Thanks for your reply and sorry for the slight delay! Would you like to present your work in a session at Founders Pledge (we have a Journal Club where we discuss relevant research and having you as a guest speaker there sounds like a good idea)?
On the cost effectiveness model On the model generality, I think my view is that even a general and simple model should not be unrealistic / systematically biased; because we trust the models more than our intuitions and generally fail—I think—to do intuitive adjustments on the model (e.g. we probably underestimate intuitively how much independence/dependence assumptions matter).
On the substance of that question, I am not sure I understand your reasoning (but see point above :)). To me it seems that when expenditure is positively correlated with success probability—what seems to be implied by a view where actors are strategic and at least mildly successful at being so—would that not (a) increase the cost effectiveness and (b) reduce the overall uncertainty?
Because we often trust models more than we should, I weakly lean towards having less models—personally, for me the conclusion “this is a really fascinating piece and now we need to think about building models for these different situations and considerations” would be fine whereas with the very rough model I see the risk of incorrect updates, e.g. people not looking into it more because they think that the model fairly represents the uncertainty and under the given range it does not look attractive for some interventions (where the benefit is less large).
But these are just personal philosophical views on modeling, weakly held.
On Baumgartner
Thanks for this, this is really clarifying.
The rephrasing makes this a lot clearer, I had originally read this as “giving more money might be useless” which would undermine the whole case for investigating advocacy charities from an EA perspective, so I am glad I misunderstood that and that this is now clearer.
I would agree that one should not update too much from this correlational evidence. Mostly, because what seems to be the dependent variable here—success probability—is itself confounded by a strategic choice to engage with issues so it is not clear that the same success probability across different levels of resources expresses no difference in strength, rather than being confounded by smaller groups not trying harder things.
On money in politics
(1) On employees as explanation of influence, I do think this is a pretty strong explanation for a couple of reasons.
(i) While employees and employers might not see eye to eye with regards to such as issues as labor policy, they have essentially the same interests with regards to the companies they own/work for—and this is where the lack of money paradox is focused, the lack of money in pork barrel political settings.
(ii) The argument does not rest on explicit voting intentions of employees, but can simply work with references to employment levels in districts, etc., it hands a very powerful argument to local business leaders vis-a-vis their political representatives.
(iii) Campaign finance as well as general influence of business leaders can lead to a situation where the political representatives are already “captured”, where there is no need for additional spending.
(iii) Empirically, it seems well-supported (or so I remember from my political science days, but I cannot find the paper so I might recall that wrongly).
(2) On lobbying equilibria, I am unsure who the relevant experts would be—but I would trust the political science / political economy literature there more than people with very local “inside view” expertise (as it is a dynamic system-level feature, something that seems more accurately to observe with data than based on individual experience).
And just to conclude on this, I think there are many cases where lobbying is probably very good, I just think that the introduction overstates this in not fully considering explanations that make this less surprising and give reasons to think that there can also be many situations where additional money will not lead to additional influence, just higher spending.
Points all well-taken. I’d love to share with FP’s journal club, though I hasten to add that I’m still making edits and modifications based on your feedback, @smclare’s, and others.
With respect to uncertainty in the CE calculation, my thinking was (am I making a dumb mistake here?) that because
Var(XY)=E(X2Y2)−E(XY)2 and Cov(X2Y2)=E(X2Y2)−E(X2)E(Y2) , then Var(XY)=Cov(X2,Y2)+E(X2)E(Y2)−E(XY)2. So if covariance is nonzero, then (I think?) the variance of the product of two correlated random variables should be bigger than in the uncorrelated counterfactual.
To me, the main value of the CE model was in the sensitivity analysis—working through it really helped me think about what “effective lobbying” would have to be able to do, and where the utility would lie in doing so. I think if it doesn’t serve this purpose for the reader, then I agree this document would have been better off without the model altogether.
Thanks for your thoughts on money in politics. Vis (1) I have to think more about this, but I do definitely view the topic a little differently. For instance, it’s not obvious to me that economic arguments and political representation do the necessary work of regulatory capture. Boeing is in Washington and Northrop Grumman is in Virginia. It seems clear that the representatives of the relevant districts are prepared to argue for earmarks that will benefit their constituents… but these companies are still in direct competition, and it seems like there’s still strategic benefit to each in getting the rest of Congress on their side. I might misunderstand- maybe we’re reaching the limits of asynchronous discussion on this topic.
