*** The Official MAGAposting thread ***

[Poopadoop]

My basic ideas of quantitative models of economics are:

a) Even if you devised the perfect model, there isn't enough information to feed into the model to allow it to make predictions. I.e., too much of the relevant input to the model is unknown or unknowable.

b) Even if you had all the relevant input, the number of variables and their interactions mean you would need a computer (literally) the size of the moon to do the calcs.

[wufwugy]

My basic ideas of quantitative models of economics are:

a) Even if you devised the perfect model, there isn't enough information to feed into the model to allow it to make predictions. I.e., too much of the relevant input to the model is unknown or unknowable.

b) Even if you had all the relevant input, the number of variables and their interactions mean you would need a computer (literally) the size of the moon to do the calcs.

Yeah that's basically right. Because of this, there is debate among economists about what economics even should be about. For example, there is a big econometrician movement. I don't like it much because it attempts to do what you pointed out can't be done. Interesting to note, Nassim Taleb often discusses how econometricians gets their statistics wrong in the first place. So that's, like, a double negative. I think econometrics is very overused. On the econometric paper I replicated for a class, I felt like the "results" boiled down to data dredging and perhaps overfitting and that the conclusion was easily wrong because of unknown confounding variables. And yet this was a good enough study it was in the textbook.

On the other side are economists who say something along the lines that economics is more about construction of logic and constraints that best model human behavior and resources that we can then derive concepts from. I agree with this side more. The model from which virtually all other economic models derive from is like this, supply and demand. That model is basically about constraints and logic.

[wufwugy]

Yeah that's basically right. Because of this, there is debate among economists about what economics even should be about. For example, there is a big econometrician movement. I don't like it much because it attempts to do what you pointed out can't be done. Interesting to note, Nassim Taleb often discusses how econometricians gets their statistics wrong in the first place. So that's, like, a double negative. I think econometrics is very overused. On the econometric paper I replicated for a class, I felt like the "results" boiled down to data dredging and perhaps overfitting and that the conclusion was easily wrong because of unknown confounding variables. And yet this was a good enough study it was in the textbook.

On the other side are economists who say something along the lines that economics is more about construction of logic and constraints that best model human behavior and resources that we can then derive concepts from. I agree with this side more. The model from which virtually all other economic models derive from is like this, supply and demand. That model is basically about constraints and logic.

Here's a good way of looking at it.

The best models in economics are VERY clear about the ceteris paribus effects of the minimum wage. So, what have the studies shown? Um, well, um, the results are mixed. In fact, certain economists who study it a ton and have an opinion on it curiously seem to find their studies showing the results they want. I'm not suggesting any foul play, but pointing out how experimental economics is very, very hard.

In my estimation, we will never know what the data say about the minimum wage. There are just too many confounding variables. Yet, the logic behind the ceteris paribus effects of the minimum wage are very clear and undisputed by economists.

[Poopadoop]

Interesting to note, Nassim Taleb often discusses how econometricians gets their statistics wrong in the first place.

This relates to my point about a) not having enough information. If there's not enough data to devise a model that gets its basic assumptions about parametrization right, your model is destined to fail.

Stats nerd alert: Taleb's argument was that making assumptions about economic models being normally-distributed when they were better described as fat-tailed is the core flaw of many economic theories, but his is just a specific example of the general principle that if you don't parametrize your model properly, you're fucked - an interesting side note is that the American Statistical Association has said the same thing recently about using p-values in scientific research, or more to the point, using them wrongly)