Rachel Waddell: GiveDirectly’s emergency cash response to COVID-19post by EA Global Transcripts (The Centre for Effective Altruism) · 2020-09-07T16:28:59.098Z · score: 3 (4 votes) · EA · GW · None comments
In this EAGxVirtual talk, Rachel Waddell provides an overview of GiveDirectly's rapid response to COVID-19 across its countries of operation. She focuses on remote responses launched in Uganda and Kenya, in partnership with local community-based organizations (CBOs) and telecommunications partners.
We’ve lightly edited Rachel’s talk for clarity. You can also watch it on YouTube and read it on effectivealtruism.org.
Nigel Choo (Moderator): Hello. Welcome to this session on GiveDirectly's emergency cash response to COVID-19, with Rachel Waddell. Following a 10-minute talk by Rachel, we'll move on to a live Q&A session, where she will respond to your questions. [...]
Now I'd like to introduce our speaker for this session. Rachel Waddell is the Partnerships Director at GiveDirectly. She joined GiveDirectly from the Overseas Development Institute, where she was Head of Strategic Partnerships for Governance and Climate Change. Prior to that, she worked on climate change and sustainable development with the New Climate Economy, the Global Green Growth Institute, and the UK Department of Energy and Climate Change. She holds a BA from University of Nottingham and an MA from the University of Leeds. Here's Rachel.
Rachel: Hi. My name's Rachel Waddell, Partnerships Director with GiveDirectly. I'm really excited to be speaking with you today, and to have the chance to share some of the rapid response work that GiveDirectly has been doing over the past few months in light of the COVID-19 pandemic.
First, we'll discuss why cash has emerged as a critical tool in the COVID-19 response, and then we’ll dig into our Uganda program, which is one of the first ones that we launched. Finally, we’ll take a step back to share some of the key challenges and questions that we've been grappling with — and what they might mean for the sector more broadly.
As we know, coronavirus has affected nearly every community around the globe. Perhaps, [more now than any other] time in history, we're all in this together. However, we also know that COVID-19 is disproportionately impacting the poorest and the most vulnerable, both in terms of direct health impacts and secondary economic impacts.
It will hit Sub-Saharan Africa especially hard. The World Bank has estimated that the pandemic will plunge the region into its first recession in 25 years. Across the countries in which GiveDirectly works, the poorest are already feeling the impact from government restrictions on movement and commerce. In Kenya, the first country in which we launched a COVID cash response, recipients whom we've spoken to have reported sobering statistics. Ninety-four percent of those interviewed have seen a reduction in income, 68% have reported that they have no choice but to go to work for income [despite the risk of getting sick], and 92% have needed to reduce how much they eat.
These statistics are echoed across the countries in which we're working. It's clear that the situation calls for immediate action at a vast scale. Without rapid support, households will inevitably be forced to engage in negative coping strategies like taking on debt and selling assets for survival, and the impacts of this will last far beyond the COVID response measures themselves.
To address this emergency, the Center for Global Development (among others) has called for the rapid scaling up of cash transfers. Governments across the world are following suit. Social support programs are now being put in place or expanded across approximately 106 countries, with cash transfers representing 65% of these social assistance schemes.
Why is cash emerging as such a critical response? We all know the fundamentals by now. Cash interventions are backed by a rigorous evidence base that shows fairly consistent positive outcomes across nutrition, education, consumption, assets, income, and other indicators of well-being.
Also, some of the evidence most pertinent to COVID is that cash supports markets. It allows individuals to make choices that address their priority needs. It avoids creating market distortions, such as the challenges posed by an incoming distribution or oversupply of particular goods, which further erodes the incomes of local vendors and producers. And critically, it supports local economies. For example, a recent study done on one of GiveDirectly's programs in Kenya shows that for every dollar of cash delivered, there has been a local economic multiplier of 2.6.
Additionally, there are the particular advantages of digital cash. As we have seen across GiveDirectly's programs, digital cash is efficient. It also allows us to rapidly, securely, and remotely — which in this context is more critical than ever — deliver cash to some of those most in need.
That's not to say that we haven't needed to rapidly redesign our programs due to COVID-19. We have. And we've needed to find ways to target, communicate, enroll, verify, and audit and follow up with recipients without meeting them. We’ve also had to deal with new risks associated with cashing out, ensuring inclusion of the most vulnerable, and reaching high standards of fraud detection without any in-person contact.
