Some (Rough) Thoughts on the Value of Campaign Contributions

I have a num­ber of non-EA friends who con­tribute to poli­ti­cal cam­paigns, and have seen a few EAs on­line won­der (sort of in pass­ing) whether cam­paign con­tri­bu­tions are worth­while. I started to won­der about this as well. How­ever, I did a bit of dig­ging around, and my ini­tial con­clu­sions are that your de­fault at­ti­tude (as an EA-minded per­son) should prob­a­bly be to not donate to poli­ti­cal cam­paigns un­less it’s an ex­cep­tion­ally good op­por­tu­nity. The vast ma­jor­ity of the time, the marginal con­tri­bu­tion of an in­di­vi­d­ual donor is neg­ligible.

My origi­nal plan was to not rein­vent the wheel and do a com­plete re­view of the rele­vant liter­a­ture. Un­for­tu­nately, as it turns out, this is quite time-con­sum­ing. So, rather than post noth­ing, I thought I’d just post some ini­tial thoughts, and let the com­mu­nity com­ment on them.

A few caveats:

  • The dis­cus­sion be­low per­tains ex­clu­sively to fed­eral elec­tions in the US (elec­tions for the House, the Se­nate, and the pres­i­dency). How­ever, I ex­pect many of the con­clu­sions will have rele­vance for lo­cal and state elec­tions, and maybe even for elec­tions in other coun­tries.

  • There are many ways you can in­fluence poli­tics with money. I’m not go­ing to ar­gue money plays no role in in­fluenc­ing poli­tics, or that all pos­si­ble ways of in­fluenc­ing poli­tics with money are quite as low im­pact as cam­paign dona­tions (for the ex­plicit pur­pose of in­creas­ing the prob­a­bil­ity your can­di­date gets into office).

  • I figure many EAs already ac­cepted some ver­sion of the con­clu­sion I reach be­low. I don’t think very many EA dona­tions, if any at all, are go­ing to poli­ti­cal cam­paigns. Still, there is some­thing valuable in pro­vid­ing EA-style analy­ses of plau­si­ble in­ter­ven­tions, if only to rule them out.

With all of that said, a ba­sic ar­gu­ment in defense of cam­paign con­tri­bu­tions might go like this:

US fed­eral elec­tions are dom­i­nated by the biggest spenders. Can­di­dates spend in ag­gre­gate $4-6 billion dol­lars each elec­tion cy­cle, and the can­di­date who spends more al­most always wins. Elected offi­cials de­vote a sub­stan­tial por­tion (be­tween 17-18%) of their time to fundrais­ing. It would seem, from all of the money be­ing gath­ered and thrown around in fed­eral elec­tions, that the amount of money a can­di­date spends, and there­fore the amount of money a can­di­date raises, is a sig­nifi­cant fac­tor in how suc­cess­ful their cam­paign will be.

More­over, the out­comes of (some of) these elec­tions are (prob­a­bly) enor­mously con­se­quen­tial. It’s hard to pre­cisely de­ter­mine just how con­se­quen­tial, but I think it’s safe to say the fol­low­ing: (1) de­ci­sions made by elected offi­cials in the US will have a large im­pact on cur­rent and fu­ture gen­er­a­tions (2) in some cases, one can­di­date will have a much greater pos­i­tive im­pact as an elected offi­cial than the other (in the coun­ter­fac­tual sense that the world in which one can­di­date is elected will be mean­ingfully bet­ter than the world in which the other can­di­date is elected).

This amounts to rough EV rea­son­ing: (1) money can in­crease the chance can­di­date X is elected and (2) can­di­date X’s be­ing elected will pro­duce a lot of good. How­ever, I think both of these premises are a lit­tle mis­lead­ing. The first is eas­ier to de­bunk than the sec­ond, so I’ll start there.

Does Money Ac­tu­ally In­fluence Elec­tion Out­comes?

