Denver gave homeless people $1k/mth. Year later, nearly half had housing
businessinsider.com
businessinsider.com
People who received $1k/month: 44% housed
People who received a $6500 lump sum followed by $500/month: 48% housed
The control group who received $50/month: 43% housed
E.g: If you have $500/month, you can afford to pay your buddy $200/month to crash on his couch. Now you have reliable access to a shower, an address to put on your resumes, a bit of cash to thrift a suit/dress clothes and a beater laptop, and it's way easier to get a job.
Most of the jobs in the world don't rely on Zoom interviews and email. You'll get a phone call and have to physically go interview with someone. Really hard to hide looking and smelling like a mess in that scenario.
* lockers. most places claim "24 hours only" but aren't so great about that, and many will let you rent lockers for 24/7 storage for trivial money.
* decent toilets, sinks, mirrors. it's hard to shower, shave, and shit in a gas station bathroom.
* you can wash clothes. gyms with a pool often have drying equipment (like the little centrifuges for bathing suits). also saunas and blowdryers. like if you only have 3 pairs of undies, you can give 2 of em a good wash and spin them mostly dry, then hang em in a locker.
* 24/7 access. hang out or workout at night, and then quietly grab a nap somewhere (mostly) secure. clean your gear, take a shower, get right, without a ton of people around. hit a couple of reps for arms and shoulders while you're at it.
* most gyms I've been to often have powerbars or other cheap meal replacements / proteins in vending machines as well, which may be your only option if you're down to < $5 a day for food
Hardly ideal, but it fit his goals.
I love how they credit themselves for “significant improvements in housing outcomes”, “an increase in financial stability and a greater ability to pay bills and reduced reliance on emergency assistance”, and “substantial cost savings in public spending and a large reduction in public service utilization” even for the $50/mo group.
Makes me wonder why they don’t have a Group D receiving nothing, or if they had one but decided not to publish results on that. (Yes, I know it’s hard to get people to participate if you give them nothing. Maybe delayed payment could work.)
I feel like it's possible to design a much better experiment than this. But my overarching concern is that I don't trust activist studies on homelessness issues, been burned too many times taking them at face value.
While the findings outlined in this report indicate overall improvement for DBIP participants across many targeted outcomes, minimal differences were observed when comparing outcomes across the different payment groups. This may suggest that consistent cash assistance of even $50 can improve many aspects of one’s life. Or, as we heard through member checking, it was the aggregate of the whole DIBP program that made a difference for individuals. This means the trust conveyed by delivering guaranteed income, attention by community based organizational partner staff, phones and phone plans, and connections.
I think this effect is seen a lot in various psychological and well-being studies, I'm not sure if it has a name. But basically, if someone demonstrates they are interested in your well-being and stays in touch with you, it can often have an outsized positive effect on one's life, especially when connection and support have been difficult to find.
Then there's the other group, people with severe mental illness, drug addiction, who aren't rational and are chronically homeless because they basically can't function in society. You could give them as much money as you want and it wouldn't make much of a difference. What they need is to be put in some kind of permanent care.
The effects on employment were much more pronounced than housing. The $1000/mo group went from 15% employment to 37%, and the $50/mo group had no change in employment at all.
It makes some sense: employment is probably more elastic to changes in financial status than housing. Just being able to afford transportation or a bit of occasional child care can immediately alter one's employment prospects.
We should target families and individuals who just lost a job and became homeless, looking for work and a place to live. Then go down the list from there.
other than that, agree. sure looks like no real effect. i wonder what % housed would be after a year for people who didn't even get $50/month? it's probably also the same. for the reasons you state.
”Logic” tells me a 20x monthly income increase would lead to more than a 1% increase in housed population.
6.5x more than _that_ only led to a 5% increase in housed population.
Sigh. Everything sucks, doesn’t it?
It’s just evidence. You wanted a clean solution and you got evidence that the thing you thought solved the problem doesn’t have the effect you’d like.
