Losing a certificate

We strongly encourage you to think of losing a certificate as if it undid the positive impact of the associated project. You might well ask: why?

Selling us a certificate does have a negative impact: it causes us to have less money. As a consequentialist, you might try to do some counterfactual calculation to decide whether that’s a good idea.

Under the certificates system, you are advised to forego the complicated calculation, and instead assume that our alternative use of the money would be about as good as the activity represented by the certificate. Under that assumption, the negative impact of selling the certificate is equivalent to the impact of undoing the activity represented by the certificate.

At the same time, you should ignore the negative consequences of us losing money—that would be double-counting the costs.


We (the certificate purchases) are indifferent between spending $X to acquire a certificate for Y, and spending $X to actually accomplish Y. So if we are willing to spend only $X purchasing a certificate for Y, that means that we think that our next-best-use for that $X is about as impactful as Y itself.

Of course, this only means that the we think that these two uses of $X are equally good—you may not be indifferent (either because you have different values, or different beliefs). By performing a more complex analysis, you can advance your values at our expense. In the simplest case, if you are convinced that you will make better use of the funds than we would, you should simply be willing to accept smaller amounts of money.

Whether it’s worth going along with the certificate system depends on how aligned the values of different buyers and sellers are, how the certificate-seller feels about being a good citizen, to what extent they accept decision-theoretic arguments for altruism, and how expensive the consequentialist analysis is / how useful the price signals provided by certificates are. It also depends on the benefits of information-sharing and collaboration, and the extent to which zero-sum competition interferes with collaboration.

In a community of donors and charities with closely shared values, the consequentialist analysis is very difficult and the certificate system will promote efficient outcomes roughly as well as correct consequentialist reasoning. In a community with widely varying values, purchasing certificates may leave some value from moral trade on the table, but the same problem afflicts conventional grantmaking. Whether we buy certificates or make grants, we need some orthogonal mechanism to capture all of the gains of moral trade. A sufficiently clever certificate-seller might do that by deviating from the certificate system, but that’s probably not going to be the most common motivation.


The economic rationale of this decision-making guideline is to ensure that the equilibrium price of a certificate is simultaneously equal to:

  1. The marginal cost of achieving the associated impact.
  2. The marginal donor’s willingness to pay for that impact (on the margin).

Property (1) ensures that cost-effective interventions are profitable. Property (2) ensures that other interventions can be replaced with a certificate, benefiting the donor and freeing up resources for more effective interventions. This is the normal mechanism by which markets promote allocative efficiency.

The economic analysis also applies when only a few people adopt the certificate system. The major caveat is that early adopters (and in particular funders) of the certificate system will subsidize non-adopters, as certificates flow to people who value them and money flows to people who don’t. We will make a (modest) effort to purchase certificates from people who seem to be taking the loss seriously, in order to reduce the size of this subsidy.

But seriously, shouldn’t a consequentialist always take the money?

In some sense, a consequentialist always prefers that they have money than that someone else have money, no matter how nice they are—after all, they can just choose the very best thing to do with it. The benefits of getting money always outweigh the costs of someone else losing money.

So suppose that Alice is buying a certificate from Bob. She has offered a price that is too low: Bob wouldn’t be willing to undo the associated project in exchange for Alice’s offer. But as a good consequentialist, Bob figures he might as well take the money anyway.

From Bob’s perspective, this is a strict improvement. After all, if Alice would have done something awesome with that money, then Bob can just do it instead.

In order to actually implement this strategy, Bob needs to think about all of the altruistic options that Alice is considering. And if Alice was considering giving money to Charlie, Bob will need to consider all of the altruistic options that Charlie was considering. And so on.

Local optimization under prices is a distributed algorithm for efficiently allocating resources. As a society, we haven’t really found any serious alternatives except in very simple cases.

So in some sense Bob should always take the money. But if Alice’s offer was too low, that’s an indication that maybe Bob should give the money back immediately, since Alice has something better to do with it. And at that point, he should probably just have declined the offer to start with.

(In reality Bob often will have strong reasons to take the money, so that he can get his way instead of Alice getting her way, but these usually amount to zero-sum conflict with Alice.)

Selfish motives

A selfish user may not care at all about the impact associated with a certificate. For example, perhaps they had a positive impact incidentally as part of a self-interested project. In that case, they should not mind selling the certificate at all.

If there are multiple buyers of certificates, however, even a selfish seller will hold out for a reasonable price—since by selling a certificate today, they lose the ability to sell it tomorrow.

The same argument applies if there may be buyers in the future, except that the reservation price is discounted by the probability that there is another opportunity to sell.

In any of these cases, the certificate-purchaser is subsidizing people who achieved a positive impact but don’t care about it as much as the certificate-purchaser (or who don’t respect the certificate system). This subsidy must be compared to the positive incentive effects, leading people who don’t care about do-gooding to do more good. On balance, we think it’s probably a good trade: the existence of people who don’t care as much about having a humanitarian impact probably makes the certificate system a better idea, not a worse one.