Hacker News Reader

Comment by asar | Hacker News Reader

Whenever I read longer articles on macro economic theories it feels like torture. Overly complicated explanations, the urge to model every event and attach a probability to it. A while back this was on the hn frontpage https://aeon.co/essays/how-economists-rode-maths-to-become-o... Definitely put things into perspective for me.

View on HN

I love that piece, I think the problem is the opposite though, not complicated enough models, in part because they are trying to model an absurdly complicated system with overly simplistic assumptions. This piece highlights that attempts to simplify the model (variable X doesn't matter!) have failed to fulfill their promise.

I think the problem is more that they are trying to use mathematics to model something that is mostly non mathematical in nature - human behaviour.

One problem I have with economics is the tendency to proclaim proof after just one data point and that too in hindsight.

Just like this particular article is doing. It proclaims people doubted government can simulate growth and now recession has proven them wrong. Isn't it in hindsight and just one data point?

The real problem with economics (as perceived by somebody that first studied applied mathematics, and then macroeconomics, and then went into management) is that it styles itself an annex of mathematics, but lacks the invariants necessary to construct the deep symmetries typical of physics or other hard sciences. We literally have nothing to work with other than those superficial relationships econometricians can measure and somehow boost into statistical significance.

Have you seen economic equations from, say an econ 3XX class? There's variables all over the place, under a fraction, inside an exponent, with zero regard to dimensional analysis or what meaning conversion constant ought to have (or the numerical tolerances of those equations). Then you take a derivative of that formula.

Yes I have. Your observations are not universal but pretty common. The whole field is littered with a hodge-podge of cobbled-together numerical coincidences fetishised to the level of universal truths. It can be quite embarrassing when one sits down and tries to actually do defensible, provable mathematics. That is why I eventually migrated towards management (of the family firm, not coincidentally). Much of macroeconomics, at least most of it prior to the agent-based models now being produced and analysed, is notionalism.

Confirmation bias may be unhealthy but that one sentence gave me a good chuckle.

Reply on news.ycombinator.com