Here’s an abstract from a paper by Erich Pinzón-Fuchs:

This paper discusses a longstanding debate between two empirical approaches to macroeconomics: the econometrics program represented by Lawrence R. Klein, and the statistical economics program represented by Milton Friedman. I argue that the differences between these two approaches do not consist in the use of different statistical methods, economic theories or political ideas. Rather, these differences are deeply rooted in methodological principles and modeling strategies inspired by the works of Léon Walras and Alfred Marshall, which go further than the standard opposition of general vs. partial equilibrium. While Klein’s Walrasian approach necessarily considers the economy as a whole, despite the economist’s inability to observe or understand the system in all its complexity, Friedman’s Marshallian approach takes into account this inability and considers that economic models should be perceived as a way to construct systems of thought based on the observation of specific and smaller parts of the economy.

Over the years, I’ve read lots of articles by people who think macroeconomics is making lots of progress. They are almost all macroeconomists.

I’m not sure I’ve ever met a non-macroeconomist with a high opinion of the field. I am also a skeptic. It’s not that macroeconomists don’t know a lot of useful things. I think they do. But their attempt to create general equilibrium models of the economy doesn’t seem to be getting us anywhere. The abject failure of the profession to respond appropriately to 2008 (where were the calls for easier money?) and the Neo-Fisherian boomlet are recent manifestations of the problems with modern macro.

I’ve always liked Friedman’s work because he seemed to reduce macroeconomics to smaller bite-sized chunks, which can be more easily digested. Let’s use a few simple (partial equilibrium) models that we do understand, and try to see how far they can take us in explaining the economy. He relied on two key models, both based on well-established micro principles:

1. A monetary model of nominal aggregates
2. A labor market model

This is also the approach I took in my book on the Great Depression (The Midas Paradox). Only instead of looking at the monetary aggregates, I looked at the global market for gold. But the basic idea was the same. Gold was the medium of account during most of this period; hence determining the value of gold was tantamount to determining the price level. In other words, I had an essentially microeconomic model of the price level. I modeled the value of gold using simple supply and demand, and then recognized that the price level was the inverse of the value of gold.

In contrast, modern models try to directly model real GDP, and then assume that if RGDP “overheats” it will put upward pressure on prices. But that theory of inflation is not well grounded in microeconomics. In micro, rapid growth in the quantity of a product does not cause its price to rise; it entirely depends on whether the rise in quantity is due to supply or demand-side factors. Better to start with a monetary model of NGDP.

Friedman and I ground our theory of employment in basic microeconomic concepts. High unemployment can occur for one of two reasons:

1. Government policies that artificially raise the cost of labor (minimum wage laws, etc.) or discourage people from working (high implicit marginal tax rates.)
2. Unexpected declines in the price level/NGDP, due to a tight money policy, combined with sticky wages.

These are two simple tools, the S&D for money and the S&D for labor. They are pretty well grounded in basic economic theory. There is a mountain of economic evidence in favor of each view. If there is a big rise in unemployment, then one of the two factors above is almost certainly to blame.

In my work on the Great Depression, I found both of these tools to be extremely useful. Not only is the Great Depression itself not at all mysterious (how could there not have been a Depression, given the massive government hoarding of gold, combined with high wage policies?) but the various zigzags from 1929-39 are also pretty easy to explain with these basic tools.

In contrast, New Keynesian models are not very good at explaining events like the Great Depression, partly because they are not very good at identifying monetary shocks. Those parts of the model that are based on basic economic theory are buried so deep they get overwhelmed by all the questionable DSGE modeling assumptions. So the economy is hit by mysterious unexplained “shocks”, which Friedman and I view as bad government policies.

That’s not to say my approach has no drawbacks. Although my supply and demand for gold model can be applied to fiat money (the monetary base), it gets much more difficult at the zero bound, where other assets are close substitutes. And while we can be reasonably sure that a very higher minimum wage, or very generous social insurance system, will reduce hours worked, we don’t have a very good grasp of the impact of smaller increases in the minimum wage, and/or smaller increases in implicit market tax rates. There is more work to be done.

Nonetheless, I often feel that macro would make more progress by returning to a few basic principles understood the Hawtrey, Fisher, and Cassel, and then adding on the innovations of Friedman (Natural Rate Hypothesis), Lucas (rational expectations) and Fama (efficient markets). And by dropping those strategies that have not proved very effective (real business cycle models, interest rate-oriented monetary models, IS-LM, etc.)

K.I.S.S. Work with models that we know to be true, because they are based on well-established microeconomic principles. That’s how we can get macro back on track. In addition, creating new markets, such as a highly subsidized NGDP futures market, would help to clarify the nature of what we really mean by a “demand shortfall” as well as what factors cause it, and what policies can cure it. We need one model of nominal shocks, and another entirely different model explaining how nominal shocks impact real variables. Right now our models try to do both at once, and fail miserably.

Screen Shot 2016-10-23 at 10.47.15 AM.png