12 ms·

Disappointing Findings on Anti-Poverty Cash Transfers in the USA

101 points by gwern 9y ago

jf 9y ago
A Conditional Cash Transfer program called "Opportunity NYC" is mentioned in this article. Given what I read in the article, I expected the amount of money in question to be on the order of hundreds or thousands a month, but according to Wikipedia, the only gave parents "$40 to $100 a month" The cash payments go to the family, almost always the mother or other female head of the household. Parents can receive from $40 to $100 a month if they keep up with responsibilities such as taking their children to the doctor or keeping them in school. https://en.wikipedia.org/wiki/Opportunity_NYC https://en.wikipedia.org/wiki/Opportunity_NYC
- wutbrodo 9y ago
  Opportunity NYC was the previous program. The current one is called Family Rewards 2.0 > On average, the program cost $13,459 per family, 48 percent of which was paid directly to families as cash rewards. This was over a three year period, which means that each family received 13459*.48/3 = $2150, or about $180/mo. So they provided between 2-5x as much money. I share your confusion though, as this is still categorically a lot smaller than I would've expected. I guess it's less of a UBI type program and focuses more on cash rewards for good health and educational outcomes. For a low income family, $2k/yr is non-trivial.
  - charlesdm 9y ago
    > 48 percent of which was paid directly to families as cash rewards. So where did the other 52% go to? Working costs of the program?
    - setgree 9y ago
      compliance and monitoring, most likely. The cash was conditioned on various attendance metrics, and the data collection process with such things is never as smooth as you'd like (I did a stint at a non-profit that ran experiments like this on behalf of economists).
      - r00fus 9y ago
        I wonder if it'd be more successful if that overhead were dropped to say 10-20% (via automation and simplification of qualifications)
        wutbrodo 9y ago
        Why would it? The study said that they found no significant effect, not that the return was poor relative to costs. It's possible that shifting the saved administrative costs into more aid would appreciably change the outcome, but it seems more likely that it's a structural problem with the program: ie, that CCTs are optimizing the wrong input.
        setgree 9y ago
        Regarding whether the effect would have been larger if the subjects got more cash -- in the original Progresa study, the ` size of the cash transfer is large, corresponding to about a 25 percent average increase in income of households living in extreme poverty' [0] (p.4). (Though in dollar amounts, if they are using the 'extreme poverty' metric of $1.25/day/person, we're talking about a lot less money in absolute terms). So there's something to this. But this is hard to square away with this newest study's findings of modest adverse impacts on labor participation rates. This, combined with null health effects, suggests that the CCTs are performing an income substitution effect, and that what people are doing with their newfound free-time is probably leisure rather than other forms of 'work' (i.e. exercising, cooking healthier meals, etc.) that policymakers might wish the subjects undertake. [0] https://www.povertyactionlab.org/sites/default/files/publications/449%20Progresa%20Health%20Gertler%202001.pdf https://www.povertyactionlab.org/sites/default/files/publica...
    - boomboomsubban 9y ago
      They also offered things to help people earn more rewards. The brief linked mentions short term tutoring, uniforms for your job, and transportation to interviews. Still seems like a lot of overhead though.
  - brudgers 9y ago
    $2k a year without strings attached is non-trivial. Tying the money to activities that may come with direct or indirect costs may make earning the money non-rational. At the granularity of paychecks, it's less than $45 a week. That's about four hours overtime at minimum wage. Or four hours of part time work at $11/hour. The value proposition of extra work is pretty straight forward. There's no bureaucratic process that might deny payment. There's no dependency on a health care provider or school filling out the paperwork correctly in a timely manner and correctly where "correctly" means to the satisfaction of bureaucratic process structured to say "no."
    - wutbrodo 9y ago
      > $2k a year without strings attached is non-trivial. Tying the money to activities that may come with direct or indirect costs may make earning the money non-rational. At the granularity of paychecks, it's less than $45 a week. That's about four hours overtime at minimum wage. Or four hours of part time work at $11/hour. Sure, but we're talking about things that most people would agree is good anyway: kids doing well in school, being healthy, etc. The goal of the CCTs isn't to make people want to do things that they vehemently disagree with: it's to shift the incentive boundary so that they're on the right side of it for things they would like to do anyway.
    - evolve2k 9y ago
      I'm really surprised now that consideration of the 'quantum of funds' was not mentioned at all in the report, even as a footnote. On face value, I would argue that $45/week extra although appreciated is not significant. As a thought experiment it's in the scale of, 1hr week spend more time help kids with homework, 1hr spend additional time job searching, 1hr deal with program administrative/reporting requirements. If a family say rewards themselves with a night out once a week for a dinner or movies etc, these funds are quickly gone. I think the scale here implies an incentive based around addressing a lack of will e.g. "Well if they just had a little more encouragement they'd spend more time helping their kids with education and seeking employment and good health". At this scale it does not address what I'll call 'incentive based on providing significant additional means for positive risk taking behaviour and structural optimization'. I define structural behavior as when a family makes significant changes to their own setup towards have an effect; like moving closer to the school, hiring a tutor, pursuing course further studies part time (2-3 days a week). Although the research process seems robust I agree the low quantum of funds discounts the greater significance of the findings. TLDR; Summary of findings: 'When we gave poor familes enough weekly funds to go out for dinner once a week alongside a bunch of criteria to seek improvement in their life, it made little difference'
- aantix 9y ago
  I found that striking too. In this program they were provided "$13,459 per family, 48 percent of which was paid directly to families as cash rewards". While I'm sure it was nice to receive a _little_ extra money, that certainly isn't going to be life changing. It's more of a tease than anything.
  - setgree 9y ago
    Going with the previous calculation of $180/month ($2160/year), that's about a 9% annual increase for a family of 4 at the poverty line (I'm seeing stats of ~24/year). One of the main ways in which redistribution programs are justified is by pointing out that a marginal dollar goes further for a poor person than a rich one. So I would hardly call this a "little" money, relative to baseline. Having said that, one of the key advantages of unconditional cash transfers is that with much lower compliance/monitoring costs, a much higher percentage of all funds can go directly to the poor. GiveDirectly provides an 'imperfect estimate' of 81%, for the record https://www.givewell.org/charities/give-directly#CashRatios https://www.givewell.org/charities/give-directly#CashRatios
