Recent Posts

Saturday, October 1, 2016

Whither Mainstream Economics?


The inertia in academic institutions provides a very good reason to expect very little progress within mainstream business cycle macroeconomics. Nevertheless, that segment of the profession could be dragged kicking and screaming back to reality by central banks, and so there is no reason for total pessimism on that score. This article is written from the perspective of an outsider analysing the field from the perspective of the history of ideas, and not arguing about the validity of any school of thought.

This article bookends my earlier article "Whither Post-Keynesian Economics?" It is no secret that I believe that post-Keynesian economics is the best way to approach economics. I discuss how post-Keynesian economics fits into my discussion here in an appendix.

How Bad is the State of DSGE Macro?


My snarky comments reflect my disagreement with his view. That said, it does not really matter for the purposes of this article: at the minimum, the DSGE macro community has a very large public relations problem. Even if the theory is not reformed, the DSGE community would have radically reform how it is presented if it is going to be accepted elsewhere. For the purposes of this article, I treat such a reform as being indistinguishable to an outsider as an true change to the underlying theory.

"I Blame Society"

Like many other outsiders that I have talked with, I am not impressed with the state of academic economics. Although I do chide academics, if one wishes to find the true culprits, one needs to look closely at the pointy-haired brain trust administering our universities.

Putting insanely high required quantitative publication targets ensures that academics have no time to think, nor to have an incentive to do. You have to cultivate a tribal identity -- so that you dump your gibberish in journals (like everyone else). Even the concept of a sabbatical has been gutted. (Please note that these problems affect all schools of economic thought, and is not confined to the mainstream.)

Drawing analogies to the history of pre-war physics is fundamentally mistaken. During those debates, the academic system had not yet deteriorated to its current level of dysfunctionality. The broad mass of academics involved actually had time to think. To be clear, some academics are able to put out high quality work despite this system; having tenure helps. However, reforms need to be put into place by the more junior academics who are producing the mass of papers, as well as acting as peer reviewers. Saying that it would be nice if mainstream journals would accept non-microfounded papers is irrelevant if said papers are still arbitrarily bounced by anonymous reviewers. No one trying to apply for positions can afford to have their papers held up for years in peer review, even if they are ultimately vindicated upon appeal.

Pure Pursuit of Knowledge Versus History of Ideas

There are two ways of pursuing this discussion.
  1. A belief that academia is pursuing the pure pursuit of knowledge, and all that matters are the ideas themselves.
  2. A history of ideas viewpoint, where the involved academics are viewed as members of tribes, and no determination is made of the ultimate truth or falsity of their debates.
From the pure pursuit of knowledge perspective, a recent article by Marc Lavoie -- "Rethinking Macroeconomic Theory Before the Next Crisis" -- is probably the best recent contribution to the "macro wars." He discusses a number of key points; he could not hope to cover all of the issues. Nevertheless, it demonstrates the patterns of weakness within mainstream macro. 

By convention, academic economists need to follow the first path of discussion; the second approach leads to very awkward discussions. For example, one may note that much of the valued sources of credentials within the field have been captured by economists following Dynamic Stochastic General Equilibrium methodology (which is based upon "microfoundations"). Was the insistence upon "microfounded" research a mechanism to lock outsiders out of those sources of credentials? I do not want to dwell on a topic that is going to be offensive to any mainstream readers of this article, but I will note that I believe that economic incentives do influence observed behaviour, and hence that I see the odds are tilted against reforms that remove the special status of DSGE models within academic institutions.

Even if we ignore the issue of incentives, an outsider should still keep an eye on the field from a "history of ideas" perspective. Academia churns out a huge volume of papers, which necessarily only represent incremental progress. We cannot hope to read them all. The hidden advantage of the current situation is that unless you are forced to use DSGE models as a result of a pursuit of credentials, it is entirely safe to ignore the literature. If the mainstream actually manages to reform itself, there would be a much larger wave of publications that one would need to keep up with.

The "Empirical Work Is  Great" Dodge

As always, I am mainly interested in business cycle macro. That is, I care little about the state of progress in areas like micro, behavioural economics, or growth economics. Correspondingly, I have no opinion about the state of the art there.

It should also be noted that the majority of "macro" research published now is empirical (which is an implicit comment on the quality of macro theory). According to the people publishing that empirical work, the quality is amazing. Given my description of the evolution of academia, we need to dial down our expectations of what "high quality" academic work represents. Relative to contemporary academic standards, the new empirical work might be quite good.

