Comments on: Thoughts on the use of Models in Economics

By: Vacslav

Vacslav — Mon, 11 Jul 2016 05:07:38 +0000

A good model captures essential and persistent properties of the system. Almost all models in economics are not models at all. Rather, they are heuristics. They capture in the form more resembling a “rule of thumb” some properties of economic systems holding sometimes under some conditions.

A good model is also good for something – to explain or to predict.

The law of electromagnetic induction is a good model. It operates with well-measured quantities such as currents, electromagnetic fields, properties of materials. And it can explain why a strong magnets slows down its fall in a copper tube https://www.youtube.com/watch?v=lnY9u8TfFKI. This law holds everywhere and every time because it deals with fundamental properties of matter.

Models in economics are heuristics and the laws of economics are not fundamental. For the law of demand to “hold” there needs to be a fundamental process to produce the result. Such fundamental processes do not exist. Good, plausible stories and rules of thumb are not fundamental.

One last thing. A decent airline or hotel reservation system is a well-designed engineering system built around “the law of demand”. A decent reservation system never assumes “all other things being equal”. Rather, it takes “the law of demand” as a good starting point, a story, and attempts to build predictive heuristics and rules of thumb and make probabilistic inferences around it.

By: Andrew'

Andrew' — Sun, 10 Jul 2016 23:05:52 +0000

In the former life, we had numerous models for the consumer product. There was the finished product model that wa used to predict performance. Ther was the model that predicted how the assembly would be formed into the finished product. The design that wedded the two together. There were process models that estimated the effects of the process on going from the build to the finished product. Then the perfirmance models. All these models were always being tweaked with constant feedback, feature upgrades, rebuilds and replacements. I never got the impression that people believed the map was the territory, but you had to have a measure of faith, but everyone was scared.

By: RAD

RAD — Sun, 10 Jul 2016 13:07:01 +0000

Arnold, I think you are slowly converging on my views about engineering and non-linear systems. “All other things equal, if the assumptions of this model hold…” is how engineers solve non-linear systems. You restrict enough variables until you have an approximation of a linear/deterministic system. You make assumptions about what variables are critical and what range of values the equations remain valid for.

The early models are always terrible and they fail miserably in unexpected ways in the real world but with each failure you can improve the model. The problem is not with attempting to model non-linear systems, the problem is the assumption that unproven models are deterministic enough to be usefully applied.

By: Andrew'

Andrew' — Sun, 10 Jul 2016 08:04:00 +0000

The problem with engineers is ultimately we have to ship, so that forces us to ultimately relief on the model, but the majority of total effort is in feedback testing and tweaking, in my limited experience.

By: Andrew'

Andrew' — Sun, 10 Jul 2016 08:01:29 +0000

“All other things equal, if the assumptions of this model hold, then we will observe that when X happens, Y happens.”

Right before I read this I was thinking this is more like how engineers usually think of models than like flight simulators. Maybe this recurring analogy is only irksome to other engineers, so if your model of engineering models is effective, carry on as you were.

By: Lord

Lord — Sat, 09 Jul 2016 15:03:20 +0000

Too often though, the model is treated as the reality and reality ignored.