This is right, and important, especially important if you're someone starting out on career in economics. If you want to contribute to debates over macroeconomic policy, the models you will learn in a mainstream graduate program will do you no good at all. noahpinion.substack.com/p/the-return-o… 
I don't however agree with the reason he gives. It is perfectly possible to make useful macroeconomics models. Policy is always based on an implicit model of some kind. As soon as you've talk about a fiscal multiplier, you're using a model. Okun's law is a model. Etc.
The problem is with the specific kind of theory mainstream theory aspires to - a model of an abstract "economy" built up from the level of individual agents, where we can identify actual outcomes with a unique equilibrium. That indeed is hard - in fact it's impossible.
What we can aspire to is a bunch of partial models for particular questions, that start from relationships between observable aggregates rather than the behavior of individual agents, embedded in a non-formalized understanding of the economy at a particular place and time.
There are too many meso-level structures in our economy for there to be any way to build a useful model up from individual behavior. Firms, markets, wages and other macro prices, geography etc. matter a lot, and have concrete history which can't be derived from first principles.
To use models effectively, they have to be embedded in a deep natural-history knowledge of the economy, to know questions they should answer, what is reasonable and relevant. Many mainstream economists do possess this kind of knowledge! But it's not what is taught in grad school.
• • •
Missing some Tweet in this thread? You can try to
force a refresh