Nevertheless, I have not been too greatly impressed with the some of the empirical work I have dealt with in detail. A common issue was that the methodology relies upon an assumption that is described in two sentences or so, but the assumption was otherwise completely ignored within the text of the paper. When it is time to replicate the work, the effects of that assumption dominates the results. If you accept the claims in the abstract and conclusion at face value, you end up with a very different view of the paper than someone who has to replicate it. Meanwhile, most people citing the original empirical paper do take the claims at face value, which implicitly weakens the later paper -- as the original assumption completely disappears from sight.

Empirical work cannot advance without some sort of theory to guide it. You could buy a chemistry set, and very methodically mix the various chemicals together. You could tabulate hundreds of experiments whose results could be replicated. But you are never going to infer the existence of the Periodic Table just by randomly mixing chemicals.

Modes of Internal Reform

The main possibilities for internal reform of DSGE macro are presented below. I have tried to cover what I view as the main possible scenarios, although this list is non-exclusive. (And as I noted above, reforms to DSGE macro might only take the form of how the theory is presented.)
  • A brand new powerful theory appears somewhere, and it pushes DSGE macro to the back burner, since it is demonstrably superior. Although I am cynical about academic politics, I believe that there is enough dissatisfaction with the status quo that an obviously superior theory would not be blocked. (This is scenario that everyone wishes will happen -- possibly excluding entrenched senior academics. As I discuss in the Appendix, I am skeptical that such a theory exists.)
  • Some innovation to the DSGE theory (or its presentation) increases the power of theory, and it is able to convince skeptics like myself that it is a progressive research strategy. (This is obviously what the mainstream wishes will happen.)
  • The DSGE community announces that it has discovered an innovation that has solved macro theory problems once and for all, and skeptics in the post-Keynesian tradition will point out that it still inherits the fundamentally flawed assumptions that have been imported from neoclassical economics. (This status quo scenario is exactly what has been happening within mainstream macro for pretty much the whole post-war period, and has to the bookies' favourite.)
  • Mainstream journals open the doors to non-microfounded research. At least some post-Keynesian research (stock-flow consistent models, say) would appear side-by-side with DSGE models in journals and academic conferences. (Professor Simon Wren-Lewis supports such an outcome. Although it appears like a reasonable possibility, I will believe it when I see it.)
In summary, the suggested course of action by Professor Wren-Lewis (accept non-microfounded papers) is the only one plausible that would allow some progress in the absence of the appearance of a new economic theory out of the æther. Since it makes no sense to arbitrarily exclude non-microfounded models, the main barrier is obviously institutional incentives. Those non-microfounded models would have no need to cite the DSGE papers published by DSGE modellers, and hence their "scholarly impact" would be at risk. Given the weakness of the DSGE modelling techniques, the slippage in "impact" could turn into a rout.

Central Banks as Islands of Functionality

My previous discussion shows why I do not expect internal reforms to change the tribal positions within academic economics. It would be easy to project that the next two decades in economic research will repeat the pattern of the post-war era. The wild card is external pressure for reform.

Central banks are islands of relative functionality adrift in a sea of academic dysfunctionality. Luckily for central bank research staffs, their top people know at least something about the research process, and are sensitive to quality. Furthermore, unlike academics who face no repercussions when they write stupid stuff about the economy, central bankers regularly meet investors who lambaste them for what the investors view as policy errors. (Admittedly, the central bankers are probably more often correct than the investors.)

The New Consensus cult elevation of monetary policy, without providing useful models for guiding that policy, has put central bankers in an extremely awkward position. Anything reality-based is going to be an improvement. Central banks are small-c conservative institutions, and hence deferential towards credentials, but they are not going to sacrifice their careers just to gratify the ego of some DSGE guru.

Whether or not the central banks run large DSGE models is a side show, and its significance is overstated. Many of my ex-colleagues were central bankers, and there was a universal cynicism towards the old (pre-DSGE) large scale models that they had to update as part of the monetary policy review process. Based on my own experience, the easiest way to get someone to hate a model is to force them to maintain one that was developed by someone else. The original developers get all the glory, and you are stuck trying to keep the lipstick on a greased pig. Therefore, the fact that the models are used does not apply that all the researchers at the central bank are particularly attached to the model. Finally, as anyone who follows the Fed's "dot plots" knows, monetary policy decisions are not based on slavishly following the output of a single large scale model.

