Jeff Bezos, the richest man in the world, caused a stir last week by announcing a major philanthropic initiative that will devote $2 billion to address homelessness and education. Called the Day One Fund, the project has two priorities: funding existing nonprofits that serve homeless families, and creating a network of new, nonprofit preschools in low-income communities.
Two billion dollars is an enormous amount of money, enough to transform many lives. The good that Bezos will do is commendable. Unfortunately, he might achieve a lot less than he could.
Interventions in early childhood have a troubled track record; it’s not clear what works, and some interventions may cause harm. The picture for homelessness programs is a little more promising, but we still don’t know much about what work it would take to achieve Bezos’s inspiring vision statement that “no child sleeps outside.”
In choosing to focus his philanthropic attention on these projects, he seems to be approaching this initiative with none of the rigor or clarity that he prides himself on at Amazon. And by starting out already committed to two causes, Bezos undermines his fund’s ability to answer the question he launched it with: “Where are the opportunities to make things better?”
The hard lessons from charitable interventions in education
Bezos’s Day One Academies Fund will spend hundreds of millions of dollars to launch and operate a line of preschools for low-income children. The problem is that this has been tried before — and we really don’t know whether it works.
We’ve been trying for a long time to do early childhood education that produces lasting results for kids. The earliest attempt in the US was Head Start, the federally funded early childhood education program for low-income families, which has been running since the 1960s. There have been some studies with promising results, but there have also been many that struggle to detect any effect size.
On the pessimistic side, randomized controlled trials (RCTs) have found that it barely improves outcomes for kids — and the gains don’t last, typically evaporating by first or second grade. A 2010 study by the Department of Health and Human Services found that “averaging across all children, the benefits of access to Head Start at age four are largely absent by 1st grade.” The study tried breaking the data down further to see if there are any subgroups the program does help. It found some promising trends — maybe the program works for black 3-year-olds, but not anyone else — but also some subgroups in which the intervention seemed actively harmful, like kids in households with a depressed parent. (Subgroup analyses like that often just turn up noise, and I’m skeptical that there are meaningful results for any category.)
Many advocates of Head Start look more optimistically towards long-term results. Some researchers found that Head Start makes people less likely to be arrested in their 20s. It’s hard to know what to make of these findings. Longitudinal studies — the only kind that would detect an effect like that — often don’t fully control for confounding variables, and it would be odd for Head Start to have no effects for 10 years and then sizable effects in adulthood. Then again, maybe early childhood education benefits students in some way we’re still figuring out how to measure.
The mixed findings aren’t limited to Head Start, either. A randomized controlled trial of the Tennessee Voluntary Pre-K Program found some effects at the start of kindergarten, but by the end of kindergarten, the other kids had caught up. There was no effect in subsequent years (by some analyses, there was actually a negative effect — on average kids in the program did worse than the control group).
One possible explanation for the mixed findings is that some preschools are good and some are bad. One of the studies that found results for Head Start concluded that the children in their sample who went to non-Head Start preschools did worse than children with no preschool at all. Of course, there are also some studies (like the famous Perry preschool experiment) that have found promising results for other preschools.
The most worrying thing about Bezos’s proposal, given that background, is its focus on scale. In his opening statement, he set forth a goal that seems inappropriately ambitious given the difficulty of making progress in this area, writing that his fund would “launch and operate a network of high-quality, full-scholarship, Montessori-inspired preschools in underserved communities” and “build an organization to directly operate these preschools.”
He doesn’t just want to fund some pilot programs trying to find schools that outperform the spotty track record of previous interventions. Instead, he is trying to build a network of schools right off the bat. But without any reason to expect something to work, there’s no reason to scale it — and starting at scale is a great way to waste a lot of money, and even do damage if your intervention turns out to be a bad idea.
The homeless initiative might be more successful — but its impact may still be limited
The picture is a little more promising in Bezos’s other focus area.
