Thanks so much for going on the podcast and for the incredible work that you and everyone involved in r.i.c.e. are doing! It’s amazing that you’ve been able to save so many lives so cost-effectively. I did have a few questions I was curious about though:
In discussing your room for more funding, you mentioned likely cost-effectiveness improvements from economies of scale as the program grows. That makes complete sense to me on the current margin, but I’m curious if you have thoughts about how the cost-effectiveness might change if the program were to scale up say an order of magnitude. It seems like both r.i.c.e.‘s staff and the medical college staff are exceptionally engaged and motivated. Do you sense of how important that’s been to the program’s success, and of whether that would be hard to maintain and replicate if the program expanded substantially more broadly? I know that for example in early childhood education interventions, there’s quite a bit of evidence of negative scale-up effects where programs that look really great in pilots have ended up looking a lot less good when implemented at large scale. (This isn’t at all intended to be a negative on r.i.c.e.’s current or near-term work, I’m just curious how optimistic or non-optimistic to be about the larger potential of kangaroo mother care programs)
In talking about why this program wasn’t done before r.i.c.e.‘s intervention and why it isn’t done in more places, you said “There are just a few paediatricians who work there, and they had just a few nurses working with them. And that’s just not going to be enough for this many babies. The government of Uttar Pradesh recognises this; the Department of Medical Education is working hard to open more new medical colleges and new nursing colleges.” Given that context, do you worry that part of what r.i.c.e is doing in the short-run is reallocating medical care from being provided to other people to being provided to the babies in the program? I of course don’t think this is an argument for r.i.c.e. being ineffective—I think there are strong reasons to think that the care you’re providing is doing more good than the counterfactual use of the nurses’ time, and that in the long-run the supply of nurses is relatively elastic. But I’m curious if this might be a reason to think the program is less cost-effective than it initially appears.
GiveWell’s moral weights take a time-relative interest account-like approach to the value of a human life, in which saving the life of a newborn infant is weighted as moderately less valuable than saving the life of an older child. This leads to GiveWell estimating r.i.c.e.’s cost-effectiveness as about 30% lower than it would have been if they had weighted preventing infant deaths as equally valuable as preventing the deaths of older children. Do you personally have any thoughts about the right approach to making these kinds of difficult tradeoffs when evaluating different charities? And how important a consideration do you think this moral weight question is for donors who are considering donating to r.i.c.e. vs. GiveWell’s current top charities.
My sense is that expanding the program at this site (or keeping it alive and well for more years at this site) has increasing returns, because we spread the administrative costs over more babies. In fact, knowing we have the funding runway to keep the program healthy lets us hire higher-quality staff with multi-year commitments. But expanding to another district would have huge fixed costs, even if the marginal cost is identical once it is up and running. We would still have a lot of our learning-by-doing, and we would have the paperwork, software, and protocols that we have developed, but we would fundamentally need some new relationships and a new entrepreneurial leader to captain that ship. We don’t currently have that person, but there is no in-principle reason that they could not be found and hired someday. More broadly, running this program has caused us to realize that cost-effectiveness in EA, philanthropy, and development economics has not paid enough attention to what microeconomic theory knows about fixed, variable, marginal, and average costs all being different. It’s on the to-do list to write about that someday.
There are three margins of behavior: The families, the government-salary doctors, and the nurses who are privately hired by the program. The families would counterfactually be bringing most babies home to poor odds. We believe the doctors are working harder and doing more, because the returns to their efforts are improved by the collaboration with the nurses and families. So what about the nurses? Government nurse jobs, for now, remain very hard to get (it’s a bad equilibrium where there both are not enough nurses and not enough public facilities to hire them). So these nurses would likely work somewhere in the private sector. Who knows the tiny general equilibrium effect on statewide nurse wages (!) but the quality of healthcare and neonatal survival for babies born in private facilities is worse, on average, than public facilities in this context, so on the margin shifting activity from private to public facilities would be likely to save lives.
If I understand the question, my own view is the opposite: all else equal, saving a neonate is 60 person-months better than saving a five year old. But I personally suspect this isn’t a huge practical deal: The uncertainties and variance in cost-effectiveness are plausibly much larger than one-part-in-three.
Thanks so much for your response, that all makes sense!
You’re understanding question 3 correctly—GiveWell’s moral weights look like the following, which is fairly different from valuing every year of life equally.