A) 1K/month (or the equivalent thereof) is not nearly enough to really make a difference in terms of increasing prospects to finding housing and the 3 groups could have just as well received $50/$51/$52 a month for all that it's worth. Or,
B) There's a roughly ~50/50 split of people who do not want to be homeless, who will do whatever they can do get themselves housed, and the other group that just don't care/unable/unwilling to find housing regardless of money, as the money will be used for other things in stead of finding housing.
some might have a cellphone, but why bother answering questions for someone who isn't going to help you?
Yes, the control group, who received less money, also did well in some metrics, but in most metrics they also started off better, so the delta in their experience is in fact smaller.
Furthermore, the control group also experienced a drop in full-time employment, while both of the subject groups experienced gains.
I agree the headline isn't a useful representation of the importance of the data, but it does seem like this was a fairly impactful program.
So it wasn't a control group at all, and the observed could be reversion to the mean.
From the executive summary of the research:
"All payment groups showed significant improvements in housing outcomes, including a remarkable increase in home rent and ownership, and decrease in nights spent unsheltered."
The actual quantitative data (1, appendix table 4) shows that there were massive leaps in the number of people housed. Less than 10% of people were renting or owning at the start of the study, and more than 40% at the end.
It didn't track any group that didn't receive payment so it doesn't tell us what the difference was, and it also doesn't give us much history as to how long they had been homeless. So that leaves open the question of how many of these people would have pulled themselves out of homelessness absent this intervention.
It definitely needs more rigorous study, but these results certainly point towards something promising, which is supported by other programs.
(1. https://static1.squarespace.com/static/64f507a995b636019ef88...)
2. They measured more than just housing attainment, and there are definitely patterns that suggest that the $1000 group had a higher impact.
3. The selection process was done through outreach agencies, who recruited from people already seeking help. That group seems likely to already be on a path away from homelessness, vs. those who do not seek external help.
Really, this is a clickbait article, written about a preliminary/inadequate study. The actual data is here: https://static1.squarespace.com/static/64f507a995b636019ef88...
You can sort of use an intervention group as a control group in clinical studies when, for example, comparing an old medication against a new one. This works because the original study of the old medication had a control group that you can infer as valid for comparison to the new medication.
You are correct that we can infer whether the amount of intervention matters, but there is no way to know the effect of those first $50. Diminishing marginal utility theory from economics tells us that the first $50 is going to be those with the greatest effect, so this study is missing a massive piece of the puzzle.
Housing attainment is just one thing they were measuring.
$50 is not a placebo or a control. Again, those have actual definitions when it comes to science, even social science. There is a damn good reason that the actual paper never called it a control group, or tried to imply it. $50 is real. For a homeless person, $50 cash in your pocket will provide an actual, not perceived effect.
I'm not saying that $50/mo can't act as an effective control. I'm saying that there is no evidence available to reach that conclusion. There are economic theories that suggest that the first $50 is more important than subsequent amounts, so it would be important to prove or disprove that in the real world.
There is also the mental effect as you pointed out, but if you read the study you would know that all of the people recruited for this study were recruited through relationships at support agencies. Meaning that they were already in touch with agencies that gave the feeling of people caring about them.
What the study shows is that the extra $950 basically makes no difference in housing attainment. That should be the headline conclusion.
$50/mo is like a control only if there is evidence to support that $50 has the same effect as $0. That evidence does not exist. We can make a guess, but it is a guess that goes against experimentally demonstrated economic theory around marginal utility.
I guess we just disagree on what a control group is. I think of it as a group who are not provided any form of effective intervention. A placebo is an intervention where the user does not know they intentionally have been given an ineffective intervention. It is impossible to not know you were given $50, so I don’t know how this could be considered a placebo. A placebo controlled study is when you give a placebo to a control group who do not know if they are receiving a placebo.
The argument I think you are making is that $50 is so small as to be irrelevant. But it is neither a control nor a placebo by almost any definition. And there is no data to suggest what you are asserting even though it feels intuitive to make the guess.
There was no control group. That term is well defined, and giving an intervention is the definition of a treatment group.
What good do you think you might be doing?
