EA - Deworming and decay: replicating GiveWell’s cost-effectiveness analysis by JoelMcGuire

Link to original articleWelcome to The Nonlinear Library, where we use Text-to-Speech software to convert the best writing from the Rationalist and EA communities into audio. This is: Deworming and decay: replicating GiveWell’s cost-effectiveness analysis, published by JoelMcGuire on July 25, 2022 on The Effective Altruism Forum. Joel McGuire, Samuel Dupret and Michael Plant Summary At the Happier Lives Institute, we have been replicating GiveWell’s cost-effectiveness analysis of deworming. GiveWell’s model assumes that the economic benefits of deworming last for 40 years with no decline over time. We noticed that this assumption conflicts with the data from the main deworming study that GiveWell uses (Hamory et al., 2021). When we looked at the effects over time, we found that the economic benefits decay by around 12% each year. Taking this decay into account shrinks the total economic benefit by 60% compared to GiveWell’s model. This would reduce the cost-effectiveness of nearly all their recommended deworming charities below their 10x bar for funding. For more details about our calculations, the data used, the choice of the decay model, and its implications for the GiveWell deworming charities, see our external appendix. We raised this issue with GiveWell. They said that they are aware that the data suggests a decay over time, but they think their model is more appropriate; they noted that they haven’t explained and defended this choice publicly. In conversation, they gave two reasons for their approach. First, they have strong prior beliefs about the effects over time. Second, they have already applied a conservative discount to deworming. In this post, we provide the context for our analysis and why we estimate a decay of economic benefits. We then attempt to reconstruct GiveWell’s reasons and explain why we find these reasons unconvincing. Finally, we provide four recommendations to improve the clarity and transparency of GiveWell’s cost-effectiveness analyses. These are to (1) publicly explain and defend their assumptions about the effect of deworming over time; (2) explain their cost-effectiveness analyses in writing; (3) illustrate the sensitivity of their results to key parameters; (4) make it clear when an estimate is subjective or evidence-based. 1. How we discovered the discrepancy between the data and GiveWell’s model The Happier Lives Institute is searching for the most cost-effective ways to increase happiness. We have started this major project by assessing GiveWell’s top interventions and charities in terms of ‘subjective wellbeing’ (e.g. self-reported happiness and life satisfaction). So far, we have conducted cost-effective analyses (CEAs) of cash transfers and psychotherapy. Now, we are focusing our attention on understanding and replicating GiveWell’s CEA of deworming. Later this year, we will publish a report comparing deworming to cash transfers and psychotherapy in terms of its effects on subjective wellbeing. In this report, however, we are solely concerned with understanding the economic benefits of deworming. Mass deworming is a large drug intervention targeting chronic parasitic infections such as schistosomiasis and soil-transmitted helminthiasis (GiveWell, 2018; WHO, 2006, 2011). These infections can cause a range of urinary, intestinal, nutritional, cognitive and developmental problems. The parasites are transmitted through soil and water because of a lack of proper sanitation. The drugs are cheap and unlikely to cause side effects so they are provided to individuals without testing for infection (which would otherwise be expensive; Hamory et al., 2021). GiveWell has reviewed deworming multiple times (2016, 2017, 2018). Mass deworming is successful in reducing worm and infection levels, but GiveWell (2018) concluded that the main impact of deworming is the potential for preventing adverse effects on children’s development - and thereby long-term consequences - rather than the effects on preventing short-term health ...