Shared publicly  - 
Imagine a crazy world in which the prices of financial assets was totally disconnected from their performance.  You could have two bonds, for example, with almost the exact same payout schedule and almost the same historical risk records, and one would cost $10 to buy and the other would cost $397.

Financial diversification would look quite different in such a world.  In particular, there should be much less of it.

Why?  Well, the intuition is that a few assets in this world are just straight out better than the rest.  When you have two bonds with similar behavior and similar prices, it makes sense to spread your money between them to create a composite asset with the same return and lower risk.  When you live in crazy world, the optimal amount of the $397 bond to buy is extremely small if not zero.

Why does this matter?  Because I would argue that charities form just such a crazy world.  The amount of good -- for almost any definition of good -- that a charity does with a $100 donation both varies widely from charity to charity and is almost entirely uncorrelated from the variance in the supplied amount of good.

(And why is that?  The primary reason is probably the absence of strong feedback loops:  no one is doing moral arbitrage by shorting donations to Harvard and going long on donations to Oxfam.  And the pluralist in me hates to say it, but the radical diversity of objective functions being pursued doesn't help either.)

And thus even if you are risk adverse in both money and goodness, you should still diversify much less your charitable giving than in your retirement portfolio.

Much much less, in fact.  Any decent savings plan will contain a stock index fund that combines hundreds to thousands of individual stocks, but if you are trying to maximize the number of lives saved per dollar (say), you might find that the best charity in the world does it for an average of 1 life / $800 and the next best does it for an average of 1 life / $1200.  Even if you are very risk adverse and even if the top charity is maximally risky (half the time they burn your money and half the time they save a life for $400), you might still prefer giving your $4800 to just the best one in order to save an expected 6 lives versus splitting your money between them equally and only saving an expected 5 lives.  And needless to say, the 99+% of charities that don't save lives or do so at worse than 1 life / $2000 or so are right out.

Finally, all of this is putting aside some very strong moral arguments that you should be risk neutral in your charitable giving.  In that case, it is a standard result that the optimal number of charities to donate to is exactly one.  (
CARE is a noble organization that fights starvation. It would like your support. The American Cancer Society is a noble organization that fights disease. It would like your support, too. Here's my adv...
Jan-Willem Maessen's profile photoMeg Muckenhoupt's profile photoFranco Baseggio's profile photoJonathan Butler's profile photo
I bet we could get Steven in for an Authors@Google talk without much trouble.
You've kind of hit on the nose why I find charitable giving difficult. It's difficult to know how much of your money actually goes toward the goal, and then how efficient they are at executing against their goal, and after that how useful in the long term their goal is.

Faced with this, one quickly reaches a point of honesty: I mostly give to charitable causes to make my self feel better that I have done something, or to causes that affect me directly, like donating to NPR.
I think it depends a little on the confidence intervals on the results of your research...
"And the pluralist in me hates to say it, but the radical diversity of objective functions being pursued doesn't help either."

And I feel like a lot of these discussions underestimate just how diverse these objective functions can be (and admittedly how selfish-ish some folks motivations can be – eg the urge to give close to home, or to people whose experiences have been similar to our own).
Measurement is a doozy. Say you have an organization that's very efficient at building wells for clean water in Africa--but the wells break down in a few years because people sell the pipes (which does happen). Are they effective or not? If they spend an increasing amount of their money fixing the existing wells instead of creating new ones, is that an efficient use of your money?
Check out IPA:  they develop and run actual RCT.  Many programs they endorse are near the top of GiveWell's rankings.  

Note, though, that your best estimates likely have material error bars that lead to similar diversification dictates if you're not risk neutral.
Add a comment...