,
7 tweets,
2 min read
Read on Twitter
1/7 Ok, here's another one. Say you want to decide how much to invest in a risky business. Do it with EE: first, consider time and write down a model for the wealth dynamic. Null model: noisy multiplicative growth.
2/7 Wealth is not ergodic in this model, so you know the expectation value doesn't tell you what happens over time.
3/7 Your task: find an ergodic observable whose expectation value does tell you what happens to wealth over time. Only one option: the ergodic growth rate. Write it down, optimize it, and you know how much to invest, given knowledge of the dynamic.
4/7 Then read the literature. What's this? Economists call the expected ergodic growth rate under multiplicative dynamics "the rate of change in expected logarithmic utility."
Ha! We have a physical interpretation of utility theory, no psychology needed.
Ha! We have a physical interpretation of utility theory, no psychology needed.
5/7 Read a bit further: Bernoulli (1738), who introduced utility long before ergodicity was known, got it slightly wrong. Laplace corrected him in 1814, but the 20th-century English translation in Econometrica of Bernoulli's Latin paper undid a good deal of Laplace's correction.
6/7 Prospect theory and behavioral economics look like someone read the Econometrica translation, not the corrected version of Bernoulli by Laplace. No wonder they need all those ad-hoc psychological biases to explain things.
7/7 And guess what: a lot of people are scientifically very uncomfortable with behavioral economics. Not because they know EE but because it just doesn't look or feel right.