Thanks so much for going on the podcast and for the incredible work that you and everyone involved in r.i.c.e. are doing! It’s amazing that you’ve been able to save so many lives so cost-effectively. I did have a few questions I was curious about though:
In discussing your room for more funding, you mentioned likely cost-effectiveness improvements from economies of scale as the program grows. That makes complete sense to me on the current margin, but I’m curious if you have thoughts about how the cost-effectiveness might change if the program were to scale up say an order of magnitude. It seems like both r.i.c.e.‘s staff and the medical college staff are exceptionally engaged and motivated. Do you sense of how important that’s been to the program’s success, and of whether that would be hard to maintain and replicate if the program expanded substantially more broadly? I know that for example in early childhood education interventions, there’s quite a bit of evidence of negative scale-up effects where programs that look really great in pilots have ended up looking a lot less good when implemented at large scale. (This isn’t at all intended to be a negative on r.i.c.e.’s current or near-term work, I’m just curious how optimistic or non-optimistic to be about the larger potential of kangaroo mother care programs)
In talking about why this program wasn’t done before r.i.c.e.‘s intervention and why it isn’t done in more places, you said “There are just a few paediatricians who work there, and they had just a few nurses working with them. And that’s just not going to be enough for this many babies. The government of Uttar Pradesh recognises this; the Department of Medical Education is working hard to open more new medical colleges and new nursing colleges.” Given that context, do you worry that part of what r.i.c.e is doing in the short-run is reallocating medical care from being provided to other people to being provided to the babies in the program? I of course don’t think this is an argument for r.i.c.e. being ineffective—I think there are strong reasons to think that the care you’re providing is doing more good than the counterfactual use of the nurses’ time, and that in the long-run the supply of nurses is relatively elastic. But I’m curious if this might be a reason to think the program is less cost-effective than it initially appears.
GiveWell’s moral weights take a time-relative interest account-like approach to the value of a human life, in which saving the life of a newborn infant is weighted as moderately less valuable than saving the life of an older child. This leads to GiveWell estimating r.i.c.e.’s cost-effectiveness as about 30% lower than it would have been if they had weighted preventing infant deaths as equally valuable as preventing the deaths of older children. Do you personally have any thoughts about the right approach to making these kinds of difficult tradeoffs when evaluating different charities? And how important a consideration do you think this moral weight question is for donors who are considering donating to r.i.c.e. vs. GiveWell’s current top charities.
Thanks so much for asking!
My sense is that expanding the program at this site (or keeping it alive and well for more years at this site) has increasing returns, because we spread the administrative costs over more babies. In fact, knowing we have the funding runway to keep the program healthy lets us hire higher-quality staff with multi-year commitments. But expanding to another district would have huge fixed costs, even if the marginal cost is identical once it is up and running. We would still have a lot of our learning-by-doing, and we would have the paperwork, software, and protocols that we have developed, but we would fundamentally need some new relationships and a new entrepreneurial leader to captain that ship. We don’t currently have that person, but there is no in-principle reason that they could not be found and hired someday. More broadly, running this program has caused us to realize that cost-effectiveness in EA, philanthropy, and development economics has not paid enough attention to what microeconomic theory knows about fixed, variable, marginal, and average costs all being different. It’s on the to-do list to write about that someday.
There are three margins of behavior: The families, the government-salary doctors, and the nurses who are privately hired by the program. The families would counterfactually be bringing most babies home to poor odds. We believe the doctors are working harder and doing more, because the returns to their efforts are improved by the collaboration with the nurses and families. So what about the nurses? Government nurse jobs, for now, remain very hard to get (it’s a bad equilibrium where there both are not enough nurses and not enough public facilities to hire them). So these nurses would likely work somewhere in the private sector. Who knows the tiny general equilibrium effect on statewide nurse wages (!) but the quality of healthcare and neonatal survival for babies born in private facilities is worse, on average, than public facilities in this context, so on the margin shifting activity from private to public facilities would be likely to save lives.
If I understand the question, my own view is the opposite: all else equal, saving a neonate is 60 person-months better than saving a five year old. But I personally suspect this isn’t a huge practical deal: The uncertainties and variance in cost-effectiveness are plausibly much larger than one-part-in-three.
Thanks so much for your response, that all makes sense!
You’re understanding question 3 correctly—GiveWell’s moral weights look like the following, which is fairly different from valuing every year of life equally.