The FEC has fairly strict reg­u­la­tions on re­port­ing cam­paign con­tri­bu­tions, so we for­tu­nately have a lot of data on how much money can­di­dates spend and how they spend it. Poli­ti­cal sci­ence re­search in the past few decades has used this data to study the re­la­tion­ship be­tween spend­ing and elec­toral suc­cess. Yas­min Da­wood sum­ma­rizes the state of the liter­a­ture in this com­pre­hen­sive liter­a­ture re­view from 2015:

Both in­cum­bents and challengers put a great deal of effort into fundrais­ing, which sug­gests that they think that rais­ing money is re­quired for win­ning elec­tions. The aca­demic liter­a­ture, how­ever, has not been able to con­clu­sively es­tab­lish a causal con­nec­tion be­tween in­cum­bent spend­ing and elec­toral suc­cess (Milyo 1999, Strat­mann 2005). In­cum­bent re­elec­tion rates are at sky-high lev­els, and be­cause in­cum­bents out­spend challengers by ap­prox­i­mately a three-to-one ra­tio, one might think that cam­paign spend­ing pro­vides an elec­toral ad­van­tage. The puz­zle, though, is that in­cum­bent spend­ing does not ap­pear to be effec­tive (Ja­cob­son 1990, Abramow­itz 1991, Milyo 1999, Ja­cob­son 2003). Cam­paign spend­ing by US House in­cum­bents does not af­fect vote shares, al­though challenger spend­ing does in­crease vote shares (Le­vitt 1994).

There are a few im­por­tant bits to draw out from this quote. The first is that the liter­a­ture seems to agree that it’s much more com­pli­cated than “spend X more dol­lars for Y more votes.” The sec­ond is that, his­tor­i­cally, challenger spend­ing has been viewed as much more effec­tive than in­cum­bent spend­ing. And while this may be true, it seems un­likely to me that the differ­ence is very sig­nifi­cant.

In fact, the study cited above (Le­vitt 1994) ac­tu­ally came to the con­clu­sion that challenger spend­ing was less effec­tive than pre­vi­ous stud­ies in­di­cated, and the rea­son is quite in­for­ma­tive. Referenc­ing work done by James Sny­der, he writes: “Failure to con­trol for can­di­date qual­ity will lead to an up­ward bias in the es­ti­ma­tion of the im­pact of challenger spend­ing be­cause high qual­ity challengers will have a greater like­li­hood of win­ning and there­fore will be able to raise a greater vol­ume of cam­paign con­tri­bu­tions.” In other words, higher qual­ity challengers (i.e. challengers who are more charis­matic, or pro­pose bet­ter poli­cies, or mesh more eas­ily with the elec­torate, or what­ever) tend to at­tract more dona­tions. Ex­plain­ing the cor­re­la­tion be­tween spend­ing and elec­toral suc­cess with the claim that spend­ing more causes elec­toral suc­cess seems to be an al­most text­book ex­am­ple of the cor­re­la­tion/​cau­sa­tion fal­lacy.

This fal­lacy ex­plains, in large part, why the aca­demic liter­a­ture has failed to find a causal re­la­tion­ship be­tween spend­ing and elec­toral suc­cess in gen­eral. Gary Ja­cob­son writes:

Ques­tions about how cam­paign spend­ing mat­ters are re­ally ques­tions about how cam­paigns mat­ter, for al­though the point is of­ten left im­plicit in the liter­a­ture, the amount of money spent is just a handy sur­ro­gate for what is re­ally ex­pected to in­fluence vot­ers, the to­tal cam­paign effect, its qual­ity as well as quan­tity. To be sure, spend­ing mea­sures cam­paign­ing with con­sid­er­able er­ror, but there is no ev­i­dence that the er­ror is sys­tem­atic, and as long as we do not pre­tend to es­ti­mate effects with greater pre­ci­sion than war­ranted (“spend an­other sev­enty-five thou­sand dol­lars and you’ll get 1,572 more votes”), we are not likely to be led too far astray.

So while spend­ing prob­a­bly has some im­pact on elec­tion out­comes in most cases, it has much less of an im­pact than one might at first sus­pect. Spend­ing is prob­a­bly still a good mea­sure of the size and pop­u­lar­ity of a poli­ti­cal cam­paign, so how much money a cam­paign spends gen­er­ally gives you a good es­ti­mate of how likely that cam­paign is to be suc­cess­ful. Spend­ing, how­ever, is only one small com­po­nent of how poli­ti­cal cam­paigns are suc­cess­ful, along with can­di­date qual­ity, the staff’s qual­ity, the cam­paign­ing strat­egy, and many other fac­tors. And, of course, if a cam­paign has no more money, that’s the end of the road for that cam­paign. But my guess is that the lack of fund­ing is usu­ally a symp­tom, rather than the cause, of the real prob­lems with that cam­paign.

The most sig­nifi­cant im­pact spend­ing has is when a cam­paign is just start­ing out (es­pe­cially for challengers, who need to get their name out there), but af­ter a cer­tain amount of money, ad­di­tional fund­ing starts to have diminish­ing re­turns.

