Jim Fearon

Small comment: at least in the case of Africa, the argument for struc adj predates Volcker shock in the form of the Bank’s 1981 (so called) Berg report. Berg was btw an old school on the ground institutionalist economist.

Here's another way to show the same thing: Suppose senators votes in the Senate were weighted by state population. Also showing how much more liberal median senator would be if use DW-Nominate ranking. (did this a while ago so it is a little out of date.)

This was also Nelson Polsby’s m.o., though without tuna/onion sandwiches as best I can recall. Nice touch.

110%. And highly relevant/ on target not only for political economy of development.

Jervis chapter 8, Perceptions of Centralization? I wudda thought FAE was perceiving self to be rationally responding to circumstance and constraints and others as character/disposition-driven, no?

DMZ You’re gonna miss me.

Yardbirds I’m a man.

don't know about now, but way, way back in the day this was THE place (for about 6-7 of us). Glad it's still there.

Congrats!!

Also, even given an assignment, there is still a distributional/fairness issue over which efficient bargain.

The bargain is "the same" independent of initial property rights in terms of efficiency (externality is efficiently internalized), but not distribution, which does depend on assignment of property rights. I didn't understand Coase to be defending the claim that any and all assignments were just.

maybe missing something here, but "Coasean" does not imply just, or having no distributional consequences, or not reflecting power relations (encoded in the property rights that shape who pays whom, eg).

Q sort of related to Keith's comment: In Wittman-Calvert models, politicians take positions that are more moderate than what they themselves prefer. What do AP folk think about whether/when/how much this is true in real life?

are more equal is an example of such an argument. Or eg Bas and Schub ISQ 2016. I suspect that there is a lot more that could be done here, but would agree that there is not a whole lot as of now.

(2) there are also in principle ways you can try to directly measure shocks that would be expected to increase p.i. between parties to a dispute, or factors that are expected correlate with more or less p.i. Blainey's arguments about mutual optimism being more likely when observable measures

more possible, in some cases. Also, while you may not be able to point predict for above and many other reasons, you may well be able to test the theory against broad empirical patterns that it does or doesn't imply. Alex Weisiger's Logics of War is terrific here.

are out of luck on that front. It's like predicting what will happen in a poker game if you are just watching it and can't see the cards. You can explain after the fact but not predict. There are other situations, eg, preventive war where shifts in power can be anticipated, where prediction may be

a couple of thoughts here: (1) we are not going to be able to "predict war" the way we can predict where Saturn will be in X years, for all kinds of reasons. Including that when p.i. is important (eg, someone's true willingness to use force on some matter), it is privately known so third parties

increasing the variance of 1's beliefs about c2 implies weakly increasing risk that the eqm offer is rejected (war, here). (At least, I think this is true -- I work it out in course notes and could have made a mistake.)

Concerning comparative statics directly on "amount" of private information, there are some fairly though not totally generally implications. eg: In tili model where state 2 has p.i. about own costs c2, which state 1 thinks is distributed by a log concave distribution (which covers a lot),

Then any comparative statics on any exogenous parameter of the model (e.g., observable indices of relative power) gives you testable implications.

Re your earlier claim: Take any model in which war occurs for some parameter values, due to p.i. and incentives to misrepresent it (in the sense that if had complete info war doesn't occur).

No Stereolab?

I would think that "theory" would typically refer to the arguments for _why_ "X causes Y" rather than the proposition alone, and those arguments are in turn what one needs to consider to interpret or set up the regression formula. But maybe that's a long-winded way of saying what you said.

So not just empirical claim but causal claim, which I guess is what is meant by "->".

yes. Or consider a theory that has as an implication that both x and z should be positively correlated with y in the data, but that the coefficient on x should diminish or go to zero if one adjusts for z.

Because these are not theories. (And don’t there exist actual theories such that y ~ x + z would be fine?)

No idea how important this is in this case, but the organizational design problem of “connecting the dots” produced by a huge number of intel sources is usually underrated. This problem is not about psychology or deception.

LfL but always thought New Values was underappreciated.

Partly motivated by concern about Saudi-China relations. Not endorsing this at all, just saying.

Counterinsurgency