10 June 2008

Lunch with the GMU guys

I had lunch with Tyler and Alex (of Marginal Revolution fame), Bryan (EconLog) and Robin (famous unto himself but also a contributor to Overcoming Bias), and they asked me their "standard questions," i.e.,

What's the best thing to study in social sciences these days?

What do you believe that others think is crazy?

[or something like that...]

and my answers were these:

I think that cooperation is the best thing to study, because methodological individualism (i.e., looking at the world from the view of the individual) fails to capture the dynamic interactions of people in groups. As part of this study, we need to know how institutions increase or decrease cooperation; see this and these.

I think that the world is headed towards an environmental catastrophe, which I defined as a 25% decline in GDP consumption quality of life; see this post.

I was slightly-surprised to see that these answers are interrelated, i.e., if we are going to prevent or cope with such a catastrophe, then we are going to need some serious cooperation, and we (economists, not psychologists, sociologists or political scientists) surely need to understand how people in groups succeed and fail to cooperate.

Bottom Line: It's fun hanging out with those guys -- and even more fun were the things that we said off the record :)

Addendum: Here's a 2006 paper [PDF] on the application of cooperative game theory to natural resource problems (via WaterSISWEB.). I tend to prefer experimental methods, since results in CGT are usually math-driven, i.e., completely specified by the researcher -- garbage in, garbage out.

5 comments:

Robin Hanson said...

I still don't see how looking at cooperation as individuals acting prevents us from seeing how institutions could help cooperation. Nor is it clear this has much to offer in avoiding enviro collapse.

David Zetland said...

I am not saying that institutions are not important, and institutions can change the way in which people interact. What I am *emphasizing* is the study of cooperation (or the interaction of groups) and how the whole is often not the sum of the parts. This is different from meth. indiv. because it pay more attention to emergent (and dynamic) behavior more than closed-form solutions from the reaction fns of individuals.

It has much to offer in the enviro area since most enviro problems take place in the commons, i.e., where the lack of private rights makes it impossible for bilateral negotiations.

See this post for more on these themes.

Dan in Euroland said...

David,

What do you mean by experimental methods? I remember you expressing some distrust towards econometrics, so I guess that leaves lab games like Tversky and Kahneman.

However, one of my pet peeves of behavioral econ is the manipulation of the experimental design (or setting.) (I think Ken Binmore as talked about this.) With math the assumptions are available for all to see. But, with experiments, determining whether the results are generally applicable, or isolated to the particular experimental setting can be quite difficult.

I trust the math, when used appropriately, over most experiments. Although the work of John List (and Binmore's own GT experiments) has certainly made me question my current position.

With respect to this question: "What do you believe that others think is crazy?"

I think Global Warming will be a net benefit to humanity.

David Zetland said...

Dan,

You crazy son!

So -- I agree that experiments can be manipulated (see my paper on just such a topic), but math is just plain foolish -- unless you think people are "well-behaved" in terms of quasi concavity, etc.

The best thing about experiments is that anyone can try to reproduce/test your results. That's harder with econometrics (new dataset!) and useless with math, which is weak (or strong) because it's built on assumptions, lemmas, functional forms, etc....

David Zetland said...

Let it be noted that my text can be misinterpreted. By "You crazy son!", I mean something like:

"You are a crazy man, amigo", not something like SOB.... :)