July 5, 2019: Weekly Forecasting Update
"A Republic, If You Can Keep It": Weekend Reading

Socialist Calculation Debate 2.0: Sparsity: On Twitter

School of Athens

Arindrajit Dube and Friends: On Twitter: Socialist Calculation Debate 2.0: "Arindrajit Dube: Prediction: there will be a socialist calculation debate do-over, but the version 2.0 will be focused on a bet on sparsity...

...Suresh Naidu: Right: instead of trying to get all the prices right, the regularization bet is that is that 99% of the benefits of planning can be realized with 1% of the prices planned by the state (e.g. fed rates).

Arindrajit Dube: "The commanding heights were a great idea but need ML to find them"

Brad DeLong: We already have such bets. Milton Friedman bet that fixing the growth of nominal liquidity services at a constant proportional rate got us 99% of the benefits. Alan Greenspan bet fixing the price of nominal liquidity services at his guess at Wicksellian neutral rate got us 99% of benefits...

Suresh Naidu: Yeah my example of fed rates wasn't accidental. Between the central bank and the tax code (Diamond-Mirrlees: tax rates=price setting) neoclassical econ already yields a sparse set of tools to manage the economy without gosplan. A good Austrian would say that is the whole problem...

Suresh Naidu: I agree with @glenweyl that a way to think about the problems with planning/bureaucracy is overfitting. But the solution to overfit isn't to not try to predict, but to regularize the prediction.

(((E. Glen Weyl))): 100%. See RadicalxChange: "Joining artists, entrepreneurs and scholars to heal political divides and inequality through political and economic mechanism design." http://radicalxchange.org I think a lot of James Scott issues can be viewed in that light...

Erik Brynjolfsson: FWIW, I don't think the calculation is the most important or hardest part (even if they're the once econ 101 models). Power and incentives matter a lot more. Decentralized asset ownership => decentralized power and incentives.

Suresh Naidu: I'm sympathetic to this, but have spent enough time around the econophysics crowd to realize a big cost of decentralization is unavoidable inequality (just from maximum entropy principles), so cant fetishize decentralization alone).

Erik Brynjolfsson: My impression is mostly the opposite. But I guess it depends on how you define and implement decentralization, and a lot of other details.

Suresh Naidu: The lesson of the thermodynamics view of markets,I think, is that economic decentralization at the individual level (even with equal initial endowments) results in quite skewed distributions of whatever is conserved by the economic system (eg money or purchasing power). Here's a survey of that literature (lot of this orbits the Santa Fe institute) https://t.co/cSMz491LEf?amp=1.

This File: https://www.bradford-delong.com/2019/07/socialist-calculation-debate-20-sparsity-on-twitter.html
Edit This File: https://www.typepad.com/site/blogs/6a00e551f08003883400e551f080068834/post/6a00e551f0800388340240a46cf3c3200c/edit
Economics Gone Right: https://www.bradford-delong.com/economics-right.html