    - kevmo 9y ago
      $2k/year isn't going to change anyone's life in America very much. 9% of a tiny amount is also a tiny amount.
      - toasterlovin 9y ago
        You can accomplish a lot with $166/month: - Mostly cover the cost of operating a car - Feed 2+ people (white rice is $34/month for 2400 calories/day; chicken is $60/month, whole milk is $80/month; mix and match to taste) While $2k/year isn't a whole lot of money, it's marginal. If you're living at the edge of what you can afford, even a 9% bump can put you in a much more comfortable place.
- DanBC 9y ago
  It's not much money. But for poor people $100 a month is the difference between eating or not eating some days, or having the heating on or going cold, or buying some books for the kids to read at home, or paying the bills on time vs paying them late. Also, about books: I don't know if there's something like this in the US, but it seems like a good idea: https://literacytrust.org.uk/support-us/help-child-fall-love-reading-valentines-day/ https://literacytrust.org.uk/support-us/help-child-fall-love...
  - djsumdog 9y ago
    I mean..even then..not really. That's still not a lot of money at all. I grew up with people who live mouth to paycheck. A meaningful amount, like $400~$500 a month could make a much more considerable impact. $100/month means "Maybe we can take the kids to a movie one a month or have a nice dinner to help distract ourselves." $500/month could mean, "We can get all those extra school supplies or put our kids in extra school programs." Seeing if people would make those decisions would be an equally important study. These current programs see more designed to encourage people to use free programs with a small monitary incentive. I probably says more about those programs than the impact of the money itself.
    - Mindless2112 9y ago
      An expectation of "we'll use the money for important things after we treat ourselves" is exactly why people oppose these kind of programs. You can't just hand-wave that away.
      - iovrthoughtthis 9y ago
        It depends on your ability to conceive of a future without the suffering you're attempting to escape. If the money is not significant enough to meaningfully impact that assessment then a short term distraction is a rational way to spend the money. It's analogous but no the same as the prisoners dilemma with a known number of games.
- jmull 9y ago
  I think this is OK. As a test. A larger amount makes more sense in terms of being surer to have an impact. But it would make a widespread program based on those amounts so expensive that there’s probably no point in answering the question. Suppose it was found that giving poor families $20K annually did make a substantial difference? It probably wouldn’t matter because a large scale program based on that amount would probably never get funded. Here, the question is if we give poor families a small (but not insignificant) amount of cash could that make a diffeeence? I guess the idea is that it might provide just enough to enable at least some of them to do something they would really like to do anyway but can’t quite manage. Unfortunately, at least according to this, the answer is no. Perhaps we can think of it as a promising idea that didn’t pan out? IDK, it sure would be nice if we could find a relatively cheap and straightforward solution to wide-spread poverty and inequality. Not exactly surprising that we haven’t, but it still seems worth continuing to try.
  - pavel_lishin 9y ago
    If we're never going to fund an amount that would make a difference, why bother doing small scale tests?
    - jmull 9y ago
      What I mean is, without doing these kinds of tests we don’t know what level will actually make a difference.
cortesoft 9y ago
Man, this type of program sounds patronizing. "Here, let us rich people tell you poor people what you need to do so you can be successful like me" Making this conditional is basically saying, "You don't know what you need, and we can't trust you to do what is best for yourself" People know what they need. If we have money to give them, just give them the money.
- linkmotif 9y ago
  Huh? Let me get this straight: you expect people to accept patronage. But you then don’t want the patrons to be patronizing? How does that work? Generally speaking, aren’t poor people poor because either they don’t know how to be not poor or can’t execute being not poor? Yes, sometimes it’s impossible to execute being rich, like if there is a systematic bias against a certain cohort. But cases like that aside, isn’t it pretty obvious that poor people either cannot or do not know how to not be poor? Your perspective seems to suggest being poor just happens randomly to random people. But that’s not true. Being poor happens to people who fail to get and keep money from other people. Any time someone talks about patronizing poor people it’s like: yeah either we’re not going to talk about them being poor like that’s a bad thing, or we’re going to talk about it like it’s a bad thing and so maybe then we should figure out what they’re doing wrong when there are other people who are not poor. Pretending like everyone’s economic actions are equal is insensitive to people who get and keep money and patronizing to people who don’t. Not all economic actions are equally good. Setting money on fire is not a good economic action. It is offensive to the person whose labor created that value. Yet I see poor people doing all kinds of offensive economic actions with money given to them by the state. I see people using benefits cards to buy groceries that I would never buy because they are so expensive or unnecessary. If you don’t want to be judged, don’t take money from other people, because that is literally patronage, which is patronizing.
  - bryanlarsen 9y ago
    Being poor is buying a 2 pack of toilet paper for $2 rather than an 8 pack for $4 because even though the 8 pack is a much better deal, the extra $2 will buy the loaf of bread you also need.
btilly 9y ago
I think that the problem is simple. People respond poorly to long-term rewards. You need to set up a short-term reward payoff cycle. https://buildingpharmabrands.com/2013/05/27/the-ad-that-created-a-habit/ https://buildingpharmabrands.com/2013/05/27/the-ad-that-crea... is inspiring that way. It shows how people who could not be motivated to solve a well-known long-term problem, tooth decay, were successfully motivated to solve a short-term problem that they previously ignored - the fact that you wake up with film over your mouth that doesn't feel very good. Long-term problems without good short-term feedback loops rewarding the proper behavior are hard, and will probably always be hard for humans. So rather than increasing the reward for the long-term issue, we need to find good short-term reinforcements for good behavior. (Yeah, easier said than done...)
- rmuesi 9y ago
  Isn't that exactly how this program was operating? Short-term reward payoff: monthly cash; long-term reward: children graduate from school.
  - btilly 9y ago