Vis (2), the “inside view” I was talking about was actually yours, as someone who thinks about this professionally- so thank you for your thoughts!
1) I don’t see incompleteness as an issue—what is good for Journal Club is bringing in lots of interesting ideas which your post certainly does, updates you made and are working on are fine. So if that would work for you, I would suggest you as a speaker for Journal Club and we could see when it would fit over the next month or so?
2) My reading of your model—which might be wrong—was that you assumed independence between variables related to cost/effort and variables of success probability. It seems to me that when they are positively correlated rather than independent, cost efficiency would increase and become more narrow, because what this says is that worlds of high spending and success will be more likely to co-occur and worlds of high spending and no success less likely to occur than under independence. Does this make sense?
3) I think on money in politics my understanding is that a couple of intensely motivated politicians—e.g. the representatives where headquarters of companies are—can be quite sufficient for pork barrel style politics because they tend to fill committee positions important for their respective economic interests and they can easily bargain with other legislators.
1) Sounds good to me! We can connect about it over DM.
2) Your reading is right. A priori, a positive correlation means lower cost-effectiveness in expectation. However, I’m not sure if it means anything generally for the median cost-effectiveness (which I tried to work with in my existing CEA), irrespective of the other model parameters. And in my existing setup, if worlds of high spending and high success are more likely co-occur, and worlds with low spending and low success are more likely to co-occur, then I believe the distribution of their product would have been more dispersed, since there would be more values at the extremes (high/high and low/low) then there would be if they were independent. But I’m pretty convinced now that a better approach would have been, as you’ve suggested, to do separate CEAs conditional on various assumed interventions. Rather than change the parameters of independent distributions as I did in the posted analysis, the true next step is probably to re-model under varying assumptions about the covariance of the different variables.
3) I have a different sense of this, but not an overwhelmingly different sense, and I’m going to think about it some more.
Hi Matt,
Thanks for your reply and sorry for the slight delay!
Would you like to present your work in a session at Founders Pledge (we have a Journal Club where we discuss relevant research and having you as a guest speaker there sounds like a good idea)?
On the cost effectiveness model
On the model generality, I think my view is that even a general and simple model should not be unrealistic / systematically biased; because we trust the models more than our intuitions and generally fail—I think—to do intuitive adjustments on the model (e.g. we probably underestimate intuitively how much independence/dependence assumptions matter).
On the substance of that question, I am not sure I understand your reasoning (but see point above :)). To me it seems that when expenditure is positively correlated with success probability—what seems to be implied by a view where actors are strategic and at least mildly successful at being so—would that not (a) increase the cost effectiveness and (b) reduce the overall uncertainty?
Because we often trust models more than we should, I weakly lean towards having less models—personally, for me the conclusion “this is a really fascinating piece and now we need to think about building models for these different situations and considerations” would be fine whereas with the very rough model I see the risk of incorrect updates, e.g. people not looking into it more because they think that the model fairly represents the uncertainty and under the given range it does not look attractive for some interventions (where the benefit is less large).
But these are just personal philosophical views on modeling, weakly held.
On Baumgartner
Thanks for this, this is really clarifying.
The rephrasing makes this a lot clearer, I had originally read this as “giving more money might be useless” which would undermine the whole case for investigating advocacy charities from an EA perspective, so I am glad I misunderstood that and that this is now clearer.
I would agree that one should not update too much from this correlational evidence. Mostly, because what seems to be the dependent variable here—success probability—is itself confounded by a strategic choice to engage with issues so it is not clear that the same success probability across different levels of resources expresses no difference in strength, rather than being confounded by smaller groups not trying harder things.
On money in politics
(1) On employees as explanation of influence, I do think this is a pretty strong explanation for a couple of reasons.
(i) While employees and employers might not see eye to eye with regards to such as issues as labor policy, they have essentially the same interests with regards to the companies they own/work for—and this is where the lack of money paradox is focused, the lack of money in pork barrel political settings.