Politics?

Another obvious wild card is politics. If the winds of politics change, academia would likely follow suit. I currently do not see such an event on the horizon, but one may note the rise in populist politics in recent years.

Concluding Remarks

The difficulty in creating a solid quantitative economic model means that one should not expect decisive internal reforms within economic academia. However, central banks are going to increasingly exert pressure on the ivory tower, so we might not see a replay of the post-war experience going forward.

Appendix: Post-Keynesian Economics

My interpretation of post-Keynesian economics leads to a somewhat pessimistic conclusion: we cannot hope to assign a single probability distribution to the future value of macroeconomic variables using a single model. (The existence of such a model would allow for a "scientific" determination of the validity of the model using statistical tests.) Please note that this is how I am interpreting the theory; individual post-Keynesians have other ways of expressing this idea (typically in a more verbose fashion).

Instead, the best we can hope for is to come up with conditional forecasts. For example:
  • If the United States avoids recession in 2017, the expected path for inflation is X.
  • If the United States falls into recession in 2017, the expected path for inflation is Y. (This estimate for inflation may or may not be generated by the same model used in the above case.)
Unfortunately, we have a hard time putting odds on the probability of recession, and so we cannot infer a single probability distribution for the inflation path. Correspondingly, we cannot apply statistical tests against the model(s) used to estimate the odds that they are indeed valid models.

If we assume that my assertion is correct, the implication is that we cannot blindly hope to apply the lessons of science (particularly Physics) to economics. Any theory that offers quantitative predictions can be shot down for one reason or another -- either problematic observed data, or theoretical incoherence. This means that there can be no "final answer" for macro theory, which is why we are stuck with battling tribes with different interpretations.

It should be noted that this theoretical ambivalence is not novel; it is effectively how market economists operate. 

The side effect is that this puts academic post-Keynesians in a bind. Since they cannot produce a model that is "scientifically true," there is no reason that their theories must be adopted on "scientific" grounds. All that can be done is package their theories in a way that makes them more attractive than their competitors. As time passes, they will be absorbed into the "mainstream." Although this process seems inevitable, the question remains whether mainstream authors will properly cite the post-Keynesian authors, or else pretend that they came from within the New Consensus tradition.

Finally, my discussion above suggests why post-Keynesians should consider packaging their theories in a way that is attractive to central bankers. One example is the reality that central bankers are bureaucrats, and do not want to inject themselves into partisan politics. As a result, they will never be able to get involved in political economy discussions, rather they have to reflect the existing window of accepted discourse. (That window does move, but central bankers do not see themselves as the ones moving it.)

(c) Brian Romanchuk 2016

6 comments:

  1. "Empirical work cannot advance without some sort of theory to guide it."

    Empirical work cannot advance because the data is already constrained by the 'mark to model' economy that generates it. The economy is essentially a man in a strait jacket and therefore the data coming from the economy are the points of reference related to a man in a strait jacket.

    The arguments then come forward that we cannot possibly remove the strait jacket because the data shows that the man has very limited limb movement.

    ReplyDelete
    Replies
    1. LOL. What a great analogy! The data shows the man has very limited limb movement so why remove the straight jacket? Excellent.

      Delete
    2. Yes. One of the other issues is that the theory guides the definition of data. Ouch.

      Delete
  2. Someone out there may be publishing a book this December that tries to give firm methodological grounds to your final considerations. The idea being that economics cannot use the methodology of, say, physical science. And that all it can hope to do is try to understand various relationships or 'schemas' instead. Some preliminary discussion here. Book is out December 16th.

    https://fixingtheeconomists.wordpress.com/2014/08/08/inet-ysi-discussion-of-a-chapter-of-my-forthcoming-book/

    ReplyDelete
    Replies
    1. I will have to keep an eye out. What formats will it be published in - hardcover, ebook?

      Delete
    2. I think hardcover and ebook yes. £20 or so I think.

      Delete

Note: Posts are manually moderated, with a varying delay. Some disappear.

The comment section here is largely dead. My Substack or Twitter are better places to have a conversation.

Given that this is largely a backup way to reach me, I am going to reject posts that annoy me. Please post lengthy essays elsewhere.