    In the context of brain psychology, short-term needs to be seconds to minutes. For example try to train a dog with rewards that come an hour later. It won't work. In the example that I gave, you brush your teeth, your mouth feels better right away. That works. By contrast if you start working on your project now, next week you won't have a miserable all-nighter. That doesn't work. Just ask anyone in college if you doubt me. That is why a monthly feedback cycle did not prove to be successful motivation here.
    - _rpd 9y ago
      Gamify anti-poverty programs? I wonder though, isn't forgoing small short-term rewards for greater long-term rewards a key aspect of escaping poverty? How can you condition that behavior with short-term rewards?
      - 9y ago
OscarCunningham 9y ago
I'd say that giving money to people is good if it makes them happy at the time. If it improves outcomes even after it's taken away then that's even better, but it's not necessary for the success of the program.
ballenf 9y ago
Seems kind of intuitive that this type of program works well in countries with weaker social safety nets than countries with stronger ones. The risk-reward calculus is just very different. Not to mention that the $ amounts don't seem to have been adequately scaled to match cost of goods.
sharemywin 9y ago
On average, the program cost $13,459 per family, 48 percent of which was paid directly to families as cash rewards. So the best we can do is 48% efficiency?
- cuckcuckspruce 9y ago
  Just because only 48% was paid directly to the family does not mean that these were the only costs to the program. For example, the program requires regular health checkups, so to make those accessible it likely either paid the total cost or most of the cost for those checkups as it's unlikely that the people using this program would be able to pay for them completely on their own.
- JackFr 9y ago
  To me that was the most interesting part. With respect to the politics of implementing these programs, it's not simply the incentives created by the 48% that lead to a programs continued adoption, but the incentive created by the 52%.
- olympus 9y ago
  While it sounds low, this is almost the same efficiency that a temporary work program for homeless people in Denver had (which initially caused me to look into the low sounding efficiency number). Think of it like this: if all of the remaining 52% went to paying a NYC social worker's average salary of $73.9k (source below), then each social worker has to manage 10.5 families to pay their salary. That's managing ten families and making sure their kids are in school, they are going to doctor's visits, etc, each week. That sounds like a manageable workload. BUT WAIT! An employee actually costs 200% of their salary to employ in terms of office rent, benefits, and other incidentals (that 200% figure can go up and down depending on company, location, etc, but is a good enough number for this thought experiment). So in reality, each social worker needs to manage 21 families to have their employment costs paid for. And tracking 21 families weekly sounds like a reasonable full time fob to me. That's getting in touch with just a little over four families each day to make sure they are meeting their conditions for the money. Since social work like this usually requires a lot of driving, it will probably fill up their day. So really, when accounting for the fact that you have to pay the case managers a fair wage to live in NYC to administer the program, 48% sounds about right. At least that single number doesn't point to there being any obvious graft going on. https://www1.salary.com/NY/Social-Worker-MSW-Salary.html https://www1.salary.com/NY/Social-Worker-MSW-Salary.html
funkythingsss 9y ago
Why is this surprising? Most anti-poverty measures are ineffective. The "war on poverty" has not worked in a single metric. Affirmative action hasn't changed a thing. The only thing that can get you out of poverty is a good family structure, a culture built on performance (Chinese, Indian culture takes this waaaay to far, but it's a start).
- tootie 9y ago
  You'll need a lot of citation for that. Also, Affirmative Action is decidedly not an anti-poverty program. It's an anti-bias program. And the comparison to China and India is apples and oranges. Aside from the fact that both countries have hundreds of millions of people in poverty, their success has been in industrializing agrarian societies. America is pretty far past that point. Our levels of absolute poverty (penury) are already near zero.
  - yasp 9y ago
    Giving someone a 200+ point advantage on their SATs based solely on the color of their skin is an anti-bias program now?
    - tootie 9y ago
      That's not really how it works, but I also didn't comment on it's effectiveness, just that it's not anti-poverty. If you're rich and a minority you get the same advantage. The test for eligibility isn't means-based. Hence, not anti-poverty. The intent is to get more minorities into colleges and into jobs. You can argue whether or not that's fair, but it probably works pretty well as intended.
      - yasp 9y ago
        > Even With Affirmative Action, Blacks and Hispanics Are More Underrepresented at Top Colleges Than 35 Years Ago https://www.nytimes.com/interactive/2017/08/24/us/affirmative-action.html https://www.nytimes.com/interactive/2017/08/24/us/affirmativ... Maybe it works as intended as long as you're trying to maximize enrollment in "any colleges", not "best colleges".
jellicle 9y ago
So if you actually read what the program is here: Provide families with a average of $179/month if and only if they jump through a whole bunch of hoops, from doctor visits (which they have to pay for) to school attendance to working more hours (each of which must be documented by the program participants), and then although the program did reduce poverty (give people money, they're less poor), it didn't make them into upper class people magically. So it didn't achieve its goals. Probably the average participant ended up spending dozens of hours doing bureaucratic tasks each month to get that $179. This is neoliberalism in a nutshell. Meanwhile, worthy people like the Koch brothers can get a $1 billion to $1.4 billion (estimated ANNUAL benefit from most recent Trump tax cuts to them personally) donation straight from the Federal Treasury and have to do exactly nothing to show that they "deserve" it.
- saint_fiasco 9y ago
  The doctor visits were probably necessary to measure the health outcomes. In a real-life version of the program, maybe they wouldn't be necessary. It's still silly that the participants had to pay for them in the study.
MechEStudent 9y ago
Welfare is engineered to fail, and it fails. You can't build a system by American political committee and expect it to actually work. There are known good solutions for poverty that work reliably and consistently well. That isn't what the US welfare system is engineered for. It is engineered to put money in the hands of political donors, not resolve poverty. The solution to resolving poverty, much like the Buffet rule for politicians, would work overnight, but will never be implemented. Buffet says a law that says no standing politician is eligible for re-election in a year when minimum true GDP year-over-year growth for the last 2 years has been below 3% would work. He is right. A similar law, based on "theory of constraints" and directly extractable from the pages of "the goal" would work for poverty, but has (sadly sadly) the same political palatability as drinking a gallon of raw sewage. Dang this lost, broken, wrecked political system and the scoundrels who are in power and abuse it.
- valar_m 9y ago