Let’s look at Uganda to see how this has been panning out. Our most recent numbers [as of June 2020] indicate that there were 450 confirmed cases of COVID in Uganda. However, since March 25, the country has essentially been in a public and private lockdown. Transportation has been suspended and all non-essential businesses have been instructed to close. UNICEF did an early modeling exercise indicating that, if unmitigated, the fallout from COVID could increase poverty rates in Uganda from about 24% to 31%. Household incomes are expected to fall, particularly among the existing poor and those who are labor-constrained or informal workers.
These impacts are being felt across Uganda, although they're particularly acute in urban, informal settlements where residents are renters, don't tend to grow their own food, and are overwhelmingly likely to be employed in the informal sector.
Therefore, we’ve decided to focus our initial relief efforts in urban and peri-urban areas. We are, however, maintaining a close watch on how secondary impacts evolve across the country. And we’re ensuring that we build flexibility into our programming so that we can respond as needed.
We've adapted our usual programming. Specifically, we’ve:
* Pivoted to remote, home-based call center operations
* Increased the capacity of our call center to take on tasks that field officers previously did through door-to-door enrollment
* Developed methods of sending mass SMS messages to recipients, enabling remote enrollment at scale
* Developed new partnerships and targeting approaches to identify those most in need
The two primary approaches that we've pursued in Uganda are telecommunications targeting and partnerships with local community-based organizations or CBOs. We kicked off the program with a remote pilot phase, building on a pilot we already had underway. Last year, we explored the possibility of using mobile phone usage patterns as a predictor of poverty, with the aim of rapidly and remotely identifying and delivering support to those most in need. This gave us an existing data set of approximately 6,000 adults to immediately enroll in the program.
The development of the poverty prediction model is still ongoing [as of June 2020]. There's significant interest in it as a remote targeting model, so we're looking to accelerate its development, and incorporate targeting approaches if the model is shown to be successful.
The first part of the project’s second phase — Phase 2A — involves partnering with telecommunications companies. We're doing this to obtain mobile phone subscriber data in known high-poverty, high-vulnerability areas, which are primarily informal, urban settlements. These subscribers are then contacted via SMS and enrolled following receipt of consent.
[Phase 2B] recognizes the challenges around inclusion and exclusion errors. We're partnering with community-based organizations to provide lists of existing vulnerable individuals which can be ingested into the GiveDirectly system to start the remote verification, enrollment, payment, and follow-up process.
The scale of our COVID response goes far beyond what we have delivered to date. To give you a sense of it, over the last 10 years we've delivered cash transfers to around 170,000 households. Now, in Uganda alone, we're looking to reach upwards of 250,000 recipients over the next four months. We're also launching similar responses in Kenya, Malawi, Liberia, Morocco, Rwanda, and potentially the DRC [Democratic Republic of the Congo], with the target of reaching almost half a million people with $100 million raised.
Here are our top takeaways around whom to target, how to identify those people, and how much to give them:
In terms of targeting, our standard program model for identifying those in the poorest areas through national poverty data doesn't apply. The most vulnerable to COVID are not necessarily the historically poorest, as borne out in the impacts we're seeing around rural versus urban households. We've also been faced with a tradeoff between dispersing cash as quickly as possible at scale and mitigating inclusion and exclusion errors. Despite our best efforts, we will not be able to reach all of the most vulnerable. One of the issues that we're grappling with is whether we can find a way to reach those without mobile phones under the current restrictions on movement.
In terms of identifying and enrolling recipients, we can't go door to door. First, we're not able to do this under COVID restrictions. But regardless, a door-to-door approach would not allow us to reach the numbers we're targeting within the timeframes we’ve set. Therefore, we’re pilot-testing — and closely monitoring — different approaches. As I mentioned, we’re trying to identify and reach people through telco companies and partner CBOs. We’re also piloting closer engagement with existing government safety net programs.
In terms of transfer sizing, for this emergency response we've moved away from our standard model of $1,000 lump-sum cash transfers. Emergency cash transfers for the COVID response are sized to allow as many people as possible to meet basic needs and avoid destitution. Therefore, we don't expect to see the same spending around livelihoods and investments for asset purchases. One of the key questions that we're discussing and evaluating in real time is what the minimum transfer requirement is to effectively meet needs, and avoid destitution and negative coping strategies — while positively impacting local economies. We recognize that there are limited resources [available to people], so we need to respond on a scale that’s both unprecedented and significant.
I look forward to hearing your questions and reactions, as GiveDirectly is in unchartered territory. We're excited to not only deliver critical support to huge numbers across our COVID response programs, but also develop and share insights on the use of digital cash as a key tool for ongoing, wide-scale social assistance. Many thanks.