Unlike preschool education, where we’re still uncertain how to get results that actually benefit kids, we know that spending money on homelessness gets people off the streets. A meta-analysis of dozens of RCTs of homelessness interventions found that housing-first programs (which connect people to homes without requiring them to qualify or comply with complex conditions), as well as intensive case-management programs, successfully keep people sheltered.
Inconveniently, we know less about which interventions are best for the specific homeless population Bezos is targeting: families. A meta-analysis of the evidence base for homelessness interventions ended up throwing out interventions targeting homeless families, because there just weren’t methodologically strong studies of interventions for this population. It seems likely that the interventions that work for single homeless people will work for families, but in philanthropy, interventions that “seem likely” to work often don’t.
On a more promising note, Bezos says he’s partnering with existing organizations to tackle homelessness, rather than trying to launch his own organization. That introduces a little less room for error. His initiative will, in all likelihood, successfully shelter some families — and that’s genuinely commendable.
Bezos’s $2 billion could make more of an impact elsewhere
For all the good it might do for the people it ends up reaching, Bezos’s Day One Fund will nonetheless achieve a whole lot less than if he were trying to do as much good as possible with $2 billion.
Is it fair to hold him accountable for that? I think so. Bezos said the founding questions of his fund are “Where’s the good in the world, and how can we spread it?” and “Where are the opportunities to make things better?” That means it’s reasonable to try to imagine how a fund that was dedicated to answering those questions would be run.
Some charitable causes are a lot more tractable, and a lot more impactful, than others. The gains from success are greater, and it takes less money and effort to make progress. This has been a slowly and painfully learned lesson in global development, where cheap, tractable, and impactful interventions — like giving everyone bednets — are getting more attention as givers sour on flashy interventions that don’t work. There’s a growing consensus in development now that aid-givers should prioritize impact — spend their limited resources on the interventions that work the best, whatever those are — that wasn’t there 10 years ago.
The idea shouldn’t be new to Bezos, though. Among tech startup CEOs, Bezos is well-known for how methodically he identified the opportunity for Amazon. He saw that the internet reflected an unprecedented new opportunity. He figured out what kind of business would be poised to take advantage of it. He ran that business, instead of running some different business.
What would that look like in the philanthropy world?
In philanthropy, like in business, there are some opportunities that are particularly important, tractable, and neglected.
When there are teams on the ground to distribute more malaria nets in Africa, but no one to provide the nets so they can be distributed, that’s a high-impact philanthropic opportunity. Supplying the nets can get more households coverage at a cost of only a few dollars a net.
When researchers think they might be able to design suffering-free and slaughter-free meat that is cost-competitive with factory-farmed meat, but need seed funding to make their endeavor happen, that’s a high-impact philanthropic opportunity.
When new research comes out suggesting that moderately sized cash transfers reduce child mortality and have effects that endure for years, that’s a high-impact philanthropic opportunity. You can arrange for studies at higher scale and make those transfers available everywhere if the results are borne out.
Most philanthropists, though, don’t seem to be looking for those opportunities. (There are exceptions.) They don’t seem to be prioritizing evidence of impact, and they don’t seem to be doing what Bezos so successfully did when he started Amazon: figuring out where his efforts will matter most by seriously thinking about what the world will look like in 20 years.
In his annual letter to shareholders this spring, Bezos exhorted them to have high standards in everything they do and to remain aware that in new fields, they might be setting the bar in the wrong place: “You can consider yourself a person of high standards in general and still have debilitating blind spots. There can be whole arenas of endeavor where you may not even know that your standards are low or non-existent, and certainly not world class. It’s critical to be open to that likelihood.”
He’s right. Wealthy individuals venturing into philanthropy are venturing into a new domain and treating it as straightforward in a way they should realize it isn’t. They’re not applying the same standards to their philanthropy that they applied to their businesses. It took the Gates Foundation decades, and billions of wasted dollars, to realize that effective action in the nonprofit world, like effective action in the for-profit world, is hard, and it takes specialized expertise.
Hopefully, Bezos will figure it out sooner.