  > There are known good solutions for poverty that work reliably and consistently well. Can you give some examples?
  - danieltillett 9y ago
    Science and technology. Has been the only thing that has gotten the world out of the default historical state of 99% poverty.
    - valar_m 9y ago
      Sounds like you're answering "What strategy may work to solve poverty?" or something close to that. OP said there are "known good solutions for poverty that work reliably and consistently well". I'm interested to hear what is known to work, not what might work.
      - danieltillett 9y ago
        No science and technology are the only things that has been shown to work. Since the industrial revolution the same playbook has been applied successfully in country after country. Look at China. It was not until they allowed science and technology to be used to its full capacity did they drag the vast majority of people out of grinding poverty. The parts of the world where poverty is still high are those where science and technology have not yet arrived.
ereli1 9y ago
It's interesting to see approaches that have been documented to fail in eliminating poverty in developing countries (https://chrisblattman.com/2013/06/27/why-cash-transfers-to-the-poor-are-not-the-next-big-thing/ https://chrisblattman.com/2013/06/27/why-cash-transfers-to-t...) failing in a developed country.
- wutbrodo 9y ago
  Most of the excitement around direct cash transfers' effectiveness in developing countries is _unconditional_ cash transfers. The incentive structure (and goals of the program) are quite different.
Leader2light 9y ago
The goal of all welfare should be temporary support. At the end of the day, its just other people paying for someone else to get free stuff.
- minikites 9y ago
  Your second sentence describes the entire concept of civilization.
- danieltillett 9y ago
  I would rephrase you last sentence as spending other people's money to give someone else free stuff.
- sdwa 9y ago
  Our societies are already so hopelessly rigged that it's hard to feel too upset about these rather tiny wealth transfers to people who need it, even if they didn't "earn" it.
- ereli1 9y ago
  welfare is only needed if the labor market is allowed to create jobs that don't provide viable living wages. If given a choice, people choose well-paid work of reliance on welfare. The problems do start when full-time employment doesn't prevent a family from being poor. This happens because wages were allowed to drop to a level that doesn't support decent human conditions. My guess that since addressing this angers people who would hope to see wages low so they can maintain their business that has grown to be dependant on cheap labor - it's much easier to talk about lump sum transfers (conditional or conditional) as those do not impact the cost of labor, unemployment or living standards.
olympus 9y ago
I suspect that the benefits of conditional cash are very non-linear, such that the benefits aren't seen until the amount of monthly cash outweighs the immediate opportunity cost of meeting the conditions. If you're offering to pay me $200 per month, but I have to keep my kid in school (meaning a big time commitment from the kid and parent), and we have to go to regular doctor's appointments, then I'm losing several hours during a typical workday over the course of a year. If I take time off to take my kids to the doctor and also get myself to the doctor, then my boss is going to get irritated. I might lose my job. If I have to go to a work education program in the evening, then I don't have time for a second job. So now you're making me choose between $200 per month and possibly losing my job and preventing me from having two jobs. That's not worth it. You know what is worth it? $3k per month. That makes it worth the risk and effort. But sadly, nobody has tried a real big money CCT program, because the good "small money" programs haven't been proven to work so nobody is willing to take the risk. But the problem is, you won't see any benefit until you get over the hump of the opportunity cost that people in poverty have.
- stevenwoo 9y ago
  I can't find the link right now but IIRC there was a study that counterintuitively showed the yearly payments program to Alaska residents (from petroleum extraction taxes) increased the likelihood of poorer people taking on more work.
  - neaden 9y ago
    Was it this one? https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3118343&utm_content=buffer0e9c8&utm_medium=social&utm_source=twitter.com&utm_campaign=buffer https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3118343&...
    - stevenwoo 9y ago
      Yes, though there was a news article this month about it and other studies that linked to all of them (they talked about different aspects) that I was searching for, it's a small impact.
setgree 9y ago
The comments on this thread drive home why pre-analysis plans[0] are awesome. It's easy to say, after the fact: 'the grants weren't large enough, the relationship between X and Y is conceptually non-linear, those aren't the right outcomes to look at,' etc. The point is whether folks would have offered these explanations in advance. [0] 'Promises and perils of pre-analysis plans', Ben Olken, https://economics.mit.edu/files/10654 https://economics.mit.edu/files/10654
SrslyJosh 9y ago
> The study found that the program did not produce the hoped-for effects on most key outcomes (g., child education, parental employment) during a two to four-year period. Measuring educational outcomes is a good idea... > a conditional cash transfer program for low-income families with at least one child entering ninth or 10th grade. ...but you are measuring the impact kids who likely have already spent over a decade in poverty?!? Why not measure younger children? Infancy and early childhood are the most critical developmental periods. > Over about a three-year period, the program provided each participating family with cash rewards that were contingent on meeting certain goals related to children’s school performance, the family’s use of preventive health care, and parents’ employment. So if you're not doing well enough, you don't get the money, which of course will help you meet the targets for the next month.... :-/ As others have said, the amount of assistance seems shockingly low for 1) the outcomes they seem to be looking for and 2) the massive dysfunction of American society.
- wutbrodo 9y ago
  > As others have said, the amount of assistance seems shockingly low for 1) the outcomes they seem to be looking for and 2) the massive dysfunction of American society. I think people are misunderstanding what they're actually trying to test here. CCTs aren't simply assistance, or UBI by another name. The intent of the study was to test the effect on incentives. The FPL for a family of 4 is 24600, or 2500/mo. An extra 10+% in disposable income at that level of poverty isn't a transformative amount of aid, but it's plausibly significant enough to shift incentives (and perhaps ability to meet the bar), which is the point of the study. As it turns out, it _doesn't_ appear to do that, but that more likely indicates structural problems with the hypothesis than the fact that the study should've started with higher amounts of aid.
ams6110 9y ago
Why are these findings disappointing? That suggests a presupposition of the outcome. The results are what they are. We learned this doesn't have the desired result, so try something else.