Nigel: Thank you for that talk, Rachel. I see we've had a number of questions come in already. The top two are somewhat related: GiveDirectly's program for US-based recipients seems to [contradict] previous reasons to donate to GiveDirectly. Why did GiveDirectly establish it?
Rachel: It's a great question — and one, of course, that I thought would come up. First, GiveDirectly isn't diverting from its core mission. Our core mission is still to work with people around the world who are the poorest. When we've done US-based disaster response before, we've had lengthy internal conversations.
Our thinking [in this case] has been that the need for COVID support around the world is immense, including in the US. The COVID pandemic has been a challenge in that we've needed to reorient our model; we had to do a rapid review and reorientation of our operations, which have generally involved door-to-door, in-person engagement, enrollment, etc.
As the pandemic hit and we paused our operations across all of the countries we support, the country that was most ready for that operational shift was the US, so we were able to move there pretty quickly.
Also, I think the nature of COVID and the pandemic has caused an incredible amount of support to flow in; people really want to do something. They're seeing the impacts of this in their own communities and countries. And a lot of them are knocking on our door, waiting and wanting to provide some kind of COVID response. Looking at our current statistics on the US COVID response, about 70% of our funders are first-time funders. We're tapping into a new crowd of people who either hadn't heard of GiveDirectly, or perhaps hadn’t fully bought into the idea of cash as an emergency response.
We're not seeing any evidence of converting funders from the international effort to the US effort, but we are attracting a new set of funders, and seeing quite a lot of evidence that people are being drawn to us through publicity like celebrity endorsements in the US. And actually, through that, they’re learning more about our mission and the work that we do in Africa, and converting into international donors.
It's certainly something we debate internally every time these situations come up. We’re monitoring it closely, and are very conscious of the fact that we don’t want to “cannibalize” funds for international efforts.
Having said that, we're seeing pretty profound impacts and getting good responses from our recipients in the US. We’re cognizant of the fact that there's a huge need there, as well as in the other countries in Africa where we're working.
Nigel: Great. Thanks, Rachel. The other question was framed in a slightly different way: Is it possible to make a strong argument in favor of donating to the COVID-19 US program rather than the Africa program?
Rachel: This comes down to personal choice. I think people are motivated, at this moment more than ever, to lend support in their direct proximity. People see the impact of COVID close to them.
Does a $30 donation go as far in the US as it does in Africa? No, it doesn’t. So is there a strong case to be made? It depends on the angle you take [in looking at this question]. We feel internally that there's a strong case for going ahead and doing the program. Part of our mission is to get as much cash, as quickly and efficiently as possible, to as many people as possible. But also, there's a top-level goal to try and shift the sector and inform [people about using cash transfers to address poverty].
You may have seen the press coming out about the US response, and there seems to be a real shift in how people are viewing cash as an emergency response. I think that we'll see a transformational impact from that over quite a long time. So, again, while $30 doesn't go as far in the US as it does in Africa, we're seeing those $1,000 donations in the US have a tangible impact on people's lives — in what's indisputably a really tough time.
Nigel: Thank you for that. Could you say more about monitoring and evaluating the US program?
Rachel: Yes. Across all of our programs, with the pandemic and the restrictions on movement that are in place, we've had to shift to completely remote monitoring. We’re using SMS messages and call centers to follow up with recipients. There is an inbound hotline to capture any adverse impacts or events, as there always is through our programs.
We just had preliminary results come in late last night on a much more in-depth survey of US recipients. I'm not in a position to share those results yet, but they will be announced publicly. I think some of the anecdotal stories are already out, but we are doing a more robust, in-depth evaluation as well, to work out how the program is impacting people's lives.
Nigel: Great — we'll keep an eye out for that.
The next question is: Do you think lockdowns in extremely poor countries do more harm than good, since they delay regular interventions for treating diseases like malaria?
Rachel: That’s an incredibly difficult question. We all saw that lockdowns, certainly in most of the countries where GiveDirectly works, came much more quickly and severely than we've seen here in Europe. There are a number of reasons for that [related to] health infrastructure and a government's ability to cope should an outbreak take off.
Having said that, it's clear that restrictions on movement — in particular in peri-urban areas — are already having a profound impact on people's lives. I shared some of the statistics from the Kenya program, but we're really seeing [similar numbers] across all of the countries that we're working in. People have very limited savings, so very limited means to survive once their income is cut off. Most rent their properties, don't grow their own food, and earn income from the informal economy. Therefore, a lockdown profoundly and rapidly impacts their ability to feed their families, not take on debt, and [refrain from selling] assets — all of these negative coping strategies.
