Monday, February 06, 2017

Much of econ has become more scientific

My employers, the editors of Bloomberg View, have an editorial out called "Why Not Make Economics a Science?". I like this post, it gets at a very important point. But I think it leaves some very important things out, too.

The article's main criticism of econ - or really, of macro, which most people casually call "economics" - is that it seems reluctant to discard theories:
[T]oo many theorists...have drifted far from the real world...Before the 2008 financial crisis, for example, the standard models more or less ignored finance...Given such spectacular failures, you’d think the profession would have gone back to the drawing board. It hasn’t. True, some tweaks have been attempted...But the error at the core of modern macroeconomics -- that mathematical consistency matters more than empirical relevance -- prevails...Reviving economics as a science will require economists to act more like scientists. If models are refuted by the observable world, toss them out. Rely on experiments, data and replication to test theories and understand how people and companies really behave.
As you can tell from the links (the post itself has many more), the editors have done their homework.

The editors are right that mainstream macro theory hasn't changed much since the crisis - the addition of finance, though important, really is a tweak to the basic structure. Central elements like the consumption Euler equation, TFP shocks, Calvo pricing, infinite forward-looking-ness, exponential discounting, profit-maximizing firms, etc. are still really common in DSGE models, despite the steady drumbeat of evidence against many of these assumptions. And DSGE models, though a tiny bit less popular than at their peak, are still really common in the literature:

The editors are right to be annoyed that the basic DSGE framework has only been tweaked, instead of rethought from the bottom up. Imre Lakatos would probably agree. Science should be about tossing out theories, not generating infinite numbers of theories to sit on the bookshelves gathering mold.

But I also think this brief editorial leaves out a few important things, which I'd like to remind people about:

1. Most economics is not macro.

In the press, we have a tendency to use "economics" as a synonym for "macroeconomics". There are a couple reasons for this. First, it's tedious to keep writing "macroeconomics". Second the only branches of econ that the public traditionally cares a lot about are macro and trade, and maybe a little about finance. Now people are starting to care a bit about labor too, which is good. But relatively few readers care about game theory, decision theory, industrial organization, development economics, public finance, economic history, environmental econ, ag econ, urban econ, etc.

The Bloomberg View editors know this distinction, which is why they specify in the article that they're talking about macro. But many economists in other fields will tend to read this editorial and get annoyed, since it leaves them out. In fact, as many of us Bloomberg View writers have noted, the non-macro parts of econ are now mostly empirical, and empirical econ is looking more and more like a standard science (i.e., very careful attention to controls). So it's important to remember that.

2. The definition of "macro" is part of the problem.

If you look at what academic macroeconomists - that is, the professors in the macro areas of econ departments - are doing, a lot of it now is pretty empirical. For example, macroeconomists might try to determine whether sticky prices or sticky wages matter more in business cycles, or why companies don't hire more young workers during recessions. Each paper of this sort will typically focus on understanding one piece of the macro puzzle. They'll have theory sections, but the theory will be limited to the phenomenon in question - it won't be a big general model of the whole economy.

Unfortunately, we often use the term "macro theory" only to refer to the big, general models of the whole economy. And since DSGE models are still the main type of big, general model of the whole economy (OLG would be a distant second, and most people don't call VARs "theory" at all), this means that "macro" now means "DSGE" almost by definition. Until someone comes up with a type of theory that isn't called "DSGE" and it gains credence as an alternative (for example, "agent-based" models), macro theory will by definition consist only of tweaks.

So while the Bloomberg View editors are right to be annoyed at the fact that mainstream macro theory hasn't changed much in the past decade, we should all recognize the big changes that have taken place not only in econ as a whole, but even within the macro area itself. No, there hasn't yet been a replacement for the basic Ed Prescott-inspired business-cycle modeling framework. But lots of other important work is being done, much of it very scientific.


  1. Ceasar9:31 AM

    Even in traditional scientific fields (e.g. physics), major scientists have gone to the grave with strong convictions in their pet theories that have been refuted in their lifetime. Progress only became noticeable after their deaths.

