Recent Posts

Sunday, June 11, 2017

Quick Comments On Ongoing Projects

This article gives a few comments on some projects that I am involved in. They are the upcoming Modern Monetary Theory (MMT) academic conference (in September), my stock-flow consistent (SFC) models guide, and my struggle to put mainstream Dynamic Stochastic General Equilibrium models into a formal mathematical framework that makes sense to me.

The MMT Conference

The call for papers for the 2017 MMT Conference at the University of Missouri at Kansas City (September 21-24th) is ending soon. Previously, the conference was billed as a "Post-Keynesian Conference" (I attended last year). The decision was made to give the conference a more decided MMT focus. (As I have previously written, MMT is a school of thought within a "broad tent" post-Keynesian tradition. Some authors define post-Keynesian economics in a "narrow tent" fashion. which leaves no good label for the "broad tent post-Keynesians." From a real-world perspective, this labelling dispute smells like silly academic politics, but as someone who attempts to write about the school of thought, it is a real pain.)

I have made a submission; I will discuss it in the next section. Unless some form of emergency crops up, I expect to be at the conference.

I have a somewhat mixed view of the narrowing of focus. From the perspective of MMT, there is definitely a need for a discussion of "what to do next?" It has to be understood that MMT is not a purely an academic exercise; the objective is to put into place a coherent policy platform that can seriously rival mainstream economics. Being involved in fratricidal academic political games about the exact provenance of economic theories is not going to advance that project.

The down side from the perspective of someone outside of the academic circuit, it makes it harder to get a broader view of post-Keynesian theory. Since MMT leans heavily on the contributions of authors who are not considered "MMT," they cannot be ignored. However, it is difficult for a non-academic to justify going to multiple conferences during the year.

SFC Models Project

My submission to the conference is an overview of the Python sfc_models package. At present, the package only has limited coverage of the models from within Godley and Lavoie's Monetary Economics. However, with the core algorithms in place, adding new economic functionality is straightforward.

In my view, the package could be very useful as a communication tool. It is powerful enough to handle most (discrete time) stock-flow consistent models that will be devised by researchers (although those researcher would probably need to learn a small amount of Python). The framework generates the equations and solves them, reducing the difficulties of tracking dozens of similar parameters in a multi-country model. The solution method is completely transparent, so that the researcher can examine the equations generated.

For users who are are not programmers or comfortable with mathematics, the framework will solve the system, which is generated with a few lines of code. The equations can be consulted if so desired, or the time series output examined either as time series plots, or in an automatically-generated CSV. It would be possible to make simple experiments by modifying an existing model with very limited programming knowledge.

In the longer-term, the system could easily have a graphical user interface added.

In my view, the framework would allow MMT researchers to share teaching models as well as research models with a wider audience. (I certainly intend to use the package as part of my writing programme.) It would be possible to have a set of "canonical models" which might be used in internet discussion. (It would also help insulate MMT against the notion that it has no mathematics.)

Currently, I am trying to finish off the "user guide" for sfc_models. There is not a whole lot of text to write, and I hope to finish it off before September.

Robert Vienneau Link

Robert Vienneau writes the blog "Thoughts on Economics." I believe that he is also trained in mathematics, and he writes extensively on modelling issues in economics. For some reason, I thought he stopped blogging, but that was not the case.

He recently linked to my article on "The Horrifying Mathematics of Infinitesimal Agents." He also linked to the "Econ job rumors" web site forum area which eviscerated my logic (allegedly). My personal favourite was the line: "So the blogger is not familiar with how decimal system works then." in response to my referring to a theorem in Kolmogorov and Fomin. This was a perfect mixture of condescension and raw ignorance. (The proof in Kolmogorov and Fomin precisely relies on showing we cannot represent all elements in [0,1] as decimals.)

The other comments were of the "read the vast literature" variety. (What was that about "mud moats" again? From the perspective of mathematics, the entire DSGE literature appears to be a mud moat.)

To be honest, my article was somewhat of a fishing expedition. All of the papers I read that covered Calvo pricing were completely and utterly wrong from the perspective of formal mathematics. I had to guess what was the alleged true mathematics behind the concept, and I generated some leads.

In case it's not clear, I am not interested in whether DSGE models are a good approximation of reality. (Certainly some readers on econjobrumors did not pick that up; they read what they thought I would write, and not my actual text. The reality is that most internet readers skim text heavily, so that is to be expected.) Instead, I am concerned whether DSGE models are mathematical models in a formal sense. There is no such thing as "an approximately true" proof.

All that would be needed to shoot down my aspersions is to formally write out the full mathematical model of the Calvo model with both production and household utility optimisation. I would be happy to receive such a model definition, but I keep running into the "someone else did the mathematics in some intermediate micro textbook," without naming names. (Admittedly, there was a somewhat interesting reference given in the comments; I have not tracked it down. However, the description of that paper in the comments suggest that mainstream economists have been doing the mathematics of a continuum of agents wrong, which is exactly my point.) If I were reassured that the New Keynesian model is in fact a mathematical model, I could then turn to examine what it predicts about the real world.

Anyway, returning to Robert Vienneau, he asked (in the above linked article):
Don't the last two bullets [one bullet is a link to my article - BR] imply that the intermediate neoclassical microeconomic textbook treatment of perfect competition is balderdash, as Steve Keen says?
I cannot comment on the other article (by Miles Kimball), but my argument is that the competitive equilibrium treatment within a macro model that features two classes of agents (households and firms) is generally not specified properly within published papers. I have no idea whether some papers exist that in fact define this equilibrium concept properly, and I still have some references to read - i.e. I have no ability to make a definitive judgement. Unfortunately for my readers, I have some consulting work (which is fortunate for me) that is cutting down on my time available for decoding mainstream economics. I will get back to it, but I prefer to wait until I digest some of the references that are piled up on my "to read" list.

Whether or not mathematical proofs on models that do not specify the formal mathematical problem to be solved is "balderdash" depends upon the reader's personal definition of "balderdash"; I am not going to go there.

(c) Brian Romanchuk 2017

1 comment:

  1. Douglas Gale has some papers on equilibrium models and game theory models described under the link below.

    Strategic Foundations of General Equilibrium: Dynamic Matching and Bargaining Games


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.