So cynically, which is happening a lot in this thread, people are saying that there is basically no benefit to providing additional money at all, which is a conclusion that can only be arrived at by assuming that $0 == $50. As I pointed out elsewhere, economic theory around marginal utility says that the first $50 is going to provide more utility than any subsequent amount. For someone on the street, $50/month may well be a massive improvement in QOL.
Put this way, since there is no control group, we can't tell how much of ANY of the effects comes from the first $50 vs. people just being able to pull themselves out of homelessness through other means. We can make assumptions, but that is a great way to come to wrong conclusions. There is a good reason that most hard sciences will not accept a study that doesn't have an actual control group.
To answer your question about what good I am doing: 1. I am educating people as to why a rigorous control is necessary to reach solid conclusions, and hopefully getting them to understand why a business insider article is a shit substitute for reading the actual data. 2. I am pointing out that we cannot conclude that $0 == $50 == $1,000 when it comes to housing attainment because that data doesn't exist (we CAN say that in this study $50/mo == $1000/mo when it comes to housing outcomes)
I want people to understand that this study is far from conclusive in regards to rental attainment.
Almost nobody has literally zero income, and those that do probably have someone taking care of them every day, i.e., are not homeless because they are living with whomever is taking care of them. I'm talking about the US because that is the only country I know well. Theft, prostitution, buying a pack of cigarettes, then selling individual cigarettes, acting as a lookout for a drug dealer and getting welfare benefits by providing a fake ID are examples of ways of getting a little income -- and my guess is that for most homeless the income easily exceeds $50 a month.
The program would need a 90% success or it means that the rules are easily breakable, there are no strings attached to the free money, and people will absolutely figure out the loopholes.
This could be a step in the right direction if there is some form of filtering out people who will misuse the money, potentially making their situation worse by buying even more drugs or other bad decisions, and like anything else it will spready by word of mouth and the damage could be much worse than the current problem.
That's what I'm doubting because it's not like giving people money or resources hasn't been tried already. Places like SF have been extremely welcoming and helpful to anyone in need yet the problem seems to only grow, and I'm not seeing comprehensive understanding of what went wrong and why.
This drastically reduces both the cost of administering the program and improves the experience and accessibility of funds (many of the desired recipients are exactly those who have difficulty navigating bureaucracy.)
These pilot programs (of course) focus on homeless or lower-income individuals, but the theory is sound: a small amount of UBI more than pays for itself in reduction in demand for other social programs.
Yet, that's now what we do. Many people view homelessness as some kind of personal moral failure. when, in the US in particular, you'll be surprised how close most Americans are to homelessness. It takes surprisingly little to go wrong to end up there.
We would rather spend money on things like policing homeless people, even incarcerating them, which is objectively more expensive than simply housing such people.
Homeless problem (exotic poor) is completely different and mostly unconnected to housing affordability problem. You can’t house drug addict or mental regardless of housing prices. They will just destroy it. Locking them up is the only solution.
Ugh, this is a horrible article. 45% of people found housing, but there was zero difference between the treatment group (given $1,000/month), a group given a large lump sum, and a control group given $50 a month!
https://www.denverbasicincomeproject.org/research
A lot of people rotate in and out of homelessness, so we should expect some percentage to find housing after a year, even if nothing is done. Indeed, this seems like pretty strong evidence that giving more money does not help people find housing (at least in the range from $50/month to $1,000/month).
They cite a decrease in public service costs, but again, there is no difference between treatment and control! See page 6 in this PDF: https://static1.squarespace.com/static/64f507a995b636019ef88...
They calculate "saved costs" by estimating the difference in costs between the start of the program (a year ago) and a year into the program (now). However, since they are tracking a particular group of people, some of them will find homes and jobs on their own, so we should expect costs for that cohort to trend down no matter what. (This wouldn't lower costs overall, because in equilibrium, they would be replaced with newly homeless people with higher costs. But this is a cohort study, so those people wouldn't be included here.)
In this case, it's "compared to $50/month" or "compared to doing nothing."
This means the delta is higher for the people receiving 1000/mo.
