People ask me how we chose oxfam to donate to exclusively. I used to say "julia did some research and thought it looked effective". Then later I could say "not only did it look effective, julia works (now worked) there and it really is well organized". I now have a third reason: their structure means they are the closest we currently get to a competitive charity market.

In general, its hard for outsiders to judge how well their money can be used. Sites like charity navigator can help you exclude truly ineffective charities (90% money to overhead? yikes!) but it's just about useless for telling the difference between good ones. All it has to go on is tax form 990. It can't tell the difference between a program that can distribute 100 anti-malarial nets for $1000 and one that can only distribute 20. Both programs would just boil down to a few statistics involving funding allocation at a very high level.

The one place I know that tries to present charity effectiveness evaluations to the people making the donations is givewell. They are in a hard position because charities are in general unwilling to make details public that would allow them to be rationally evaluated. This makes some sense for charities: it's a lot of work to collect the data, and if it makes you look bad you'll lose donors. This means that givewell is only able to review a small number of charities and has only been able to measure charities that have pretty simple programs. While its top charities (village reach, stop tb partnership, nurse family partnership, and knowledge is power program) are good, and the model of fully external evaluation is a good one, I think in the current charity market you can still do better.

Oxfam is a big charity working on lots of things. This is not in and of itself good, or even a good sign. It's work on monitoring and evaluation of projects has a big advantage over givewell's, though: it's internal. This allows them to be highly critical of their projects and even admit failure when appropriate [1] without losing donors. It also gives them far more leverage to get more information out of their projects than some place like givewell can get out of arbitrary charities they wish to review.

I don't like that oxfam spends so much money on emergency relief when I think development work does more good in the long term [2]. But I haven't found any other large charity that does as good a job of managing evaluation of internal projects as oxfam does, and I believe that in the current charity market that's what it takes to do the most good.

(Looking back over this, it looks suspiciously like trying to justify a decision I've already made. That worries me.)

Update 2011-10-17: I no longer think this makes sense. While a charity that had really effective internal monitoring might be able to beat single issue charities with external monitoring, I don't know how we could tell they really were. I now think it makes more sense to use givewell's recomendations.

[1] I don't like that donors are scared away by failure. Projects at all charities succeed and fail, what matters is that you expand and replicate the successful ones and cut your losses on the failures. Charities should not be held to the standard of "never waste a penny of donor money; it's not yours to waste", where 'wasting' is defined roughly as spending money that doesn't end up helping someone. Instead they should be held to the standard of "try and spend money to maximize the good you can do", and expanding 'wasting' to include spending money on something that does good but does not do as much good as it could be doing.

[2] It's possible that people motivated by disasters give enough to oxfam that their disaster relief work ends up subsidising their development work, if only indirectly by paying for some of the overhead.