Economic modeling

John Creighto

Mod note: this thread was originally part of http://www.physicsforums.com/showthread.php?t=601879 but has been split off due being off-topic.

I would prefer the word understand, rather then model. Well a models serve both to illustrative and verify our understanding, from an instrumentalist perspective -- that is a perspective which only asks is a model useful -- many applications of economics give bad results. Well you may say that it does well in certain situations (like perhaps micro-economics) these aren't the areas where economics makes the largest impression on people.

Sure, you can apply math to economics. Alternatively you can ask why. Economics was once much more of a philosophical field then it is now. The math is great for illustrating concepts and giving precision but it isn't the whole store.

Pyrrhus

John,

How come you understand if not modeling?

Even biological processes are modeled, infectious diseases, airplanes, the outcome of financing systems, and many more.

Definitely policies should not be based solely on "intuition".

You must model, you must collect data, and test your models.

This is what economics does now. Years ago, economics was just ideas. There was no application of the scientific method.

John Creighto

Well, first what do we mean by modeling? If I model an economic system say with an auto regressive model based on some fit to historical data then; two types of knowledge arise. The first type is synthetic knowledge in that we are applying, a prior, principles which exist independently of the system. Through these principles we measure real things but without a basis for our assumption we aren't doing much more then curve fitting.

We can of course try and justify these assumptions but in order to do this we have to make higher order assumptions. For instance we can apply various model tests and while this may give us more confidence in our model; by applying these principles we are just making more general assumptions which we again need to justify. Well we may think good results to statistical tests means a high confidence, the problem of being fooled by such tests is a problem which was well known long ago and the most popular example, is the chicken who inductively assumes the farmer won't wring his neck.

This is not to say that there is no merit to making higher order assumptions or even that it is the job of the scientist to be concerned with them. We need a form of meta-physics. We need a method of acquiring knowledge as Aristotle would call it. We may choose as our method the, scientific method and decide not to go any deeper into metaphysics as this method has proven sufficiently useful for most people to not be in need of justification.

However, in the field of economics often predictions are statistical. If results are random how can we falsify them? We would need an additional principle from which to apply the scientific method to our random results. Well in testing any theory we can certainly gain confidence through empiricism, without irrefutable evidence it is not unreasonable to ask for some rational justification beyond tenuous correlations. Our assumptions in some sense should stand on their own. As a minimum they must be plausible.

When it comes to economics we can`t uniformly embrace empiricism to justify all of economics. So often so called experts are found to support claims which in the context of history seem irrational but as far as the immediate data seems concerned they may think that they are justified.

Of course the ability to reduce complexity through the process of obtaining empirical knowledge through synthetic methods, is helpful -- in that it gives us a form of descriptive knowledge (descriptive knowledge being the second type of knowledge from blind modeling) which may help us build intuition. Without further justification it is just data.

So if empirical knowledge isn`t always reliable how is it that economics as a profession has come to accept the amount of theory it has now amassed. This didn`t happen overnight by randomly conjecturing models and empirically testing them. Much of the principles we accept today which form the edifice which justifies our economic system arose from centuries of debate about what are the underlying principles of economics.

Early economics didn`t just draw graphs and call their work done. They appealed to history, they appealed to principles of philosophy and they looked for example after example of their principle to make their ideas more concrete. All I can say is that if you want to see a different way of doing economics study older economists from history. You may not prefer their methods to those we use today but it was their hard work and debate which helped to ground the concepts we take for granted today and which are glossed over quickly in introductory economics courses.


I don`t dispute any of this. Of course well we can`t base policies solely on intuition; we can`t base them either on blind empiricism. We must strike a balance between rationalism and empiricism. To me it is the rational part from which understanding comes from; well it is the empirical part which helps to present the data in an orderly way so that we can understand it.

We will of course pursue to the greater extent, the method of acquiring knowledge which appeals more greatly to us but; both methods are essential in our pursuit of knowledge.

Pyrrhus

John,

I fail to understand your quibble. You do not discard the scientific method, and you proceed to mention that before the mathematization of economics, they ("economists" of the "past") appealed to history, philosophy, and others to make their ideas "more concrete". Also asking to me study older economists ideas.

Many economists know what you said too well, and that's why we have an order in terms of our experiments. We value higher Natural experiments, which are found in some situations. However, when we cannot we use our econometric machinery to try to tackle the other cases.

Now, note that economists are not inductive researchers, but deductive. First, there is an hypothesis. A model is developed for a very simplifying assumptions such as those found micro textbooks, and then we proceed to see how well it fits with data collected. The results are written in a paper and published. These hypothesis go through a "natural selection" process, and only few survive across the years similar to other sciences. Those that work eventually are found in books, and are taught to future students (undergraduate and/or graduate). How else is it to be done? This is science, not philosophy.

There is a reason economics became a science, and not philosophy. We need quantities, We need directions, We need to test whatever we are thinking, or else it is just a bunch of pages of conjectures.

By the way, we should probably start a new thread on this topic.