The views expressed here are my own, not those of my employers or reviewers of the draft.

I have collected Ambitious Impact’s (AIM’s) cost-effectiveness estimates of the animal welfare and global health and development interventions they recommended for Charity Entrepreneurship’s incubation programs until 2024. AIM’s original cost-effectiveness estimates of the animal welfare interventions are optimistic because they assume a welfare range conditional on sentience of 1, so I adjusted them based on Rethink Priorities’ median welfare ranges. In addition, I assumed an increase of 100 welfare points in humans (WPs) and 1 quality-adjusted life-year (QALY) are each equivalent to averting 1 disability-adjusted life-year^[1] (DALY), and aggregated the main estimates of each report using the mean.

The table below contains the mean, geometric mean, 5th percentile, median, and 95th percentile cost-effectiveness of the animal welfare and global health and development interventions. All the 5 stats suggest the interventions in animal welfare are more cost-effective than those in global health and development. In particular, in terms of the 5th percentile, which is the stat I considered arguably best proxying AIM’s marginal cost-effectiveness in each area, animal welfare is 48.7 times as cost-effective as global health and development.

Joey Savoie (AIM’s CEO) noted:

[AIM’s] welfare points are far less certain estimates when compared to our global health [and development] estimates. This matters a lot, e.g., I would regress weaker CEAs [cost-effectiveness analyses] by over 1 order of magnitude [making them less than 10 % as large] even from the same organization [AIM?] using similar methods, and it could be 3+ orders of magnitude across different orgs and methods. AIM in general is pretty confident e.g. that our best animal charities are not 379x better than a top GiveWell charity even if a first pass CEA might suggest that.

I understand cost-effectiveness estimates across different areas are not directly comparable, even if they are unbiased, because the methodologies differ. However, I believe the large regressions to a lower cost-effectiveness Joey mentioned only apply given a strong (low uncertainty) prior that animal welfare interventions are much less cost-effective^[2]. It is unclear to me why one would have such a strong prior, as cost-effectiveness is usually thought to vary significantly across cause areas. All in all, I would still say this post provides some evidence that the best interventions in animal welfare are much more cost-effective than the best in global health and development.

Thanks to Filip Murar, Morgan Fairless and Vicky Cox for feedback on the draft^[3].

1. ^{^}

    AIM estimated an annual welfare of a human in a high income country of 82 WPs. AIM’s cost-effectiveness estimates of the new animal welfare interventions they recommended for Charity Entrepreneurship’s 2025 incubation program are expressed in suffering-adjusted days (SADs) per $, not WP/$.

2. ^{^}

    Based on inverse-variance weighting, the regressed cost-effectiveness would be the mean between the prior and estimated cost-effectiveness weighted by the reciprocal of their variance. For the regression to be large, the prior cost-effectiveness has to be much lower than the estimated one, and the variance (uncertainty) of the prior cost-effectiveness much smaller than that of the estimated one.

3. ^{^}

    I listed the names alphabetically.