101 points by gwern 9y ago

jf 9y ago

A Conditional Cash Transfer program called "Opportunity NYC" is mentioned in this article. Given what I read in the article, I expected the amount of money in question to be on the order of hundreds or thousands a month, but according to Wikipedia, the only gave parents "$40 to $100 a month" The cash payments go to the family, almost always the mother or other female head of the household. Parents can receive from $40 to $100 a month if they keep up with responsibilities such as taking their children to the doctor or keeping them in school. https://en.wikipedia.org/wiki/Opportunity_NYC https://en.wikipedia.org/wiki/Opportunity_NYC

wutbrodo 9y ago
Opportunity NYC was the previous program. The current one is called Family Rewards 2.0 > On average, the program cost $13,459 per family, 48 percent of which was paid directly to families as cash rewards. This was over a three year period, which means that each family received 13459*.48/3 = $2150, or about $180/mo. So they provided between 2-5x as much money. I share your confusion though, as this is still categorically a lot smaller than I would've expected. I guess it's less of a UBI type program and focuses more on cash rewards for good health and educational outcomes. For a low income family, $2k/yr is non-trivial.
- charlesdm 9y ago
  > 48 percent of which was paid directly to families as cash rewards. So where did the other 52% go to? Working costs of the program?
  - setgree 9y ago
    compliance and monitoring, most likely. The cash was conditioned on various attendance metrics, and the data collection process with such things is never as smooth as you'd like (I did a stint at a non-profit that ran experiments like this on behalf of economists).
    - r00fus 9y ago
      I wonder if it'd be more successful if that overhead were dropped to say 10-20% (via automation and simplification of qualifications)
      - wutbrodo 9y ago
        Why would it? The study said that they found no significant effect, not that the return was poor relative to costs. It's possible that shifting the saved administrative costs into more aid would appreciably change the outcome, but it seems more likely that it's a structural problem with the program: ie, that CCTs are optimizing the wrong input.
      - setgree 9y ago
        Regarding whether the effect would have been larger if the subjects got more cash -- in the original Progresa study, the ` size of the cash transfer is large, corresponding to about a 25 percent average increase in income of households living in extreme poverty' [0] (p.4). (Though in dollar amounts, if they are using the 'extreme poverty' metric of $1.25/day/person, we're talking about a lot less money in absolute terms). So there's something to this. But this is hard to square away with this newest study's findings of modest adverse impacts on labor participation rates. This, combined with null health effects, suggests that the CCTs are performing an income substitution effect, and that what people are doing with their newfound free-time is probably leisure rather than other forms of 'work' (i.e. exercising, cooking healthier meals, etc.) that policymakers might wish the subjects undertake. [0] https://www.povertyactionlab.org/sites/default/files/publications/449%20Progresa%20Health%20Gertler%202001.pdf https://www.povertyactionlab.org/sites/default/files/publica...
  - boomboomsubban 9y ago
    They also offered things to help people earn more rewards. The brief linked mentions short term tutoring, uniforms for your job, and transportation to interviews. Still seems like a lot of overhead though.
- brudgers 9y ago
  $2k a year without strings attached is non-trivial. Tying the money to activities that may come with direct or indirect costs may make earning the money non-rational. At the granularity of paychecks, it's less than $45 a week. That's about four hours overtime at minimum wage. Or four hours of part time work at $11/hour. The value proposition of extra work is pretty straight forward. There's no bureaucratic process that might deny payment. There's no dependency on a health care provider or school filling out the paperwork correctly in a timely manner and correctly where "correctly" means to the satisfaction of bureaucratic process structured to say "no."
  - wutbrodo 9y ago
    > $2k a year without strings attached is non-trivial. Tying the money to activities that may come with direct or indirect costs may make earning the money non-rational. At the granularity of paychecks, it's less than $45 a week. That's about four hours overtime at minimum wage. Or four hours of part time work at $11/hour. Sure, but we're talking about things that most people would agree is good anyway: kids doing well in school, being healthy, etc. The goal of the CCTs isn't to make people want to do things that they vehemently disagree with: it's to shift the incentive boundary so that they're on the right side of it for things they would like to do anyway.
  - evolve2k 9y ago
    I'm really surprised now that consideration of the 'quantum of funds' was not mentioned at all in the report, even as a footnote. On face value, I would argue that $45/week extra although appreciated is not significant. As a thought experiment it's in the scale of, 1hr week spend more time help kids with homework, 1hr spend additional time job searching, 1hr deal with program administrative/reporting requirements. If a family say rewards themselves with a night out once a week for a dinner or movies etc, these funds are quickly gone. I think the scale here implies an incentive based around addressing a lack of will e.g. "Well if they just had a little more encouragement they'd spend more time helping their kids with education and seeking employment and good health". At this scale it does not address what I'll call 'incentive based on providing significant additional means for positive risk taking behaviour and structural optimization'. I define structural behavior as when a family makes significant changes to their own setup towards have an effect; like moving closer to the school, hiring a tutor, pursuing course further studies part time (2-3 days a week). Although the research process seems robust I agree the low quantum of funds discounts the greater significance of the findings. TLDR; Summary of findings: 'When we gave poor familes enough weekly funds to go out for dinner once a week alongside a bunch of criteria to seek improvement in their life, it made little difference'
aantix 9y ago
I found that striking too. In this program they were provided "$13,459 per family, 48 percent of which was paid directly to families as cash rewards". While I'm sure it was nice to receive a _little_ extra money, that certainly isn't going to be life changing. It's more of a tease than anything.
- setgree 9y ago
  Going with the previous calculation of $180/month ($2160/year), that's about a 9% annual increase for a family of 4 at the poverty line (I'm seeing stats of ~24/year). One of the main ways in which redistribution programs are justified is by pointing out that a marginal dollar goes further for a poor person than a rich one. So I would hardly call this a "little" money, relative to baseline. Having said that, one of the key advantages of unconditional cash transfers is that with much lower compliance/monitoring costs, a much higher percentage of all funds can go directly to the poor. GiveDirectly provides an 'imperfect estimate' of 81%, for the record https://www.givewell.org/charities/give-directly#CashRatios https://www.givewell.org/charities/give-directly#CashRatios
  - kevmo 9y ago
    $2k/year isn't going to change anyone's life in America very much. 9% of a tiny amount is also a tiny amount.
    - toasterlovin 9y ago
      You can accomplish a lot with $166/month: - Mostly cover the cost of operating a car - Feed 2+ people (white rice is $34/month for 2400 calories/day; chicken is $60/month, whole milk is $80/month; mix and match to taste) While $2k/year isn't a whole lot of money, it's marginal. If you're living at the edge of what you can afford, even a 9% bump can put you in a much more comfortable place.
DanBC 9y ago
It's not much money. But for poor people $100 a month is the difference between eating or not eating some days, or having the heating on or going cold, or buying some books for the kids to read at home, or paying the bills on time vs paying them late. Also, about books: I don't know if there's something like this in the US, but it seems like a good idea: https://literacytrust.org.uk/support-us/help-child-fall-love-reading-valentines-day/ https://literacytrust.org.uk/support-us/help-child-fall-love...
- djsumdog 9y ago
  I mean..even then..not really. That's still not a lot of money at all. I grew up with people who live mouth to paycheck. A meaningful amount, like $400~$500 a month could make a much more considerable impact. $100/month means "Maybe we can take the kids to a movie one a month or have a nice dinner to help distract ourselves." $500/month could mean, "We can get all those extra school supplies or put our kids in extra school programs." Seeing if people would make those decisions would be an equally important study. These current programs see more designed to encourage people to use free programs with a small monitary incentive. I probably says more about those programs than the impact of the money itself.
  - Mindless2112 9y ago
    An expectation of "we'll use the money for important things after we treat ourselves" is exactly why people oppose these kind of programs. You can't just hand-wave that away.
    - iovrthoughtthis 9y ago
      It depends on your ability to conceive of a future without the suffering you're attempting to escape. If the money is not significant enough to meaningfully impact that assessment then a short term distraction is a rational way to spend the money. It's analogous but no the same as the prisoners dilemma with a known number of games.
jmull 9y ago
I think this is OK. As a test. A larger amount makes more sense in terms of being surer to have an impact. But it would make a widespread program based on those amounts so expensive that there’s probably no point in answering the question. Suppose it was found that giving poor families $20K annually did make a substantial difference? It probably wouldn’t matter because a large scale program based on that amount would probably never get funded. Here, the question is if we give poor families a small (but not insignificant) amount of cash could that make a diffeeence? I guess the idea is that it might provide just enough to enable at least some of them to do something they would really like to do anyway but can’t quite manage. Unfortunately, at least according to this, the answer is no. Perhaps we can think of it as a promising idea that didn’t pan out? IDK, it sure would be nice if we could find a relatively cheap and straightforward solution to wide-spread poverty and inequality. Not exactly surprising that we haven’t, but it still seems worth continuing to try.
- pavel_lishin 9y ago
  If we're never going to fund an amount that would make a difference, why bother doing small scale tests?
  - jmull 9y ago
    What I mean is, without doing these kinds of tests we don’t know what level will actually make a difference.

