Paul Romer's Criticisms of Mainstream Macro Are Weak

Paul Romer recently raised a stir with a paper "The Trouble with Macroeconomics." In my view, the implications of this paper are being misunderstood. The interpretation I have seen is that a senior practitioner of mainstream macro is pointing out the deficiencies in DSGE modelling, and that this will cause some soul searching and changes of habits in the economic mainstream. However, this paper seems to be best understood as a venting of professional grievances by someone leaving academia to take a senior policy position, and there is nothing in the paper that is already obvious to any halfway competent mainstream economist.

What Are The Criticisms?

I see two main areas of criticism within the paper.

1. We cannot identify DSGE models (too many parameters versus the amount of available data).
2. Certain senior economists associated with Real Business Cycle models are political hacks.

I discuss these in turn.

Identification - Yeah, What Are You Going To Do About It?

If we want to create an economic model that covers the entire macroeconomy, we need to write down equations (more generally, constraints) that relate the various economic variables that we wish to model. Those equations will contain parameters that determine how the system behaves -- what is the quantitative effect of changing one variable upon the others?

Unfortunately, this is a big problem,

• We need at least four sectors (households, central government, businesses, external sector), and realistically need more than that. (Capital goods versus consumer goods, financial sector, de-consolidating the central bank, high-skill versus low-skill labour, etc.) .
• There are multiple financial instruments that govern the interaction of sectors, and real quantities such as inventories and capital (and post-Keynesians may immediately start yelping about how we measure "capital".)
• The more sectors and instruments that you have, the more equations (constraints) are needed to describe the interactions. (As Romer describes, going up with the square of the number of sectors.) Accounting considerations can eliminate some equations, but typically only one of them for each sub-system.
• Unless we want to have linear functions determining behaviour, we need multiple parameters describing behaviour.

We very rapidly end up with more parameters than we have independent observations.

However, this is true for any conceivable modelling approach. Romer correctly points out that the DSGE researchers are effectively pulling parameters out of their nether regions, and obfuscating this by burying the description under Bayesian sludge. Although true, there is no alternative if we want to develop a whole economy model -- any identification technique is effectively going to involve researchers pulling parameters out of their nether regions. That said, no one is going to say that in print in an economic journal, as that would be "unscientific." (I am willing to write things like that, but I am a victim of a terminal case of being able unable to lie convincingly.)

The "neoclassical" project in business cycle macro might be summarised as:

1. Fiscal policy is a priori evil, and to be avoided.
2. This means that theory must scientifically prove that central banks are able to control the business cycle. (Insert disclaimer about the zero lower bound.)
3. Monetary policy almost entirely consists of a committee setting a single control variable (interest rates). This is central planning at its finest, and so the central planners need a scientific model that allows them to set that interest rate.
4. Therefore, research staffs at central banks must have "scientific" models of the entire economy.

There is nothing in the Romer paper that suggests breaking that inexorable chain of logic. The only implication of his paper is that the current failed identification procedures be replaced by some other obfuscatory procedure that mainstream researchers can present as "scientific."

That said, mainstream researchers are churning out thousands of new papers every year. including papers on identification. In other words, this issue is already being dealt with (from the perspective of mainstream researchers).

Real Business Cycle Theory Is Political Hackery

Paul Romer has some professional beefs with some of the authors associated with Real Business Cycle (RBC) theory. Gosh, an academic is involved in academic politics -- what a shocking development! Seriously, no academic should view this anything other than an academic political spat. (Apparently, one post-Keynesian compared another's editorship of a journal to the rulership of Robert Mugabe, so Romer's beefs are pretty weak sauce.)

I do not really care about what is happening within mainstream economic journals (where RBC apparently still has a foothold), but it is not taken seriously elsewhere -- other than as for teaching models, since they are simpler than New Keynesian models. Central banks do have DSGE models, but they are New Keynesian models. The only people who appear to care about Lucas et al are free marketeers who do not want to join the Austrian economics cult, and this seems to be pretty well understood.

Concluding Remarks

Although this paper has generated a fair amount of headlines, the criticisms are extremely weak from the perspective of mainstream economics, and there is no reason to expect it to cause changes in behaviour. Whatever debates it spawns will be a distraction from the deeper problems associated with DSGE macro.

Post-Script: Real Effects Of Monetary Policy

There is a discussion of the real effects of monetary policy within the paper (part of the dispute with the RBC crowd). However, Romer points to the ability of the Fed to raise interest rates as evidence of this.

This would make no sense to anyone who is unaware that people believed (as Romer appears to do) that the Volcker Fed was setting the level of the money supply, and the interest rate settled at its own level. This interpretation is exactly the opposite of the endogenous money view, which argues that the Fed was truly setting interest rates in an attempt to control the growth of the money supply. (In exactly the same way that central banks now set the interest rate to try to control the level of inflation.)

I think the best description of reality is the endogenous money view; I believe there was even some later research by the Fed (or other well known mainstream authors) which admitted that the "money supply targeting" was a deliberate obfuscation of reality, because the Fed did not want to admit it deliberately blew the economy out of the water with outrageously high interest rates. (I do not have the reference any more.)

I am supposed to be writing about the "endogenous money" debate some time later, so I do not want to go into that here. However, it looks like another reason for post-Keynesians not to celebrate this paper.

(c) Brian Romanchuk 2016