Expectations of a Macroeconomist
Monday evening at AEI; My Perspective on Robert E. Lucas, Jr. and macroeconomics
[Note: For various reasons, I will not be holding live Zoom sessions the next few Mondays. I do plan one for Wednesday, May 31. Monday the 21st, I will be a discussant at AEI at an event starting at 5:30 PM New York time. The topic is a book on monetary theory and practice, written by Paul Sheard. I do not share the book’s certainty about macroeconomics and money and banking. If you tune it, there is perhaps a 5 percent chance you will see me lose my temper.]
A number of economists have offered their reflections on Robert E. Lucas, Jr., the Nobel Laureate who recently passed away. I can refer you to Tyler Cowen, David Henderson, John Cochrane (and here), Timothy Taylor, Scott Sumner, or Noah Smith. I appreciated the illustration Smith put on his post, which I assume is an allusion to the Lucas “islands model.”
Consider a macroeconomic theory that rests on the following assumptions.
There is a “representative agent.” There is one consumer, one worker, and one firm in the economy.
Wages and prices adjust instantly.
When the worker is surprised by low prices, he sets his wage rate too high. The firm cannot afford that high wage. For a moment, let us think of workers as plural, and some of them are laid off and become unemployed. Conversely, when the worker is surprised by high prices, he keeps his wage rate too low. The firm hires more workers.
The worker’s expectations for prices are backward-looking. If the Fed undertakes an expansion, high prices will surprise the worker.
I see Lucas as having done two things. First, he exposed (1) - (4) as the “microfoundation” implicit in textbook Keynesian macroeconomics. Second, he objected to (4) on the grounds that, contrary to everywhere else in economics, it takes people to be systematically irrational. Someone who used purely backward-looking expectations will be fooled more often than he should be. Instead, expectations ought to be forward-looking, and “rational.”
When I explained rational expectations to a freshman class at MIT, one of the students said, “That’s like saying the batter knows what the pitcher is going to throw before he throws it.”
That gets the idea. The batter does not know exactly, but he should be able to better than thinking, “Last pitch was a curveball, so I assume the next pitch will be a curve ball.” Instead, he uses all the available information to guess, and he will guess fastball if that’s the best prediction to make.
Rational expectations had a surprising implication, which Lucas emphasized. Think of money growth as having an expected component (which the Fed controls but the market anticipates) and a random component, which by definition the Fed does not control. Any historical correlation between faster money growth and lower unemployment must be due to the random component! It will appear in hindsight that the Fed was able to reduce unemployment deliberately, but this is not the case.
Because policy is ineffective in the model with rational expectations, Lucas convinced macroeconomists that they had to pay attention to microfoundations. If they ignored the implicit assumptions, how could they know which of these assumptions might come back to bite them? How could they know that they were not deceiving themselves about the effectiveness of monetary policy?
Most macroeconomists went along with the idea that macroeconomics needed microfoundations in order to be reliable. It became expected (pun intended) of macroeconomists to work with models that assumed rational expectations.
But the “saltwater” economists made a point of rejecting (2). These “New Keynesians” said, in effect, “rational, shmational, wages and prices are ‘sticky.’” No matter how well the worker anticipates the Fed, he cannot adjust his wage quickly enough to neutralize a Fed expansion.
My own view instead goes back to (1). I think it is wrong-footed to describe the economy as if it were a single GDP factory. Instead, I emphasize patterns of sustainable specialization and trade.
One argument against my approach is that it is a description, not a model. To have an economic model, you need a simple representation of the economy, and PSST does not do that.
I think of post-Lucas macroeconomic theory as mathematical masturbation, because the representative agent assumption is unrealistic about how jobs are created and destroyed. But PSST is not tractable enough to satisfy the requirement to have a model. So there you go.
This essay is part of a series on human interdependence.
Substacks referenced above:
My wife's architecture firm has to bid on projects that take several years. If costs increase faster than expected they eat a loss.
Coming out of COVID they were getting a lot of work and couldn't hire enough people. Later, it turned out a lot of that work ended up not being as profitable as they thought, because the input cost on construction went up.
While this did cause them to increase activity temporarily it didn't lead to a more effective allocation of resources (some of those projects shouldn't have been done, some of those employees should have worked somewhere else). In addition it's now becoming more common to build big COLAs into the contracts for protection, and a lot of work that should be going into design and planning goes into COLA prediction and negotiation. Many projects, especially with government or other grants, can't change their COLA rates and so they've wasted a lot of time going after these contracts only to learn that they are underfunded.
*I'm using COLA because I can't think of a better acronym for cost inflation adjustment.
Ultimately seesawing inflation does not lead to more sustainable patterns of trade in their industry and diverts a lot of energy into inflation prediction rather than planning and construction. So the Fed can "succeed" by surprising them but that surprise makes them worse at their jobs.
Wonderful post, one of the better appreciations of Robert Lucas. Thank you.