  2. Anonymous12:36 PM

    Appreciate the link to the Forsythe paper as I have not seen that before.

    Making economics a science would be difficult due to the lack of constant variables. There are assumptions in macro that can lead to a good estimate but the variables don't remain constant because a lot of those variables are constantly changing in a micro sense. I also don't think there is a really good tool to measure wealth yet; GDP, PPI, CPI, all seem to have some sort of hole and the data collection isn't to the point where I believe economists can make scientific proofs. As data mining continues to get better, this can all change. You seem to have a strong background in understand the macro subject matter and is one of the reasons I follow this blog.

    Disagree on the interest in game theory. At a grad level it may not be as interesting but in business and on a micro level it is the most powerful subject matter I have seen. In mathematical applications it may be boring but in engineering applications, it can create powerful mechanisms.

  3. Great post, I would add that it's actually business cycle theory that's the problem and it is a subset of macro. Macroeconomic s traditionally includes theories of growth and there has been a ton of new original stuff there.

  4. There are all sorts of sciences. One of the problems I've seen is that economists are trying to be like chemists or condensed matter physicists when the field is more like cosmology or evolutionary biology. Chemists and condensed matter physicists have the advantage that they can do all sorts of desk top experiments that give directly usable answers. Cosmologists and evolutionary biologists have to deal with nature as they find it and can only do the most limited and indirect of experiments. Economists have to deal with society as they find it and are limited in their experimentation.

    Science often starts out as stamp collecting, and only as the collection grows do patterns emerge. I often get the impression that economists haven't collected enough stamps before claiming to have found patterns. There are just too many things that economists have no explanations for, and too many areas in which their work seems to be extrapolation untempered by observation. We know, for example, that the laws of accounting are pretty solid, but too often economics and accounting produce inconsistent results.

    I think your critique of macro is on target. They've built all those big models without learning from the smaller ones. It wouldn't matter so much, but there are forces that have used economics' scientific reputation to back pernicious political views. Economics needs to actually become more scientific, even if it becomes less influential in the short run.

  5. Anonymous2:29 PM

    Personally, instead of reviving Auguste Comte's social scientific pipe dream once again, I reckon economics should try and become more of a profession, like dentistry even. One obvious barrier is the tolerance of shysters and quacks, be they TV talking heads spouting tosh in line with their paymaster's political line or, as Inside Job" made abundantly clear, (impervious) paragons of academic virtue whoring out their credentials for cash. Mebbe sort that out before reverting to the physics envy chat?

  6. Bill Ellis11:46 PM

    Wouldn't it be ironic if Trump caused a trade war that really ended up hurting Mexico's economy in a big way...causing a new wave of Mexican economic refugees to migrate to the US ?

    Maybe Trump's "wall" will keep them out... But one thing to keep in mind is they are saying it will take a decade to build...
    And a second point...When it comes to wall vs desperate people...i'll put my money on the desperate people every wall people seem to lack imagination...

  7. Nobody should be doing Macro anymore. It cannot be made more "scientific". All theory and data analysis should cease. Patience, young padawan, patience.

  8. Models are never really refuted by empirical observations in macro. it's impossible. Empirical models are never true. They are only either (incrementally) useful or not.
    The problem with current macro is that it's, by and large, completely useless. Finance will not help. Empirics will not help. Stay away from reality.

    1. Anonymous3:20 PM

      Thanks for your opinion. Both people who care are impressed.

    2. Nobody needs to be impressed. Everybody should ask themselves the question if the new strategies will work. Or more crucially: how would they know if they work or not?

  9. Great post. I believe the purpose of positive economics is to make accurate statements about economic phenomena. I think one of the most common mistakes in modern economics is too much time spent building complex formal economic statements and not enough emphasis testing the validity of those statements. Economists spend hours trying to put their statements into formal language according to academic standards. I believe economic statements do not need to be complex in order to be testable.