Thursday, January 28, 2016

Jason Smith — Models and frameworks

If you are into scientific modeling and economic modeling in particular, physicist Jason Smith is up with a foundational post. Even if one doesn’t agree with his approach, he illumines the issue, which many conventional economists either ignore or are ignorant of.

Information Transfer Economics
Models and frameworks
Jason Smith


Random said...

Tom, Jason says he doesn't really understand MMT. Could you leave a short comment explaining it. Cheers. I left a comment on endogenous money on his "how money transfers information" post.

"Hello Random,

You ask:

Do you assume the money multiplier or the MMT view (loans create deposits)

This is not an entirely germane question in the information theory view as it assumes output is in information equilibrium with some monetary aggregate (which one is an empirical question), but causality goes both ways. Additional money opens up economic state space (expands the opportunity set) for occupation by agents, but additional occupied states in the economic state space can end up causing money to be printed.

Empirically it appears the growth rate of physical currency (literally paper bills) sets the scale for nominal output as well as inflation in the short run.

Although due more recent results, I'm now under the impression that money has nothing to do with inflation or output and it's all about the size of the labor force:

Money in that view may be an intermediary as described above, but to first order the effects are invisible at the macro scale.

As an aside: the latter two links in your comment don't seem to be empirical evidence -- there is no explicit model calculation compared to data or quantification of model error.

I do admit that I don't really understand MMT."

Random said...

So perhaps I have helped cause this post :)

Tom Hickey said...

Jason is working on a different basis (information transfer) than MMT is (accounting). What both approaches share is recognition that there is something fundamentally misguided about conventional approaches and conventional economists don't seem interested in addressing it.

I think that Jason's work is important for two chief reasons.

First, he understands mathematical modeling in science at a deep level from the POV of theoretical physics. He is not an academician, so he is also concerned with models that work rather than doing "mathematical philosophy."

Secondly, Hayek called attention to the key role of information in economics and assumed that market prices were packages of information. That is a basic framework as an approach for developing models. Jason picked up on that as economic hobbyist with considerable experience in dealing with the framework of information transfer.

IN my view it would be a category error to assume that the MMT approach is the only sound approach to macroeconomics. Jsson is exploring another approach that some here may find of interest, especially the scientists and engineers who are used to dealing with information transfer.

Random said...

Would he be interested in MMT though?

Tom Hickey said...

I would assume it is not his kind of model, being accounting based.

Neil Wilson said...

The problem I have with physicists is that they struggle with emergent behaviour, and the models they come up with reflect that.

Plus empirically fitting a model to a system that has been running under neo-classical regime for 40 years will give you a model that looks a lot like neo-classical economics.

I see Jason's approach as just another attempt to use System Dynamics to see things. It may come up with interesting epidemiological relations, but it won't get to the underlying mechanisms because it is too high level.

And of course it fails the essence of the Lucas Critique. The point of economics is to develop policy, and policy works on individual entities, not aggregated groups. The response of the individual entities and the change that makes to the aggregation is not mechanical.

Tom Hickey said...

I think that the basic assumption he is working with is that there is some consistency to human behavior and that it can be modeled using a fairly simple model. That's what "rationality" means in econ.

Of course it doesn't explain everything. Neither does MMT or any other economic theory. There is no grand theory as comprehensive general theory.

Physics can put a man on the moon and send a robot to Mars. Biology has done wonders in preventing and curing disease.

There just isn't anything like that in the psychology or social sciences including economics yet. We are still exploring around the edges because scientific understanding of "human nature" is inchoate.

I would say that the underlying presupposition of contemporary science is that the universe is an algorithm, which if were known would be a theory of everything with explanatory and predictive power bordering on omniscience. That is to say, the universe is a big package of information, which what unified field theory is driving at.

One aspect of that package is emergence and one aspect of that is a type of information package that is capable in principle of knowing the universal algorithm as a whole. One theory to rule them all.

The so-called laws of nature discovered so far as aspects of that algorithm formulated as equations that are testable at least in principle.

Moreover, it's pretty fundamental that probability rules, pace Einstein.

As I understand Jason, he is not denying that the socio-economic world is complex adaptive and emergent. He is saying that simple models are not only still possible but useful. Prefer simple tools if they are useful, and then add layers of complexity as needed to flesh out the model.

Right now he has developed a framework based on information transfer and playing with models within that framework to see what emerges. I find it quite interesting and I post some things here that I think others might, too.