Considering its size and seriousness, the EA community lacks a great deal of information about itself, particularly with regards to the financial preferences of EA donors. This post proposes a novel[1] way to elicit marginal donor preferences and provide greater incentives for discovery of donor-relevant information without requiring factual validation.
Matching Credits
A matching credit is a financial instrument that provides $1 to a specified charity when the holder donates $1 of their own. The charity for the matching credit is specified on creation and cannot change. Matching credits are tradable and enter the market through an initial offering by their creator. I will assume the creator also operates the sole exchange on which credits trade. The creation of credits requires a subsidy on the part of the creator. This subsidy varies based on how much the credits can be initially sold for, but I believe this will always be at least slightly less than $1.
With these parameters, we can imagine a useful market in which prices reflect how much subsidy participants require to donate, and participants are rewarded for correctly anticipating price movements or uncovering information relevant to pricing.
For example, imagine a market with two participants, Alice and Bob, trading only Fivethought[2] matching credits. Fivethought is a new, small, and poorly understood charity. Alice believes donating to Fivethought is only half as valuable as her best option, so she values the credits at zero. Bob believes Fivethought is 80% as good as his best option and therefore values a $1 matching credit at $0.60[3]. If the market price settles below $0.60, Bob has an incentive to argue publicly for a higher valuation and thereby raise the price.
Several complications must be resolved for the market to function as described. First, participation would need to be restricted. For an EA funder, subsidizing credit creation and operating an exchange is only worthwhile if the market produces information about EA donors, which likely entails restricting participation to the EA community.
Second, limits are needed on how much capital each participant can deploy in the market. This is necessary for several reasons. In order for market outcomes to be reflective of EA funding at large and not the people most interested in the market, participants’ market budgets should be proportional to their historical EA donations[4]. Additionally, total participant capital must be lower than the total redeemable value of outstanding credits. Credits must also only be usable within the system. This ensures that participants are actually making decisions under scarcity, and that they cannot assume all credits will be redeemed.
Third, participants must face restrictions on when credits can be redeemed. The market is only informative if participants cannot immediately exit by redeeming credits. In order to prevent this, redemptions should be restricted to one month in the year, likely December. Relatedly, credits should either never expire or have very long durations. This would be more expensive for the credit provider but would allow participants to express a wider range of views about timing.
Because no market of this specific type has existed, several design parameters remain uncertain. They include the following:
- What is the appropriate ratio between required donation and matching value? I’m assuming a 1:1 ratio, but this allows only a limited range of expressed views. (prices cannot express more than a 2× difference between the best and worst options). A 3:1 match ratio could express up to a 4× difference but would also reduce the amount of capital circulating in the market, potentially weakening incentives.
- Is it feasible or desirable to issue credits with different match ratios?
- Should holding credits rather than using them be subsidized through yield? How would such a yield mechanism operate?
- To what degree is it reasonable to subsidize price discovery for donation opportunities the community values much less?
- Should participants be allowed to withdraw funds from the market? This would create stronger incentives for some participants, but also dramatically increase the incentives for bad behavior.
- How open can market participation be?
Beyond the implementation details of the market, there are broader questions about its viability and usefulness.
- Is this a better way of generating information about collective preferences than alternatives like quadratic budgeting?
As a piece of information, how valuable is the marginal participant’s willingness to donate[5] to a specific charity? Would such a market primarily help low-information donors such as myself?
- The market could incentivize new research and public dissemination of information, but it is unclear how cost-effective this would be relative to alternatives such as funding research directly or running essay contests. Would a market complement these approaches?
- Would the proposed market be tax-efficient and legal under US law?
- Would there be sufficient activity for the system to function as a market rather than a sequence of bilateral negotiations?
While I think this idea is very intellectually interesting, I’m not sure it has much merit as a practical solution. There are likely cheaper ways to generate socially valuable information about funding decisions in EA, such as surveys or contests. I think better aggregation of already public information could also be cost effective and useful, and I hope to write a separate post about this.
- ^
To my knowledge.
- ^
25% better than Forethought
- ^
The value of a credit is 2 x (relative effectiveness to best option in %) - 1. Below 50% relative effectiveness to best option, the credits have a value of $0.
- ^
This is clearly inegalitarian but is necessary so that the market can reflect the EA funding landscape, which is also inegalitarian.
- ^
Note that with the conditions already specified, prices are also downstream of the marginal buyer’s speculative belief in the future value of the credit. I think this is a positive, but it does complicate things.