A while ago, I discussed the limits of the deductive approach in economics (and, for that matter, in anything). Basically, one can be really clever and derive how the world could or should work. This might or might not have anything to do with how the world actually works.
Robert Waldmann takes Matthew Yglesias to task for deducing, when he should be inducing about the effects of state income taxes on the decision by rich people to leave that state:
You specifically write “The tax competition issue is real, but limited, and the further you get from New Jersey the less real it becomes.” “Real” in this context must be some sort of quantitative qualifier* If so you contest the work of Cristobal Young and Charles Varner forthcoming in June 2011 in The National Tax Journal. (warning pdf)
They write
The New Jersey millionaire tax experiment offers a potent testing ground, given the magnitude of the policy change and the relative ease of relocating to a different state tax regime without leaving the New York or Philadelphia metropolitan areas. Using a difference-in-difference estimator, we find minimal effect of the new tax on the migration of millionaires.
What is less real than “minimal” ?
I am not upset because you missed the article especially since it isn’t, technically published yet. I wouldn’t be surprised if other researchers vehemently contest the conclusions of Young and Varner. In fact, I have no doubt provided one counts Koch financed hacks as researchers.
I am distressed, because you wrote ” I actually think this kind of question is better investigating initially through theory.” I don’t see how a question about the real world could be investigated through theory. I thought the idea that we could learn about the word through theory without evidence was proven to be nonsense in the 17th century. I think you mean to write “we should think about this issue before deciding which data to look at” but the theoretical work is the pre-investigation work, not another kind of investigation.
I am an economist and I know many economists who think that theory without evidence constitutes investigation. I frequently tell them otherwise.
Amen. Selah.
An aside: If I’m not mistaken, Waldmann has had some training as a biologist, which might explain his predilection for those stupid fucking natural history facts.
mikethemadbiologist
Thanks for the link. Yes I have a BA in biology.
I was very hard on Yglesias over what I now think was basically a slip of the keyboard. I he had said we should develope hypotheses before testing them with data, I would have had no problem. That’s what he meant.
Robert Waldmann:
“If he had said we should develope hypotheses before testing them with data, I would have had no problem. That’s what he meant.”
That’s what I thought he meant, and I have no problem, either.
However, I think that that is wrong. There is a lot to the idea that look-see experiments without having formed a hypothesis is not the best way to advance science. Science is, after all, a battle of ideas.
At the same time, while hypotheses can help us to organize data and focus research, they can also blind us. There is a place for Beginner’s Mind, as they say in Zen.
Do higher top marginal tax rates cause those affected to flee the jurisdiction? Do you really need a theory to answer that question?
I often feel that the social sciences could use less theory and more hard facts.
Well, yes. Yes you do.
If higher taxes make people flee, then why are there any people left in higher-tax states? Because that is not everything by a long shot. Other things that matter include proximity of family, recreational opportunities, quality of health care, availability of decent jobs (if your job more than makes up for the tax-rate difference, stay put), etc.. A decent theory will figure out the mixes of all these sorts of factors.