cortesoft 9y ago

Man, this type of program sounds patronizing. "Here, let us rich people tell you poor people what you need to do so you can be successful like me" Making this conditional is basically saying, "You don't know what you need, and we can't trust you to do what is best for yourself" People know what they need. If we have money to give them, just give them the money.

linkmotif 9y ago
Huh? Let me get this straight: you expect people to accept patronage. But you then don’t want the patrons to be patronizing? How does that work? Generally speaking, aren’t poor people poor because either they don’t know how to be not poor or can’t execute being not poor? Yes, sometimes it’s impossible to execute being rich, like if there is a systematic bias against a certain cohort. But cases like that aside, isn’t it pretty obvious that poor people either cannot or do not know how to not be poor? Your perspective seems to suggest being poor just happens randomly to random people. But that’s not true. Being poor happens to people who fail to get and keep money from other people. Any time someone talks about patronizing poor people it’s like: yeah either we’re not going to talk about them being poor like that’s a bad thing, or we’re going to talk about it like it’s a bad thing and so maybe then we should figure out what they’re doing wrong when there are other people who are not poor. Pretending like everyone’s economic actions are equal is insensitive to people who get and keep money and patronizing to people who don’t. Not all economic actions are equally good. Setting money on fire is not a good economic action. It is offensive to the person whose labor created that value. Yet I see poor people doing all kinds of offensive economic actions with money given to them by the state. I see people using benefits cards to buy groceries that I would never buy because they are so expensive or unnecessary. If you don’t want to be judged, don’t take money from other people, because that is literally patronage, which is patronizing.
- bryanlarsen 9y ago
  Being poor is buying a 2 pack of toilet paper for $2 rather than an 8 pack for $4 because even though the 8 pack is a much better deal, the extra $2 will buy the loaf of bread you also need.

btilly 9y ago

I think that the problem is simple. People respond poorly to long-term rewards. You need to set up a short-term reward payoff cycle. https://buildingpharmabrands.com/2013/05/27/the-ad-that-created-a-habit/ https://buildingpharmabrands.com/2013/05/27/the-ad-that-crea... is inspiring that way. It shows how people who could not be motivated to solve a well-known long-term problem, tooth decay, were successfully motivated to solve a short-term problem that they previously ignored - the fact that you wake up with film over your mouth that doesn't feel very good. Long-term problems without good short-term feedback loops rewarding the proper behavior are hard, and will probably always be hard for humans. So rather than increasing the reward for the long-term issue, we need to find good short-term reinforcements for good behavior. (Yeah, easier said than done...)