The actual research paper is available, and while it suffers from a lack of a control group, it shows that there were indeed some pretty remarkable changes in the lives of participants, and that there were statistically significant differences between the low pay group and the high pay group (which... duh, $11,500 extra is going to make a huge difference). Of course, without a control group, it is pretty hard to say what the effect of the income is vs. other programs, self-help, external factors, etc.
https://static1.squarespace.com/static/64f507a995b636019ef88...
However universal really has to be universal e.g an entire country. This lets one measure the demand frenzy and resulting inflation.
What works really well is supplementing incomes for the poor.
Giving money out seems like the easiest thing to do, but much more effecting is increasing supply for basic goods and services. E.g housing, education, food, healthcare.
* a sociopathic nation state thats hell bent on bombing and relocating the original inhabants of that land,
* a private prison industry that uses its profits to lobby for more punitive sentences for nonviolent crimes
* a military industrial complex that destabilizes countires around the world
god forbid some small sliver of my tax dollars gets used to give homeless people some dignity and shocker keeping them off the streets which helps increase land value in the properties around where they used to hang out.
The real cost to taxpayers was either 0 or an influx of dollars due to slightly boosted local economy, not a loss.
I think they should try doubling the payments and multiplying the number of recipients by 10.
Probably don't have the money for that though and won't consider it even if they did have it.
The thing to look for are the charts that have the yellow star on them (statistically significant, p < 0.05). AFAICT, that means that those are the only 'real' findings in the study. I'll summarize for everyone:
- All 3 groups found a positive Change in Perception of Stable Housing
- All 3 groups found a positive Change in Financial Well-Being
- The Lump sum group found a positive Change in Health
- The $50/mo group found a negative Change in Energy (?)
- The lump sum group found a positive Change in Sleep Quantity (6 to ~7 hours/night)
- The $50/mo and $1k/mo groups found a negative Change in Sleep Quality
- The lump sum group found a lowering of Food Insecurity
- The $50/mo and $1k/mo groups found a lowering in Change in Parenting Distress
- The $50/mo group had lower Hope Scores
- The $50/mo group had lower Agency Scores
- All groups had less Hours Per Day Spent Accessing Resources
- the $50/mo and $1k/mo had greater Hours Per Day Spent for Social and Leisure
- The $1k/mo had higher Transportation Security
- The lump sum and $50/mo groups had lower Client Connection and Satisfaction with DBIP Partner Agency
The actual data of the study starts on pg 58.
A key finding is on the very last page; Changes in Public Service Costs. From there we find that the TOTAL COST SAVINGS was -$589,214. I'm not entirely sure of what is going on in the chart, but I believe (and please correct me) that a negative cost savings means that the costs then increased? So this program increased, by about $600, usage of Public Services. One wonders if they corrected for inflation over the time period.
Overall, an amazing study. Honestly, I kinda can't believe they got it funded. I think it's added a lot to the conversation about UBI and has, like any good study, prompted a lot more questions than answers.
Personally, I think muddied the waters more than anything. None of the measures of getting people off the streets were statistically significant, but that may be me misreading the data here. And a lot of the less key, but still good measures had people less well off than before (Client Connection and Satisfaction, Agency Scores, Hope Scores, Sleep Quality, Energy). It seems to me that most of the groups had an overall sense that this study was helping them out, kinda. But overall, I'm not all that sure it really did help anyone get off the streets here. It did seem to help with employment. This study, to me, really speaks to the difficulty that people have in becoming housed once homeless. That there are a lot of obstacles and that even an extra grand a month for a year isn't really enough.
I'd like to see a study that does the Mythbusters approach and sees what it takes to get people off the streets, just upping things until it happens. But, per this study, that is going to take a lot of money to accomplish. If anything, this work kinda bends one towards the idea that addressing homelessness isn't really worth it; it's just too hard of a problem. So, maybe another study is needed on how to keep people from being homeless in the first place?
Again, amazing work here. I feel this is going to be a seminal study for policy planners to argue over for decades.
That link might be a bit hard for people to follow...
I don't doubt that there was some degree of reduction but if you are looking at it as a "50% cure rate" or something like that, you shouldn't be claiming things like "fact" or "success/failure" regarding the topic, without at least adding some more context.
Providing $50/month is just a different experiment.
Edit: well I thought it was funny