This page outlines a number of important considerations that Fund management teams take into account when recommending grants. We think it’s useful for donors to have a sense of some of the conceptual models that inform how we think about effective grantmaking.
These models are tools that are used as part of an overall approach when making judgements about which projects to fund. For example, grantmakers may not sit down and conduct explicit expected value calculations for every potential grant. But having the concept of expected value helps to clarify thinking around when it’s a good idea to take a risk, and when it’s better to play it safe. This page also doesn’t attempt to document every conceptual model exhaustively, but outlines some of the major considerations.
Much of our general approach towards grantmaking has been heavily influenced by GiveWell and the Open Philanthropy Project, two grantmaking organizations with significant experience and a strong track record of making highly effective grants.
Expected Value
Expected value is a way of comparing different types of grants with different risk profiles.
There are many possible projects that each EA Fund could recommend grants to, all of which have different risk profiles. But how do you compare something that’s high-risk / high-reward with something that’s a safer bet? Expected value is a method of comparing between projects with different outcomes. Essentially, you’re multiplying how good it would be if your plan works out by the probability that it actually does.
As a simplified example, let’s say someone offered you the following two bets:
- A 1% chance of winning $100,000
- A 50% chance of winning $2,000
While our intuitive reactions to these bets might be different, both bets have exactly the same expected value, of $1,000. If you played each game enough times, you’d expect to win the same amount of money in the long run.
In the same way, EA Funds grantmakers try to look at both of these terms – risk and reward. Some grants are very likely to work out. This can lead to some grants that seem strange or unlikely to work. However, if the potential payoff is big enough, it can be worth taking these chances.
As with any conceptual model, explicit expected value estimates should not be made too literally, or without accounting for other factors, but they do provide a useful heuristic for comparison.
Marginal Impact and Room for More Funding
Many projects could easily find a use for an extra dollar, but wouldn’t be able to cost-effectively deploy an extra ten million dollars (let alone double or triple their current budget). In economics this is known as ‘diminishing marginal returns’, and in grantmaking this is sometimes referred to as a project’s ‘room for more funding’.
There are many reasons why projects may hit these bottlenecks – for example an organization may be too capacity-constrained to scale up quickly enough, or perhaps the beneficiaries of a project become progressively harder to reach because they live further away from urban centres. Even a project that is doing extremely high-impact, cost-effective work, will eventually hit these constraints.
Grantmakers aim to think about how a project could productively use additional money, not just how effectively they’re currently using their existing funds.
Counterfactual Impact
A counterfactual is a thought experiment where you imagine how a situation could have turned out if you had acted differently (or not acted at all). With a fixed amount of money, grantmakers must make difficult tradeoffs between potentially high-value grant opportunities. It’s important that grantmakers consider not just how good their grants are in the abstract, but how good they are compared to other alternatives.
One way that we should think counterfactually is to consider opportunity cost. The opportunity cost of a grant is the value of the best available alternative use of the grant money. Where possible, grantmakers should consider alternative uses of the money they’re disbursing, and if one of those alternatives is higher-impact, they should consider switching to it instead.
It’s also important to consider what would happen in cases where grantmakers chose not to make a particular grant. For example, there is the question of replaceability – how likely is it that someone else will fund the project anyway? If the project was already likely to hit its funding goals, and you’re just displacing someone else’s money, then you haven’t had much counterfactual impact. In these cases it may be preferable to let the other party make the grant, and then look for alternative opportunities.
Track Record
Grantmakers need to make informed decisions about potential grant recipients, but are often operating on incomplete information. Sometimes a project has never been tried before, or not been tried in a particular context. Sometimes a project seems promising, but the grantmakers don’t know much about the team who plan to deliver it.
Grantmakers try to integrate information from a range of sources about the credentials of a project. Projects that are run by established teams with a history of successful execution on similar projects, and who are viewed positively by relevant experts will probably have a higher chance of success.
Information Value
Sometimes grantmakers will not expect a particular grant to have a large direct impact by itself, but that it will lead to more information that will be useful for future grantmaking. For example, a project could test out a new approach and learn valuable lessons that future projects could learn from and thus become more effective. A small seed grant to an as-yet unproven organization could enable to project to run for long enough to give the grantmakers a better sense of whether it makes sense to make further investments.
Direct and Indirect Effects
Grantmakers should be trying to take into account both the direct effects of a grant, as well as indirect effects. Sometimes most of the impact of a grant can come from the indirect effects. Sometimes the direct effects of a grant look positive, but the negative indirect effects may outweigh them (in which case, it’s likely that the grant shouldn’t be made at all).