Rethinking the same damn thought

Noah Smith at Bloomberg, same post we looked at yesterday. Noah writes

... economists have known for decades that recessions might not be random, short-lived events, but the idea always remained on the fringes. One big reason was simple mathematical convenience -- models where recessions are like rainstorms, arriving and departing on their own, are mathematically a lot easier to work with.

"Recessions are like rainstorms". Nice alliteration. The phrase reminds me of something I read in The Roaring 80s:

... in the Great Depression, economists wrote about unemployment as if it were a bad hailstorm; then the Keynesian revolution gave some hope that nations could do something about the 'economic blizzards' that had previously been considered as random as the weather.

According to Noah, we're back to random as the weather.

... easier to work with. A second was data availability -- unlike in geology, where we can draw on Earth’s whole history, reliable macroeconomic data goes back less than a century. If economic fluctuations really do have long-lasting effects, it will be very hard to identify those patterns from just a few decades’ worth of history.

Skipping ahead a couple paragraphs:

If the only tools available were the ones that prevailed in 2007, it might be best for economists to simply throw up their hands and declare that the problem of spotting, preventing and fighting recessions is best left to our distant descendants.

But fortunately, this isn't the case. New tools are available that could help shed some light on the processes behind recessions. Now that almost all economic activity is electronically recorded, economists are able to observe much more about the behavior of businesses and consumers than they were even a decade ago.

Notice Noah's implicit assumption that recessions must be the result of people's behavior but general changes in people's behavior cannot be the result of changed aggregate quantities like private debt becoming excessive or financial cost reaching a problematic level.

Better microeconomic data will help researchers make better models of behavior. Those more realistic models -- sometimes called “microfoundations” -- will then be able to replace the old, simplistic assumptions that prevailed in previous generations of macroeconomic theories.

Now that almost all economic activity is electronically recorded, researchers can make better, more realistic models of behavior.

So now it's behavioral microfoundations.

The microfoundations will be better, Noah says, and so the macro will be better. Because everybody knows that what happens in the economy happens because of people's behavior.

What ever happened to the whole is greater than the sum of its parts ?


Remember Milton Friedman's best-remembered thought: Inflation is a monetary phenomenon. If prices are going up, it is because people are willing to pay higher prices for things. So there is behavior involved, to be sure. But if there was no behavior, there would be no transaction and no economy. If you are studying the economy, you are already studying behavior.

The results of behavior show up in the aggregates.

Friedman said that what matters is the quantity of money. That's macro. He did not say what matters is the behavior of the individual spenders. That would have been micro.

After 2007 the Federal Reserve increased the quantity of money several hundred percent. But ten years later, inflation still has not broken the 2% barrier. Was Friedman wrong? I'd say no. I'd say his statement was not general enough. It describes conditions in a "normal" economy, but not the economy of the last ten years.

You could say inflation is a spending phenomenon. You could say it is the spending that influences prices, not the money. I've said as much, myself. And then you could argue that spending is behavior and therefore the need is to focus on behavioral microfoundations. Here, my inclination is different. If spending behavior suddenly changes for the population as a whole, I expect to find the cause in the aggregates. Why? Because people respond to economic conditions, and the aggregate numbers describe economic conditions.

Aggregates are macro. Behavior is not. Behavior is the Brownian motion of gas molecules in a container. Macro is the behavior of a container of gas, measurable conditions like pressure and temperature that do not even apply to individual molecules. Boyle's law is macro.


We go shopping, or we don't. We spend money, or we don't. Only a fool would dispute that. But to insist on building up macro from scraps of human behavior is just so micro, so utterly small-minded. And to suggest that macro can be improved by including "all" the scraps of human behavior in the models, well that is plain ridiculous.

Nor is "all the scraps" a re-evaluation of what economists have been thinking for the past 40 years. It is, rather, a strengthening of commitment to the old "some scraps" idea. Using a more advanced technology to think the same old thought is not a way to rethink macroeconomic policy.