.

      This is really serious -- why doesn't anyone looking at macroeconomics look at empirical data?

      Why do we just get iteration after iteration of political priors and economic terminology?

      No one appears to want to solve anything.

      The whole idea of "schools of thought" is ridiculous. It happens in fields like literary criticism or philosophy because no one ever reaches a conclusion. It happens in economics because no one ever wants reach the conclusion that their idea is wrong.

    2. "This is really serious -- why doesn't anyone looking at macroeconomics look at empirical data?"

      I'd love for you to get more attention from people who do look at empirical data... if such people exist in the econ world. Then they can come over here and tear you apart... where somebody like me would be easily snowed or bamboozled. And if they failed... then great!... but I'd like to see them try.

      "It happens in economics because no one ever wants reach the conclusion that their idea is wrong."

      Yes, I was going to say something similar, as my personal "conspiracy theory" ... but you said it for me. I imagine John Cochrane might accuse me of engaging in a "Bulverism" if I'd brought that up on his blog (John likes that word).

      Perhaps the way I would have phrased it is slightly different: it's not in any economists' best interests to specify what empirical data would demonstrate that their ideas are wrong, because the market place for economic ideas does not really require that of anybody currently, so what's the upside? That's my speculative theory.

      Perhaps with enough pressure a Gladwellian "tipping point" will be reached, and overnight, testing against empirical data will become all the rage. I'd be nice if an economics interested billionaire in the mold of James Randi would put together some kind of public challenge with a potential monetary payoff. Then people could at least say the equivalent of "You're a psychic, eh? Then why haven't you collected your million dollars?" If I were a billionaire that would be enormously fun to do... have this great big challenge out there ($10 million? $25 million?)... like a big embarrassing neon-sign pointing out what's missing in the field.

      But there are empirical data centered ideas in economics, right? I'm thinking of econometrics. But in that case there's no theory, except implicit linear relations, ... correct? (If I've understood your point about this that you've expressed before)

      It's rare to see both a true theory or theoretical framework and a dedication to comparisons with empirical data.

      Re: stick poking: you have the potential with a post like this of gathering some material for your own version of Noah Smith's bestiary.

    3. Your answer to Noah Smith's bestiary looks like a "derpiary."

  2. What about Austrians and "Paleo Keynesians?" (Nick Rowe sometimes speaks of Paleo Keynesians, but I don't know if they actually exist). Also, is Steve Keen in his own category at this point? I know he's related to PKE. Are neo-Fisherites (both of them? Williamson & Cochrane) under one of those categories, or their own thing at this point? I think they'd say they're NKers basically, no?

    1. Perhaps there are some paleo-monetarists as well? And then John Handley... he'll probably object to my limited understanding characterization, but a monetarist/NK fusion or sorts for him? (NK with money)

    2. Paleo Keynesian economics is standard economics minus DSGE

      New Keynesian economics is basically the same as monetarist economics except at the ZLB.

      I'm not entirely sure I understand Austrian economics.

    3. Tom, I think Roger Farmer describes himself as a Neo-Paleo-Keynesian, if that counts.

  3. That's right, Post Keynesianism has a lot about political economy and have political positions.

    But funny you call it standard macroeconomics + politics. It's a complete rejection of standard macroeconomics.

    But that's a minor point. The bigger point is that every one has a political position and you look amateurish claiming you are some neutral sort of a guy or that standard macroeconomics has no political position.

    1. I'll be more amenable to your position when you've put a theoretical curve through some empirical data.

  4. Is empirical success really all that important, though? (I may be turning to the dark side of realism, even if I only appeal to Hume)

    1. My humble request is that empirical data be given a weight w > 0.

      But your post brings up interesting philosophical issues. What do you do when you think there are more variables but there is insufficient data to reject more complex models?

      Also: my info econ 101 vs a naive econ 101 supply and demand analysis of the minimum wage was not intended to say "economics is wrong". It was meant to be:

      naive econ 101 analysis is incorrect (therefore there's more to it)


      info econ 101 analysis says from the outset there is more to it

      Basically, it's not good if a naive analysis can lead you astray at the outset. It's better if it captures ambiguity.

    2. "What do you do when you think there are more variables but there is insufficient data to reject more complex models?"

      Well, let's say that our hypothesis is that

      y = f(x,z,u)

      but we can only directly observe x, y, and z and we know that there are no other variables possible. In this case, assuming we're working with aggregate time series' and can't do a laboratory experiment, the model basically is invincible to the data; we can just assume that the unobservable (u) adjusts to whatever makes our model true, even if the actual relationship is only

      y = f(x,u)

      This is basically Market Monetarism; there is no possible test of the model because expectations are unobservable.

      Does unfalsifiability in this case mean that the model is wrong? What if the actual relationship is indeed y = f(x,z,u)?

      Of course I guess you could Occam's razor if you have a model that works as well but only has y = g(x,z); the second model (e.g., ITM) is falsifiable and simpler than the previous one (MM), so it can be rejected.

      Also, if such a model exists, then it can be inferred that u = h(x,z) since both models are empirically accurate. This would mean that y = f(x,z,h(x,z)) = g(x,z).

      If there is no alternative model (and if the list of possible causes isn't arbitrarily cut down to three), I guess you'd have a problem, especially since closed systems are basically impossible with econ.