Initial Comments On Roger Farmer's Macro Models

Professor Roger E. Farmer has been developing a macro theory that is built around the notion of animal spirits. (This article on the Phillips curve is an example that recently popped up.) I had some misgivings about his approach, and so I never got around to pursuing his research. However, I am now starting to think about writing a book on business cycles, and I realised that his analytical work is pretty much in line with what I have been thinking about. I cannot claim to completely understand his policy logic, but I believe that my views diverge from his with regards to the application of his model towards policy. Another issue that will be of interest to some of my readers is how to contrast and compare his models with existing post-Keynesian models. I will worry about that topic when I get around to writing the book itself.

This article represents my initial opinions on the topic, based on my earlier readings of Professor Farmer's work (and reading the front chapters to his book). I am being deliberately vague in order to avoid misrepresenting his thinking. I am supposed to be finishing off my book on breakeven inflation (as well as waiting for big holes to be knocked into the side of my house), so I am keeping this brief. I will address his models and thinking in a review of his book "Prosperity for All", as well as technical articles that will be folded into my next book.

Roger Farmer's hit my personal blind spot in economic theory: it was close enough to DSGE macro to annoy me, but it was different enough from DSGE macro to be useful in my discussions of that topic. One of my complaints about DSGE macro is the completely non-standard mathematical notation used, as I discussed at length in earlier articles (link to best summary). It has been awhile since I last looked at his papers, so I am not sure whether his work has the same issues.

His models looked interesting, but I never pursued them because I was still thinking in a conventional manner. My diagnosis at the time was that applying such models to business cycle forecasting would face problems. However, my thinking has evolved -- I realise that I want to work with models that are hard to apply to the real world. Furthermore, his models are built around what I view as the key concept -- animal spirits.

If one wants to be sarcastic, one could liken "conventional" approaches to macro as believing in ghosts. The "ghosts" are various not-directly-measured variables that drive the models. The key examples being:

• the natural rate of interest;
• potential GDP;
• NAIRU (or equivalent).

(As a disclaimer, I appealed to those concepts back when I was churning out conventional market research.)

Serious economists have gotten used to dealing with this ghostly quantities, and many in fact will not treat people who reject them as being serious. Unfortunately, all of these variables are defined in such a way that they generate easily testable predictions -- and we can empirically reject those predictions. This has left the true believers using methodologies which result in these variables tearing rapidly around, allowing them to fit any observed data set (making their use non-falsifiable).

The idea is to add "animal spirits" to the list of these not-directly-measured variable. The beauty of animal spirits is that it is by definition vague, and so has a lot more flexibility to avoid rejection. The question is: can we build mathematical models around it?

This is of utmost importance for business cycle models. It takes a very large amount of education to believe that standard DSGE macro models offer any insight to business cycle analysis. ("Everyone decided to take a vacation in 2008!") The models are built around household sector optimisation over time -- the business sector only blindly reacts to prices (both spot and forward), and does not plan ahead. This obviously bears no resemblance to real world behaviour.

I have no idea who coined the term "animal spirits" in conjunction with the business cycle, but Keynes certainly popularised it. The concept is straightforward -- businesses ramp up hiring and investment if they think the future will feature higher demand. This creates a self-reinforcing impetus to growth. Unfortunately, that self-reinforcing nature works backwards in a recession. This mechanic is bizarrely missing from standard DSGE macro.

Where I believe that Professor Farmer and I depart theoretically is the measurement of animal spirits. He argues that stock market prices ought to embody the level of animal spirits, and so we can use the stock market as a proxy for this variable. If the stock market operated in even a slightly rational fashion, the logic behind that case is strong. Nevertheless, I have doubts about the link, largely based on my observations of stock market behaviour. Since it has been awhile since I have read his analysis of the topic, I will defer the discussion of policy to later articles.

In particular, he is part of the growing chorus of voices in favour of sovereign wealth funds. In his case, he believes that the sovereign wealth fund can act to stabilise risk asset prices, and therefore animal spirits (according to his framework). My feeling is that sovereign wealth funds have evolved to be the main free lunch available to policy makers in the "neo-liberal consensus" (which means I should get around to writing about them).

(To tie into my recent articles, I instead argue that credit markets are more important practically. To a certain extent, the Canadian government blundered into a policy that explicitly targets animal spirits: its guarantee of mortgage insurance via CMHC guarantees. The housing market has become an economic laboratory for the Canadian federal government.)

One final issue that I face is the question of how to relate his work to the existing post-Keynesian theory. In his book, Professor Farmer refreshingly acknowledged the existence of post-Keynesian economics, which is unfortunately unusual among "mainstream" economists. He sets out his reasons for disagreement with post-Keynesian authors. I would not be able to comment on those disagreements until I delve deeper into his work.

(c) Brian Romanchuk 2018