Additionally, the majority of existing aid interventions have needed to be halted, at least to some degree, which exacerbates those impacts. But at GiveDirectly — despite those restrictions — we’ve found ourselves in the strange and somewhat fortunate situation of being able to shift our operations and actually deliver support effectively, safely, and securely on a scale that goes beyond what we've done before. And I think the level of support and the scale that we've managed to reach is a testament to [cash transfers as an intervention].
But yes — various governments are [faced with] huge, difficult decisions. And of course, it’s very hard to know how severe and widespread those lockdowns should be, and for how long.
Nigel: Thank you for your response. [Given the] restrictions around movement and current programs’ remote logistics, how did GiveDirectly overcome the issue of providing cash transfers to the unbanked, or those without mobile banking, in markets like India?
Rachel: This is a great question and one of the real challenges. For example, in Uganda, we’re primarily identifying significant numbers of highly vulnerable people through partnerships with telecommunications companies. We’re using cell tower data.
However, the glaring challenge with this is that if somebody doesn't have a cell phone, you're not going to pick them up. Therefore, we’ve been building partnerships with local, grassroots, and community-based organizations that are working with particularly vulnerable groups to understand how we can reach them.
If there are severe restrictions on movement, it's very difficult to provide somebody with a mobile phone, a SIM card, or access to funding. But through these CBO partnerships, we're able to bank folks along the way — people who we usually assist through door-to-door enrollment. It is certainly a challenge.
Nigel: Thanks for that. Zooming out from the COVID-19 response, based on GiveWell's selection of most effective charities, GiveDirectly ranks comparably low; it’s last within the top charities. Why do you think that is?
Rachel: Again, that’s an interesting question. I think there are a number of factors:
1. It's worth noting that it’s inevitable, in a sense, for GiveDirectly to be at the bottom of that list, because GiveWell uses cash as a benchmark to exclude any other interventions unable to demonstrate that they are more cost-effective than cash. It's not that GiveDirectly is at the bottom of the list because there aren’t any giving options that are worse. There are.
2. GiveWell’s evaluation model also uses moral weightings that are assigned to different outcomes. And with unconditional cash transfers, that [presents a] challenge, in that we spread our impact across a number of different applicants [editor’s inference: impact being spread out could mean that some outcomes will be too small or unusual for GiveWell’s model to account for them — but it’s not clear to me whether that’s what the speaker meant].
3. I think the final factor is that there's a broader role for cash in the sector as a benchmark. But I think cash is [inevitably] undervalued in that ranking, because there isn't an effective means of tracking how it shifts the sector to be more effective. Cash [transfers] encourage other interventions to demonstrate that they are able to beat that minimum cash benchmark.
Nigel: Great. I think that answers the question. Thank you. Here’s the next one: How could the EA community support GiveDirectly? What expertise and connections could you benefit from?
Rachel: We have a huge amount of support from the EA community, and have for a number of years. That support comes in a lot of different ways:
* We really value and encourage an open conversation and being challenged. Keep the questions around the US COVID response coming. Keep holding GiveDirectly to account. We think about [these issues] really carefully internally, and the EA community across the world has been a brilliant thought partner. That's highly valuable; individual members of the EA community have supported GiveDirectly by giving pro bono advice or engaging in informal discussions on particular topics that we're grappling with.
* The community has provided a huge amount of funding support over the years, which is invaluable. I think we find ourselves in interesting times with the COVID pandemic. I mentioned that the scale that we have reached through this pandemic is unprecedented for us. We've been a roughly $50-million-a-year organization for a few years; this is the first year that, already, we've exceeded $100 million. That is huge.
* There will be an important period of reflection following this one [during which we’ll appreciate your input]. [We’ve branched away from] our standard model of giving large, lump-sum cash transfers. [We’re using] a different type of disaster response model, and at the moment are working around the clock on delivering this aid. But what will it mean for us post-pandemic? [We’ll need to reassess] where we should be working and whether we should be doing more in the humanitarian space. We’ll call on the EA community. Send us your thoughts, talk to us, and join donor events and webinars, because you’re a smart group whose input is always really valuable.
Nigel: Great. Thank you.
Rachel: If there's something I haven't said, then folks should email me (firstname.lastname@example.org). If we’re missing a trick, and folks have some support in mind that they'd like to give, or a way that they'd like to be engaged, then feel free to suggest that as well.
Nigel: The next question circles back to your presentation. How representative was the survey? How many people were interviewed? Could you share a bit more?
