My quick response to Volker Nitsch on May 9


Thanks for your paper, which I just read through quickly.  While I haven't

read it carefully, I wanted to give you some quick feedback.


I don't think I have any serious issues with what you've done.  While I

could quibble on details, most of it seems reasonable enough.  I do think

that you lose a lot of efficiency by not pooling the data across years

(which gives a lot of imprecise estimates which wander around), but that

may be a matter of taste.


I think you misunderstand me in the large though.  I've never taken my

estimate of 1.2 literally, and I don't recommend you do either.  The point

of the original paper was just to say that the effect is large, which I

believe it is (and most of your analysis bolsters that claim).  I found the

tripling to be too large for intuition when I first wrote the paper, and

all you're really doing is agreeing with me.


A few other things.  I've regenerated data from scratch twice more since

the original paper was written; once for my work with van Wincoop, and

again for my paper with Glick.  So while I readily admit to minor data

errors (e.g., the New Zealand/Netherlands issue), I've been trying to

correct them.  The stuff on language, I leave to others; I wasn't aware

that the CIA website could be so misleading; ditto with the WTO.  More

importantly, I've sort of given up on pure cross-sectional stuff, since I

think that fixed-effects estimators are the way to go (that's in my current

work with Glick).


Anyway, I enjoyed the paper, which was fun and well-written.  Please

continue to send me stuff --