rmuesi 9y ago
Isn't that exactly how this program was operating? Short-term reward payoff: monthly cash; long-term reward: children graduate from school.
- btilly 9y ago
  In the context of brain psychology, short-term needs to be seconds to minutes. For example try to train a dog with rewards that come an hour later. It won't work. In the example that I gave, you brush your teeth, your mouth feels better right away. That works. By contrast if you start working on your project now, next week you won't have a miserable all-nighter. That doesn't work. Just ask anyone in college if you doubt me. That is why a monthly feedback cycle did not prove to be successful motivation here.
  - _rpd 9y ago
    Gamify anti-poverty programs? I wonder though, isn't forgoing small short-term rewards for greater long-term rewards a key aspect of escaping poverty? How can you condition that behavior with short-term rewards?
    - 9y ago

OscarCunningham 9y ago

I'd say that giving money to people is good if it makes them happy at the time. If it improves outcomes even after it's taken away then that's even better, but it's not necessary for the success of the program.

ballenf 9y ago

Seems kind of intuitive that this type of program works well in countries with weaker social safety nets than countries with stronger ones. The risk-reward calculus is just very different. Not to mention that the $ amounts don't seem to have been adequately scaled to match cost of goods.

sharemywin 9y ago

On average, the program cost $13,459 per family, 48 percent of which was paid directly to families as cash rewards. So the best we can do is 48% efficiency?

cuckcuckspruce 9y ago
Just because only 48% was paid directly to the family does not mean that these were the only costs to the program. For example, the program requires regular health checkups, so to make those accessible it likely either paid the total cost or most of the cost for those checkups as it's unlikely that the people using this program would be able to pay for them completely on their own.
JackFr 9y ago
To me that was the most interesting part. With respect to the politics of implementing these programs, it's not simply the incentives created by the 48% that lead to a programs continued adoption, but the incentive created by the 52%.
olympus 9y ago
While it sounds low, this is almost the same efficiency that a temporary work program for homeless people in Denver had (which initially caused me to look into the low sounding efficiency number). Think of it like this: if all of the remaining 52% went to paying a NYC social worker's average salary of $73.9k (source below), then each social worker has to manage 10.5 families to pay their salary. That's managing ten families and making sure their kids are in school, they are going to doctor's visits, etc, each week. That sounds like a manageable workload. BUT WAIT! An employee actually costs 200% of their salary to employ in terms of office rent, benefits, and other incidentals (that 200% figure can go up and down depending on company, location, etc, but is a good enough number for this thought experiment). So in reality, each social worker needs to manage 21 families to have their employment costs paid for. And tracking 21 families weekly sounds like a reasonable full time fob to me. That's getting in touch with just a little over four families each day to make sure they are meeting their conditions for the money. Since social work like this usually requires a lot of driving, it will probably fill up their day. So really, when accounting for the fact that you have to pay the case managers a fair wage to live in NYC to administer the program, 48% sounds about right. At least that single number doesn't point to there being any obvious graft going on. https://www1.salary.com/NY/Social-Worker-MSW-Salary.html https://www1.salary.com/NY/Social-Worker-MSW-Salary.html

funkythingsss 9y ago

Why is this surprising? Most anti-poverty measures are ineffective. The "war on poverty" has not worked in a single metric. Affirmative action hasn't changed a thing. The only thing that can get you out of poverty is a good family structure, a culture built on performance (Chinese, Indian culture takes this waaaay to far, but it's a start).

tootie 9y ago
You'll need a lot of citation for that. Also, Affirmative Action is decidedly not an anti-poverty program. It's an anti-bias program. And the comparison to China and India is apples and oranges. Aside from the fact that both countries have hundreds of millions of people in poverty, their success has been in industrializing agrarian societies. America is pretty far past that point. Our levels of absolute poverty (penury) are already near zero.
- yasp 9y ago
  Giving someone a 200+ point advantage on their SATs based solely on the color of their skin is an anti-bias program now?
  - tootie 9y ago
    That's not really how it works, but I also didn't comment on it's effectiveness, just that it's not anti-poverty. If you're rich and a minority you get the same advantage. The test for eligibility isn't means-based. Hence, not anti-poverty. The intent is to get more minorities into colleges and into jobs. You can argue whether or not that's fair, but it probably works pretty well as intended.
    - yasp 9y ago
      > Even With Affirmative Action, Blacks and Hispanics Are More Underrepresented at Top Colleges Than 35 Years Ago https://www.nytimes.com/interactive/2017/08/24/us/affirmative-action.html https://www.nytimes.com/interactive/2017/08/24/us/affirmativ... Maybe it works as intended as long as you're trying to maximize enrollment in "any colleges", not "best colleges".

jellicle 9y ago

So if you actually read what the program is here: Provide families with a average of $179/month if and only if they jump through a whole bunch of hoops, from doctor visits (which they have to pay for) to school attendance to working more hours (each of which must be documented by the program participants), and then although the program did reduce poverty (give people money, they're less poor), it didn't make them into upper class people magically. So it didn't achieve its goals. Probably the average participant ended up spending dozens of hours doing bureaucratic tasks each month to get that $179. This is neoliberalism in a nutshell. Meanwhile, worthy people like the Koch brothers can get a $1 billion to $1.4 billion (estimated ANNUAL benefit from most recent Trump tax cuts to them personally) donation straight from the Federal Treasury and have to do exactly nothing to show that they "deserve" it.

saint_fiasco 9y ago
The doctor visits were probably necessary to measure the health outcomes. In a real-life version of the program, maybe they wouldn't be necessary. It's still silly that the participants had to pay for them in the study.