Rachel: Great question. I'm presuming this refers to the early stats on Kenya. At the beginning, it was a fairly small sample group of a few hundred people. When we first kicked off the program there, we did internal follow-ups. We're continuing to do those, so actually, these statistics change as they’re updated over time. We've now reached almost 20,000 people in Kenya. All of these metrics are measured in real time. So those stats will change. It's an internal tracking tool where we follow up with all recipients from our call center. Now, because of the scale, we're doing a lot via SMS. But we are doing sample follow-up calls from the call center as well, and we're tracking all of that.
There are some other [studies] done by organizations such as BRAC. In the first few weeks of COVID, people were trying to rapidly get a sense of what was going on in the places where we work. Some of those [samples were] fairly small, but they give us an indication of what the secondary impacts of COVID have been during what, otherwise, would be a tricky time to complete a study.
Nigel: So, in a way, that expanded the sample size.
Rachel: Yes. We're collecting that data in real time as we're enrolling and talking to more people. The percentage with whom we follow up varies a bit across programs, depending on the scale. In Uganda, we aim to reach over 100,000 people — up to 250,000 and beyond. There, we'll follow up with 1% of recipients through phone call surveys.
Nigel: Great, thank you. On to the next question: What is the key idea you would like people to convey when talking about GiveDirectly?
Rachel: I think there are a few. The primary one [centers on] using unconditional cash transfers to empower recipients to meet their own needs — the idea that the poor know better how to meet their own needs than others do. They have proven time and again that they can effectively meet those needs if [given the opportunity to do so]. I think that's the primary message.
Nigel: Since GiveDirectly's model is a one-time cash transfer model, does that mean people who got cash before [the pandemic] are not entitled to a COVID-19 cash transfer?
Rachel: This is a great question again. Because of the nature of the targeting, it's unlikely that they would overlap. One of the things I think I mentioned in the presentation is that it's a bit of a different demographic group. GiveDirectly’s standard $1,000 lump-sum cash transfers go to people in the poorest rural villages. And for the time being (although we're closely monitoring the situation), our COVID response efforts are focused on urban and peri-urban areas. Based on the data that we've collected to date and [other available] analyses, the people in those areas are the most affected by the restrictions on movement.
Hypothetically, they wouldn’t be excluded from GiveDirectly’s standard program. Our COVID response is intended as emergency relief for three months. The amount varies a bit by country, but is around $25 to $30 a month. There is also, quite frankly, something about wanting to maintain the highest efficiency rates.
There are a lot of resources being pumped into this program, but it's obviously insufficient to meet the scale of the need. And there is an element of needing to accept [occasional] inclusion/exclusion errors in order to reach as many people as we possibly can, as quickly as we possibly can.
Nigel: Thank you for that. This question is somewhat related: Can you share some details of the indicators used to determine vulnerability? Are you just looking at mobile money records?
Rachel: Again, great question. It varies a bit across different countries. I mentioned that we're digging into a remote mobile data pilot. We actually started that pilot pre-COVID; it was originally designed to help us rapidly, securely identify and get support to people in the wake of disasters like droughts. We’re partnering with IDinsights on it. They're building an algorithm that uses mobile phone data usage patterns as a proxy predictor of poverty. We are still testing it; we're not actively rolling it out as a COVID response. But if that program is proven to work, it will be an exciting approach.
In other places, we're using a combination of approaches. One is geotargeting through cell tower data for areas that we know are some of the poorest informal settlements in certain cities. But we are overlaying that with information from local organizations, CBOs, and NGOs, and with government data on vulnerability and poverty. It's tricky, given that a lot of the poverty surveys that were done pre-COVID don't necessarily pick up people who have been impacted the most by the COVID-19 restrictions; that’s why we’re using a combination of sources and monitoring them closely. We're building in flexibility and remaining agile within our programs so that we can respond as things change.
Nigel: Thank you. I think we have time for just one last question: How valuable is lobbying state or national governments to try a full national or state-level trial?
Rachel: This is a great question — and a hard one to answer in 30 seconds. There is a lot of government interest, in particular in our COVID response, and I think we find ourselves in a stronger position than ever before to inform government strategies going forward.
A lot of governments, as I mentioned, are rolling out expanded or new social assistance programs in the wake of COVID. Cash is playing a critical role in that, and a lot of governments are looking to us to share learnings from our work that they can use to inform their own programs. We are engaging governments really closely on that. I'll leave it at that.
Nigel: Yes. Thank you very much, Rachel. I think that was a great point on which to end the conversation. That concludes the Q&A part of this session.
Rachel: Thank you.
Comments sorted by top scores.