Here’s a GiveWell blog post from 2009 that engages with this question.

The answer to this is probably that in most cases, the welfare realized by that counterfactual life and/or their children might not be huge, but on average still positive. This can be exemplified by the commonly used life satisfaction scale, which ranges from 0-10, and aims to capture all possible wellbeing states. Therefore it follows that the extreme lowest scores represent states worse than not existing at all. There have been several attempts to determine this “neutral point”, above which life is better lived than not. 

A survey conducted in Ghana and Kenya estimated the neutral point at 0.56 (IDinsight, 2019). Results from Brazil, China and the US suggest a neutral point of 25 on a 100-point scale (Referenced by HLI, 2022). Another in the UK suggested 2/10 (referenced in Krekel & Frijters, 2021). While people disagree where exactly to locate this neutral point, there is some agreement that it could be between 1.5-2. 

When looking at country-level average happiness levels (happiness is closely related to life satisfaction), the only country deceeding 2 is Afghanistan. So while there might be a case for that argument in Afghanistan, in most other countries there will be no problem in expectation. That said, there is of course variance in life satisfaction within countries, so there is still the possibility of edge cases where the intervention benefits a person whose life satisfaction is far below the country-average and deceeds the neutral point. Some within-region life satisfaction averages are provided here.