MechEStudent 9y ago

Welfare is engineered to fail, and it fails. You can't build a system by American political committee and expect it to actually work. There are known good solutions for poverty that work reliably and consistently well. That isn't what the US welfare system is engineered for. It is engineered to put money in the hands of political donors, not resolve poverty. The solution to resolving poverty, much like the Buffet rule for politicians, would work overnight, but will never be implemented. Buffet says a law that says no standing politician is eligible for re-election in a year when minimum true GDP year-over-year growth for the last 2 years has been below 3% would work. He is right. A similar law, based on "theory of constraints" and directly extractable from the pages of "the goal" would work for poverty, but has (sadly sadly) the same political palatability as drinking a gallon of raw sewage. Dang this lost, broken, wrecked political system and the scoundrels who are in power and abuse it.

valar_m 9y ago
> There are known good solutions for poverty that work reliably and consistently well. Can you give some examples?
- danieltillett 9y ago
  Science and technology. Has been the only thing that has gotten the world out of the default historical state of 99% poverty.
  - valar_m 9y ago
    Sounds like you're answering "What strategy may work to solve poverty?" or something close to that. OP said there are "known good solutions for poverty that work reliably and consistently well". I'm interested to hear what is known to work, not what might work.
    - danieltillett 9y ago
      No science and technology are the only things that has been shown to work. Since the industrial revolution the same playbook has been applied successfully in country after country. Look at China. It was not until they allowed science and technology to be used to its full capacity did they drag the vast majority of people out of grinding poverty. The parts of the world where poverty is still high are those where science and technology have not yet arrived.

ereli1 9y ago

It's interesting to see approaches that have been documented to fail in eliminating poverty in developing countries (https://chrisblattman.com/2013/06/27/why-cash-transfers-to-the-poor-are-not-the-next-big-thing/ https://chrisblattman.com/2013/06/27/why-cash-transfers-to-t...) failing in a developed country.

wutbrodo 9y ago
Most of the excitement around direct cash transfers' effectiveness in developing countries is _unconditional_ cash transfers. The incentive structure (and goals of the program) are quite different.

Leader2light 9y ago

The goal of all welfare should be temporary support. At the end of the day, its just other people paying for someone else to get free stuff.

minikites 9y ago
Your second sentence describes the entire concept of civilization.
danieltillett 9y ago
I would rephrase you last sentence as spending other people's money to give someone else free stuff.
sdwa 9y ago
Our societies are already so hopelessly rigged that it's hard to feel too upset about these rather tiny wealth transfers to people who need it, even if they didn't "earn" it.
ereli1 9y ago
welfare is only needed if the labor market is allowed to create jobs that don't provide viable living wages. If given a choice, people choose well-paid work of reliance on welfare. The problems do start when full-time employment doesn't prevent a family from being poor. This happens because wages were allowed to drop to a level that doesn't support decent human conditions. My guess that since addressing this angers people who would hope to see wages low so they can maintain their business that has grown to be dependant on cheap labor - it's much easier to talk about lump sum transfers (conditional or conditional) as those do not impact the cost of labor, unemployment or living standards.

olympus 9y ago

I suspect that the benefits of conditional cash are very non-linear, such that the benefits aren't seen until the amount of monthly cash outweighs the immediate opportunity cost of meeting the conditions. If you're offering to pay me $200 per month, but I have to keep my kid in school (meaning a big time commitment from the kid and parent), and we have to go to regular doctor's appointments, then I'm losing several hours during a typical workday over the course of a year. If I take time off to take my kids to the doctor and also get myself to the doctor, then my boss is going to get irritated. I might lose my job. If I have to go to a work education program in the evening, then I don't have time for a second job. So now you're making me choose between $200 per month and possibly losing my job and preventing me from having two jobs. That's not worth it. You know what is worth it? $3k per month. That makes it worth the risk and effort. But sadly, nobody has tried a real big money CCT program, because the good "small money" programs haven't been proven to work so nobody is willing to take the risk. But the problem is, you won't see any benefit until you get over the hump of the opportunity cost that people in poverty have.

stevenwoo 9y ago
I can't find the link right now but IIRC there was a study that counterintuitively showed the yearly payments program to Alaska residents (from petroleum extraction taxes) increased the likelihood of poorer people taking on more work.
- neaden 9y ago
  Was it this one? https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3118343&utm_content=buffer0e9c8&utm_medium=social&utm_source=twitter.com&utm_campaign=buffer https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3118343&...
  - stevenwoo 9y ago
    Yes, though there was a news article this month about it and other studies that linked to all of them (they talked about different aspects) that I was searching for, it's a small impact.

setgree 9y ago

The comments on this thread drive home why pre-analysis plans[0] are awesome. It's easy to say, after the fact: 'the grants weren't large enough, the relationship between X and Y is conceptually non-linear, those aren't the right outcomes to look at,' etc. The point is whether folks would have offered these explanations in advance. [0] 'Promises and perils of pre-analysis plans', Ben Olken, https://economics.mit.edu/files/10654 https://economics.mit.edu/files/10654

SrslyJosh 9y ago

> The study found that the program did not produce the hoped-for effects on most key outcomes (g., child education, parental employment) during a two to four-year period. Measuring educational outcomes is a good idea... > a conditional cash transfer program for low-income families with at least one child entering ninth or 10th grade. ...but you are measuring the impact kids who likely have already spent over a decade in poverty?!? Why not measure younger children? Infancy and early childhood are the most critical developmental periods. > Over about a three-year period, the program provided each participating family with cash rewards that were contingent on meeting certain goals related to children’s school performance, the family’s use of preventive health care, and parents’ employment. So if you're not doing well enough, you don't get the money, which of course will help you meet the targets for the next month.... :-/ As others have said, the amount of assistance seems shockingly low for 1) the outcomes they seem to be looking for and 2) the massive dysfunction of American society.

wutbrodo 9y ago
> As others have said, the amount of assistance seems shockingly low for 1) the outcomes they seem to be looking for and 2) the massive dysfunction of American society. I think people are misunderstanding what they're actually trying to test here. CCTs aren't simply assistance, or UBI by another name. The intent of the study was to test the effect on incentives. The FPL for a family of 4 is 24600, or 2500/mo. An extra 10+% in disposable income at that level of poverty isn't a transformative amount of aid, but it's plausibly significant enough to shift incentives (and perhaps ability to meet the bar), which is the point of the study. As it turns out, it _doesn't_ appear to do that, but that more likely indicates structural problems with the hypothesis than the fact that the study should've started with higher amounts of aid.

ams6110 9y ago

Why are these findings disappointing? That suggests a presupposition of the outcome. The results are what they are. We learned this doesn't have the desired result, so try something else.