(ii) The argument does not rest on explicit voting intentions of employees, but can simply work with references to employment levels in districts, etc., it hands a very powerful argument to local business leaders vis-a-vis their political representatives.
(iii) Campaign finance as well as general influence of business leaders can lead to a situation where the political representatives are already “captured”, where there is no need for additional spending.
(iii) Empirically, it seems well-supported (or so I remember from my political science days, but I cannot find the paper so I might recall that wrongly).
(2) On lobbying equilibria, I am unsure who the relevant experts would be—but I would trust the political science / political economy literature there more than people with very local “inside view” expertise (as it is a dynamic system-level feature, something that seems more accurately to observe with data than based on individual experience).
And just to conclude on this, I think there are many cases where lobbying is probably very good, I just think that the introduction overstates this in not fully considering explanations that make this less surprising and give reasons to think that there can also be many situations where additional money will not lead to additional influence, just higher spending.
Points all well-taken. I’d love to share with FP’s journal club, though I hasten to add that I’m still making edits and modifications based on your feedback, @smclare’s, and others.
With respect to uncertainty in the CE calculation, my thinking was (am I making a dumb mistake here?) that because
Var(XY)=E(X2Y2)−E(XY)2 and Cov(X2Y2)=E(X2Y2)−E(X2)E(Y2) , then Var(XY)=Cov(X2,Y2)+E(X2)E(Y2)−E(XY)2. So if covariance is nonzero, then (I think?) the variance of the product of two correlated random variables should be bigger than in the uncorrelated counterfactual.
To me, the main value of the CE model was in the sensitivity analysis—working through it really helped me think about what “effective lobbying” would have to be able to do, and where the utility would lie in doing so. I think if it doesn’t serve this purpose for the reader, then I agree this document would have been better off without the model altogether.
Thanks for your thoughts on money in politics. Vis (1) I have to think more about this, but I do definitely view the topic a little differently. For instance, it’s not obvious to me that economic arguments and political representation do the necessary work of regulatory capture. Boeing is in Washington and Northrop Grumman is in Virginia. It seems clear that the representatives of the relevant districts are prepared to argue for earmarks that will benefit their constituents… but these companies are still in direct competition, and it seems like there’s still strategic benefit to each in getting the rest of Congress on their side. I might misunderstand- maybe we’re reaching the limits of asynchronous discussion on this topic.
Vis (2), the “inside view” I was talking about was actually yours, as someone who thinks about this professionally- so thank you for your thoughts!
Hi Matt,
1) I don’t see incompleteness as an issue—what is good for Journal Club is bringing in lots of interesting ideas which your post certainly does, updates you made and are working on are fine. So if that would work for you, I would suggest you as a speaker for Journal Club and we could see when it would fit over the next month or so?
2) My reading of your model—which might be wrong—was that you assumed independence between variables related to cost/effort and variables of success probability. It seems to me that when they are positively correlated rather than independent, cost efficiency would increase and become more narrow, because what this says is that worlds of high spending and success will be more likely to co-occur and worlds of high spending and no success less likely to occur than under independence. Does this make sense?
3) I think on money in politics my understanding is that a couple of intensely motivated politicians—e.g. the representatives where headquarters of companies are—can be quite sufficient for pork barrel style politics because they tend to fill committee positions important for their respective economic interests and they can easily bargain with other legislators.
1) Sounds good to me! We can connect about it over DM.
2) Your reading is right. A priori, a positive correlation means lower cost-effectiveness in expectation. However, I’m not sure if it means anything generally for the median cost-effectiveness (which I tried to work with in my existing CEA), irrespective of the other model parameters. And in my existing setup, if worlds of high spending and high success are more likely co-occur, and worlds with low spending and low success are more likely to co-occur, then I believe the distribution of their product would have been more dispersed, since there would be more values at the extremes (high/high and low/low) then there would be if they were independent. But I’m pretty convinced now that a better approach would have been, as you’ve suggested, to do separate CEAs conditional on various assumed interventions. Rather than change the parameters of independent distributions as I did in the posted analysis, the true next step is probably to re-model under varying assumptions about the covariance of the different variables.
3) I have a different sense of this, but not an overwhelmingly different sense, and I’m going to think about it some more.