Happy to see activity toward this. I agree that, for some targets at least, cash transfers are likely more effective than targeted services or programs. I don’t agree that this is true for all impoverished persons, and I don’t understand how to tell the difference at scale.
Part of my confusion/disagreement is that these tend to get framed as populations, rather than individuals, but the pilot programs are about a small number of individuals. There seems to be no recognition that individuals come and go from populations, and there’s a ton of overlap/leakage between a served and an unserved population for any given program. IMO, very small pilots are often successful not because of the underlying mechanisms, but because of observation and selection effects.
Thanks for the comment! I also do not think cash will be more effective for—every—impoverished person. I do think, however, that most are a lot closer to almost all with the only exceptions being people that are suffering from extreme drug addiction and/or mental illness. Actually, there’s some spotty evidence (we need a good RTC study on this) that cash transfers could actually be a cost-effective way to reduce drug usage. (check out Simon, a case study who was a 20-year homeless heroin addict).
individuals come and go from populations, and there’s a ton of overlap/leakage between a served and an unserved population for any given program
I 100% agree with your comment that we should make sure to look at the individuals as well as the larger data. I may be misunderstanding your point, but I think you might be misunderstanding how guaranteed income programs work. Some are household-based, but I vastly prefer individual-based (it means larger families get more support and individuals are empowered to, for example, escape abusive relationships).
Either way, the individuals or households in the population selected do not leak between served and unserved during the duration of their transfers. The whole point of guaranteed income is its consistency for every person in the program.
very small pilots are often successful not because of the underlying mechanisms, but because of observation and selection effects
Nonetheless, spending on drugs & alcohol went down by 39%, and the pilot was a massive success. It appears like larger, better pilots with fewer observation and selection biases don’t have worse results than the small ones.
Happy to see activity toward this. I agree that, for some targets at least, cash transfers are likely more effective than targeted services or programs. I don’t agree that this is true for all impoverished persons, and I don’t understand how to tell the difference at scale.
Part of my confusion/disagreement is that these tend to get framed as populations, rather than individuals, but the pilot programs are about a small number of individuals. There seems to be no recognition that individuals come and go from populations, and there’s a ton of overlap/leakage between a served and an unserved population for any given program. IMO, very small pilots are often successful not because of the underlying mechanisms, but because of observation and selection effects.
Thanks for the comment! I also do not think cash will be more effective for—every—impoverished person. I do think, however, that most are a lot closer to almost all with the only exceptions being people that are suffering from extreme drug addiction and/or mental illness. Actually, there’s some spotty evidence (we need a good RTC study on this) that cash transfers could actually be a cost-effective way to reduce drug usage. (check out Simon, a case study who was a 20-year homeless heroin addict).
I 100% agree with your comment that we should make sure to look at the individuals as well as the larger data. I may be misunderstanding your point, but I think you might be misunderstanding how guaranteed income programs work. Some are household-based, but I vastly prefer individual-based (it means larger families get more support and individuals are empowered to, for example, escape abusive relationships).
Either way, the individuals or households in the population selected do not leak between served and unserved during the duration of their transfers. The whole point of guaranteed income is its consistency for every person in the program.
This is an argument I see pretty frequently, and I agree that (specifically for homeless people) more research is needed. That said, in the largest GI Randomized Controlled Trial experiment for homeless people, the (115) pilot participants were not carefully overseen. “Participants completed questionnaires at 1 month and then every 3 months. To better understand individual experiences, participants also completed open-ended qualitative interviews after 6 and 12 months.”
Nonetheless, spending on drugs & alcohol went down by 39%, and the pilot was a massive success. It appears like larger, better pilots with fewer observation and selection biases don’t have worse results than the small ones.