The clear­est con­clu­sion we can draw from all of this is best pro­vided by Jeffrey Milyo:

One im­por­tant im­pli­ca­tion of re­cent em­piri­cal re­search on cam­paign fi­nance is that the marginal value of a cam­paign con­tri­bu­tion is quite small, both to the re­cip­i­ent and to the donor, be­cause cur­rent law limits the size and source of con­tri­bu­tions and be­cause marginal cam­paign spend­ing ap­pears to have lit­tle effect on elec­toral out­comes.

How much of a pos­i­tive im­pact could your poli­ti­cian have?

I think you can, in a sort of naïve and in­com­plete way, ap­ply the STN frame­work to in­di­vi­d­ual poli­ti­cal cam­paigns. Although al­most all of the time your con­tri­bu­tion will make lit­tle-to-no differ­ence, per­haps you could view cam­paigns as in­di­vi­d­ual in­ter­ven­tions, and as­sume they vary sig­nifi­cantly in effec­tive­ness, and figure that there are some cam­paigns (few-and-far-be­tween) where your con­tri­bu­tion may mean­ingfully in­crease the prob­a­bil­ity that a re­ally good can­di­date is elected to office (and there­fore be worth­while from a marginal cost-effec­tive­ness point-of-view):

(1) How knowl­edge­able and com­pe­tent is the can­di­date and how con­se­quen­tial is the office they’re run­ning for? What poli­cies do they sup­port?

(2) How effec­tive are the cam­paign­ing strate­gies they are us­ing (and how well is the cam­paign or­ga­nized in gen­eral?)

(3) How early are you con­tribut­ing in the elec­toral pro­cess and how much po­ten­tial for fund­ing does the can­di­date cur­rently have?

(1) is scale/​im­por­tance, (2) is tractabil­ity (how effec­tively will your con­tri­bu­tion be used, how plau­si­ble is it that the can­di­date wins, etc.), and (3) is ne­glect­ed­ness/​room-for-more-fund­ing. The main differ­ence be­tween this frame­work and STN is that nor­mally, an in­ter­ven­tion only has to be un­usu­ally satis­fac­tory along one axis of STN to qual­ify as “high im­pact”, whereas it seems like a cam­paign will need to perform well along all three of these axes to qual­ify as “high im­pact.” If the can­di­date does not seem es­pe­cially com­pe­tent, or the cam­paign is poorly or­ga­nized, or the cam­paign is quite late in the elec­tion pro­cess and already has a lot of fund­ing, mak­ing a dona­tion would prob­a­bly not be worth it.

So let’s ac­cept that you’ve iden­ti­fied one of these rare cam­paigns, and (while we’ve seen ear­lier than con­tri­bu­tions are not gen­er­ally very effec­tive) a con­tri­bu­tion might be effec­tive here. Why might you still be hes­i­tant to con­tribute to the cam­paign?

Well, the first rea­son is that the can­di­date is just not par­tic­u­larly coun­ter­fac­tu­ally effec­tive. While I don’t have any data here to sup­port this, I do have a strong feel­ing that many peo­ple vastly over­rate the coun­ter­fac­tual im­pact of their fa­vored can­di­date. My per­sonal bias is that this has been hap­pen­ing a lot in the 2020 Demo­cratic pri­mary cam­paign. Some­one will say some­thing like: “if my can­di­date is elected, we’ll re­ally start to see some profound struc­tural/​sys­temic changes!” They usu­ally ne­glect to take the fol­low­ing sorts of ques­tions se­ri­ously: what sort of struc­tural/​sys­temic im­ped­i­ments are there to your can­di­date im­ple­ment­ing said policy? How con­fi­dent are we that this can­di­date’s policy will have its de­sired effect? What sort of knock-on, sec­ond-or­der effects might the policy have? What sort of bi­ases might be caus­ing me to over­rate how much bet­ter this can­di­date is from their com­pe­ti­tion? And so on.

As usual, we need to think about im­pact in a coun­ter­fac­tual way. And I think this is re­ally hard to do. Some good con­sid­er­a­tions may be: How likely is a given can­di­date con­tribute to/​alle­vi­ate the pos­si­bil­ity of par­tic­u­lar x-risks? Does this can­di­date have un­usual fo­cuses or abil­ities that will be un­likely to be matched by other can­di­dates (such a can­di­date may sup­port farm an­i­mal welfare or dis­cuss his con­cern for the global (not merely na­tional) poor)? My un­for­tu­nate hunch is that many poli­ti­ci­ans – un­less they’re dis­tinctly quirky/​ex­cep­tional! – are some­what in­ter­change­able.

