When I talk about the poor distribution of funds in charity, people in the effective altruism movement sometimes say, "Didn't Holden Karnofsky show that charity is an efficient market in his post Broad Market Efficiency?"

My reply is "No. Holden never said, and doesn't believe, that charity is an efficient market."

What is an efficient market?

An efficient market is one in which "one cannot consistently achieve returns in excess of average market returns... given the information available at the time the investment is made." (Details here.)

Of course, market efficiency is a spectrum, not a yes/no question. As Holden writes, "The most efficient markets can be consistently beaten only by the most talented/dedicated players, while the least efficient [markets] can be beaten with fairly little in the way of talent and dedication."

Moreover, market efficiency is multi-dimensional. Any particular market may be efficient in some ways, and in some domains, while highly inefficient in other ways and other domains.

Charity as an inefficient market

Financial markets are relatively efficient. It's rare for players to consistently beat the market by a large margin. You can beat the average by investing in a low-fee index fund, but not by a lot, and it's hard to beat hedge funds.

Philanthropic markets appear to be less efficient than financial markets in many ways. In charity, one can consistently beat the market by a wide margin simply by giving to GiveWell's recommended charities, which achieve far greater returns (in social value) per marginal dollar than the average charity does. However, Holden points out that it has been surprisingly difficult for GiveWell to find ways to beat well-run large foundations like the Gates Foundation's work in global health.

Why should we expect charity to be less efficient than financial markets?

For one thing, most people giving to charity don't even seem to care what returns (in social value) they're getting with their investments. That's why, when proto-GiveWell initially contacted a bunch of charities to ask for evidence of positive impact, some of those charities reported that nobody who gave them money had ever asked that question before. And when charities sent proto-GiveWell their internal reports about effectiveness, they were so inadequate that they "led [proto-GiveWell] to understand that the charities themselves did not know whether they were helping or hurting a given situation" (Stern 2012).

For another thing, "market incentives of the nonprofit world push charities toward happy anecdote and inspiring narrative rather than toward careful planning, research, and evidence-based investments" (details here).

Also, as Brian Tomasik notes, "Efficiency in the realm of charity is inherently less plausible than in financial markets because in charity there’s not a common unit of what 'good' means... Indeed, one man’s good may be another man’s bad (e.g., abortion, gun control, extinction risks)." But even when we focus on relatively common units of 'good' (e.g. human welfare, QALYs, or DALYs), charity is still relatively inefficient: we can easily purchase more QALYs per dollar via AMF than via, say, the popular Make-a-Wish Foundation.

What is "broad market efficiency", then?

If Holden agrees that philanthropic markets are relatively inefficient in the sense that it's easy to consistently and substantially beat average market returns by giving to GiveWell's recommended charities, then what does he mean by "broad market efficiency"? Holden introduces "broad market efficiency" as a term for the spectrum of market efficiency, but remains uncertain as to where charity falls on that spectrum of market efficiency.

Brian Tomasik worried that the term "broad market efficiency" would confuse some readers into thinking Holden was claiming that philanthropic markets are relatively efficient and thus that "it doesn’t really matter where you donate." Holden said he wasn't worried about this, saying, "I don’t think 'broad market efficiency' is a common phrase or one with a clear meaning." But I think the phrase is confusing, that many readers interpret it as meaning "market efficiency," and indeed that people in economics and finance sometimes use it that way: search for the phrase "broad market efficiency" here, here, here, and here.

The research ahead

So how efficient is the charitable market, and in which ways? My own guess is that it's far less efficient than financial markets, but GiveWell's research has provided valuable and surprising (to me) information on this topic, and I look forward to future discoveries.

Crossposted from LessWrong