31 January 2012

Economic models are so unrealistic



Sometimes listening to criticisms of models feels a bit like this. 
Posted by Lee Labels: , , ,

17 comments:

Ranil Dissanayake said...

I think the bigger criticism that is implied is 'that by being so simplistic, often they contrive to either miss the interactions that are most interesting or reduce themselves to stating the trite and obvious'. As an economist, I think this is deeply unfair on the best models; as a historian I think it's accurate for many.

February 1, 2012 at 7:01 AM
Unlearning Economics said...

This is funny, but...

In Milton Friedman's paper he effectively advocated building a centre made out of jelly then extrapolating that up to a larger scale whilst ignoring friction. There is a difference between a good assumption and a bad one.

February 1, 2012 at 10:41 AM
rovingbandit said...

Right - but he then says that you evaluate a model not by the realism of its assumptions but by its predictive/explanatory power. You test it empirically. Is it useful. "Good" or "bad" assumptions are irrelevant. 

February 1, 2012 at 10:45 AM
Unlearning Economics said...

The last point is an assertion and simply not how science works. In science, models are only deemed as good as their assumptions are realistic - even if they have good predictive power, there is always room for improvement by abandoning assumptions where practically feasible.

In economics this is not possible because it is often unclear exactly what the impact of the assumptions is on the analysis. In many cases (such as Arrow-Debreu), relaxing certain assumptions simply mean the model collapses. If this happened in engineering the model would not be used to build a bridge. Assumptions need to have a clear impact on the analysis and simplify rather than create a fantasy world - often they simply eliminate a known variable.

http://unlearningeconomics.wordpress.com/2011/12/01/milton-friedmans-methodology-a-critique/ 

February 1, 2012 at 12:45 PM
rovingbandit said...

I had some sympathy with this post-Keynesian stuff in university but kind of lost patience with it. Ultimately I just get a bit bored thinking about theory and methodology all day. Folks like Dani Rodrik and Sam Bowles were the watershed for me - methdology is not political - you can do mainstream neoclassical economics and have whatever kind of radical left wing policy conclusions you choose. I'm interested in applied work with policy relevance, whether theoretical or empirical (but mostly empirical), on the most important question of our time - global poverty. And there just doesn't seem to be that much work being done on this by post-Keynesians, who are mostly preoccupied with a theorist who has been dead for over 50 years, and a slightly tired and esoteric methodological debate.

February 1, 2012 at 2:12 PM
Unlearning Economics said...

I too am interested in global poverty but I feel there are enough people commenting on public affairs at surface level and the real way to mend things over the long term is to fix the broken economic models and skewed view that neoclassical economics has created. This may involve highly abstract theorising and splitting hairs over how issues are farmed, but only by meticulously demonstrating that neoclassical economics is flawed can we begin to shift the paradigm.

As Samuelson said: "I don't who writes a country's laws, as long as I can write its economics textbooks."

February 1, 2012 at 2:26 PM
rovingbandit said...

Copernicus created a revolution because he had a new and better idea. I'm not sure that rehashing Keynes counts as a new and better idea to modern mainstream economics. 

February 1, 2012 at 2:29 PM
Unlearning Economics said...

Keynes was completely misinterpreted by Hicks (admitted by the man himself) and many of his conclusions are still relevant today.

However, I wasn't simply calling for a rehash of Keynes - he's just part of a larger effort to re-write economics.

February 1, 2012 at 2:37 PM
Matt said...

""Good" or "bad" assumptions are irrelevant."

This is all I can say about that:
http://www.xtranormal.com/watch/7873033/mulberry-bag-will-develop-for-everyone 

February 1, 2012 at 2:39 PM
rovingbandit said...

My impression is that 90% of heterodox economics is about destroying neoclassical economics, and that very little is constructive. And any good bits are simple incorporated by the mainstream. Is there really a viable totally alternative paradigm?

February 1, 2012 at 2:40 PM
rovingbandit said...

Behavioural economics = incorporated by mainstream = done. 

February 1, 2012 at 2:42 PM
Unlearning Economics said...

They are but just because of that it doesn't mean they aren't constructive - if models are flawed then it's destructive to use them and constructive not to. I'm not sure about the 'good bits being incorporated' as even assuming it's true, it still protects the core of neoclassical economics and writes off the other contributions as anomalies.

Perhaps there isn't. Post-Keynesian economics is lacking in a coherent methodology, for a start, something I want to draw attention to. However, there are plenty of alternative theories and it's mostly a matter of bringing them together.

February 1, 2012 at 2:46 PM
Unlearning Economics said...

Is it? I've got an economics textbook in front of me and the index does not contain 'loss aversion' 'status quo bias' or the word 'heuristic' at all.

February 1, 2012 at 2:59 PM
rovingbandit said...

Wrong textbook obviously :)

Dean Karlan and Jonathan Zinman have one coming out soon which will be full of good behavioural and empirical development things.

But yeah, generally speaking, textbooks suck balls. 

February 1, 2012 at 3:03 PM
Unlearning Economics said...

That much we can agree on! Interesting about the new textbook, will look it up when it comes out.

February 1, 2012 at 3:09 PM
Matt said...

But it *hasn't* yet. I'm going to a seminar this afternoon on the economics of suicide, and the presented is using a standard rational choice Becker model: I have perfect knowledge of my future utility, and I I commit suicide if I'm better off in utility terms (minus the cost of suicide) for doing so. No uncertainty. No, well, mood swings. 

I find it strange for you to make the prediction argument - it lends to more simplification, not only of theoretical models, but empirical analysis. Someone who is quite naive might say "I think Christmas is caused by excessive spending in December". I could say "That's bullshit, just because they are extremely correlated, doesn't mean there is causation there." Then they could reply "If spending perfectly predicts Christmas, it doesn't matter if we're capturing the right relationship."

February 2, 2012 at 1:04 PM
Unlearning Economics said...

That suicide model is the most hilarious thing I have heard in a while. Thanks.

February 2, 2012 at 11:55 PM

Post a Comment

Subscribe to: Post Comments (Atom)
loading..