There’s been some EA rea­son­ing in this do­main. I found this ar­ti­cle from 2016 where Rob Wiblin writes about the im­mense im­pact one pres­i­dent might have over an­other. I think a lot of his com­ments are spot on, but there are a few ways they don’t ap­ply here:

  • Much of what he’s talk­ing about is ab­solute im­pact, not coun­ter­fac­tual im­pact.

  • Most peo­ple would agree Trump was an anoma­lous can­di­date. The differ­ence be­tween Hilary and Trump is likely much greater than that be­tween your av­er­age cen­ter-right Repub­li­can and cen­ter-left Demo­crat (and even less be­tween two cen­ter-left Democrats, like many in the pri­mary).

  • The im­pact of the US pres­i­dent is prob­a­bly greater (by at least an or­der of mag­ni­tude) than a mem­ber of Congress.

  • Yes, if you were con­fi­dent that one can­di­date had an X% lower chance of caus­ing a war (and es­pe­cially of caus­ing an x-risk, as men­tioned pre­vi­ously) that would be a good rea­son to as­sume they’d have a greater coun­ter­fac­tual im­pact.

You should also be con­cerned that, aside from coun­ter­fac­tual im­pact, you might just be wrong about ab­solute im­pact. Holden Karnofsky has also made some in­ter­est­ing and rele­vant com­ments about the risks of “ad­ver­sar­ial philan­thropy” in a blog post from 2013:

I’ve long been wary of giv­ing op­por­tu­ni­ties that in­volve tak­ing on other peo­ple as ad­ver­saries, and I think a lot of our au­di­ence shares my mis­giv­ings. One rea­son for this is that pro­jects with ac­tive, in­tel­li­gent op­po­si­tion are likely to have more difficult—and un­pre­dictable—paths to suc­cess. Another rea­son is the po­ten­tial difficulty of be­ing on the right side. When work­ing on con­tro­ver­sial is­sues, one can eas­ily be blinded by per­sonal bi­ases and ide­ol­ogy into be­liev­ing a par­tic­u­lar change is more de­sir­able than it is, with the re­sult that even a “suc­cess” can end up do­ing more harm than good.

He goes on to say his fears have been less­ened be­cause he’s be­come more aware of the fact that there are “nat­u­ral asym­me­tries of money and or­ga­ni­za­tion be­tween differ­ent sides on a policy is­sue. The side fa­vored by a con­sen­sus of in­formed hu­man­i­tar­i­ans can be sig­nifi­cantly (and im­por­tantly) un­der-re­sourced rel­a­tive to the side with a struc­tural ad­van­tage.” This is a good re­vi­sion for many other kinds of philan­thropic in­ter­ven­tions in poli­tics, but I think “nat­u­ral asym­me­tries” ap­ply less in the case of cam­paign con­tri­bu­tions (of­ten, as men­tioned in the pre­vi­ous sec­tion, if a can­di­date has a real chance, they have enough fund­ing). The “difficulty of be­ing on the right side” is still very real.


If you were to as­sume money buys votes, and ar­gue against con­tri­bu­tions from a marginal cost-effec­tive­ness point-of-view (spend $X to have a Y% greater chance of Z im­pact), I think you’d be suc­cess­ful. But it’s a tougher case to make than what I’m say­ing. Spend­ing $X on most cam­paigns prob­a­bly won’t budge that Y% at all. Even if that Y% budges, your coun­ter­fac­tual im­pact Z is likely to be quite low.

The one thing I didn’t add, be­cause EAs have dis­cussed it quite a bit, are the rep­u­ta­tional ad­van­tages of stay­ing apoli­ti­cal. I haven’t thought too much about these, but they also seem to fa­vor my po­si­tion.

Fi­nally, in this post, I’ve as­sumed that all your money is buy­ing you is an in­creased chance your can­di­date is elected. One set of rea­sons to con­tribute might be if your money is buy­ing you some­thing else. For in­stance, maybe you’re in a po­si­tion to benefit from sig­nal­ing to the right tribes. Your dona­tion may in­fluence other donors to con­tribute, or, if you con­tribute enough, maybe you’ll get to meet the can­di­date and in­fluence them in some way. If you’re in this unique po­si­tion where your con­tri­bu­tion is buy­ing you these “ex­tra bonuses”, then you should ac­count for whether they’re worth the cost. For most of us, they don’t ap­ply.