THE HAGUE — As the United Nations’ Centre for Humanitarian Data took shape last year, its founders had to think hard about what it would not be doing. It would not be the place where all humanitarian information, from cash transfer volumes to vaccination rates to rainfall, is analyzed. It would not handle big data, and it would not collect data itself.
Instead, after consultations, the goal set was to increase the use and impact of data across the aid sector. The overarching aim of the center, which is part of the U.N.’s Office for the Coordination of Humanitarian Affairs, is that “everyone involved in a humanitarian response has access to the data they need, when and how they need it, to make responsible and informed decisions.”
One long-term ambition is to create a “crisis dashboard,” so that anyone can promptly get a comprehensive assessment of how a region is faring in order to plan a humanitarian response. In the short term, however, its challenge is to meet a set of self-imposed progress indicators and prove to its backers that it is a necessary investment amid a growing field of organizations working on humanitarian data.
The Data Guardians Series
The collection of data on digital platforms has become ever more central to aid work, as organizations strive to ensure their interventions are as efficient, effective, and targeted as possible. But while data can be a transformative tool, it also comes with risks. As more data about more beneficiaries is gathered, are the rules around how we use it keeping pace? Are aid organizations being too cautious in their use of data — or are they not being cautious enough? As scandals over data protection push these concerns further up the agenda, Devex’s Data Guardians series explores the issues affecting aid organizations as they work to protect their beneficiaries’ data, and the debates and practicalities around what more can be done.
HDX & HXL
In December 2017, the center launched with three years of funding worth $5.5 million from the Dutch government and 400,000 euros ($474,300) from the city of The Hague. It has since added a grant from the Education Above All foundation, while the Tableau Foundation provided licenses and training worth about $270,000.
From the start, the center has four areas it wants to improve: Data interoperability; data literacy; data policy — designed to improve the sharing of information during emergencies; and network engagement.
“We need a product,” the lead for the center, Sarah Telford, told Devex at its head office in The Hague recently. “I don’t think we could have a center that was just about thinking.”
The most prominent of these products is the Humanitarian Data Exchange, or HDX, OCHA’s data sharing platform that launched in 2014 and is now managed through the center. HDX lets any approved organization upload data to an open database on subjects such as population levels, food insecurity, infrastructure, violent crime, or a country’s funding needs. It has 30,000 monthly users, of whom 60 percent are in headquarter locations, including New York, Brussels, and Geneva. So far, it includes around 6,700 datasets, covering every humanitarian crisis as well as development data. For Myanmar, for instance, you can find out the number of students and teachers in the country from UNESCO, health indicators from the World Health Organization, and the path of every storm in the Asia-Pacific region for the past 50 years from OCHA.
But HDX is not just about aggregating data, Telford said. “It has to be about what the data adds up to, what the insights are, and then how that connects to decisions and more effectiveness,” she said. The center’s key performance indicators, drafted with the Overseas Development Institute, a London-based think tank, acknowledge that “it is not within the center’s mandate to be responsible for how data can lead to better decisions.” Rather, the hope is that by improving the quality and interoperability of data, users can discern new insights for themselves.
That is where the Humanitarian Exchange Language, or HXL, comes in. By adding common hashtags such as #sector, #affected, and #reached to different humanitarian datasets, the idea is to make all the information easily accessible for different users. In June 2017, when the center took its baseline measurements, it had 903 HXL-tagged datasets from 39 providers. Now, it has 1,216 from 48 providers. By 2020, it wants to triple the number of datasets and double the number of providers from 2017 levels, but “eventually we would like all relevant data to be in HXL,” Telford said. “We believe that is the key to unlocking cleaner, interoperable data that can be combined quickly and sustainably for insight.”
Any organization uploading data to HDX must specify when it was collected. If it goes out of date, the center can send a nudge for the owner to provide an update. Time matters in an emergency, too. During a hurricane response, for instance, most data concerning which organizations are doing what is only uploaded to the HDX in the second month of the response. The aim is to make sure that, in future, this happens in the first week.
Yet the center is still grappling with how to ensure quick, easily uploaded data is also safe. Amid high-profile scandals around data protection, the greatest threat Telford sees to humanitarian data initiatives is a loss of trust. “The policy and doctrine gaps relating to data privacy, data sharing, and other areas will take years to address, but a failure to do so will create major risks for the people we serve,” she wrote in a recent article for Devex.
Devex's Data Guardians series explores the issues affecting aid organizations as they work to protect their beneficiaries' data, and the debates and practicalities around what more can be done. Here, Devex spoke to experts about how aid organizations can prepare for the onset of the most stringent data protection regulations worldwide.
At the same time, the European Union’s new data protection regulation, entering into force this month, has brought the need to safeguard personal information front of mind for NGOs and aid agencies.
At the moment, the center has a duty officer constantly online, in charge of checking datasets as soon as they are published to ensure there are no names, phone numbers or social security information included, which would breach the HDX terms of service.
Telford acknowledged that the current system is imperfect. “In that moment [before it is reviewed] the data could be accessed because it’s already open, so it’s almost too late,” she said.
So far, no personal data has been inadvertently revealed in this way, she said, but in an effort to address this weakness, the center makes a point of repeatedly emphasizing its terms of service with approved organizations, and is hoping to add “speed bumps” to the process of uploading data that would prompt organizations to make additional checks.
The greater challenge, she said, comes with potentially community identifiable or demographically identifiable information, since anonymizing data is not always enough to protect individuals. “You actually have to do analysis on that to understand the risk of reidentification, and that’s where we run into trouble,” she said. “That takes time and there’s a certain level of expertise. We’ve only been able to uncover that sometimes when we’ve been asked to build a visualization, and then we look at the data and you can see that for this question there are only five responses, and you could tell that it’s a widow [and] you could probably figure out who that might be if you lived in that village.”
There have been two or three cases in the four years of HDX’s existence where running shared data through a statistical disclosure control, or SDC, revealed a high risk of re-identification, Telford said.
To protect against this, the center recently applied for funding from the European Commission’s humanitarian arm, ECHO, to create a separate workflow for survey or sensitive data. Known as HDX Secure, the limited access, secure infrastructure would run data through an SDC process before publication.
“We believe this service could be useful to organizations that do not have this level of expertise and could improve responsible data practices across the sector,” she told Devex. “The secure infrastructure could also be used to transfer sensitive data within OCHA offices or between partners but there are obvious risks to opening it up to too many users.”
Partners or competitors?
Telford leads two dozen people working as part of the center’s team, including one OCHA employee in The Hague and two in Geneva, with 20 consultants working from places including Nairobi, Dakar, Copenhagen, and New York.
The network engagement component of the center’s mission envisages information sharing, project collaboration, and secondments with other organizations, designed to “build and engage an active community in support of its mission and objectives.” In the first months, that has meant a lot of meetings.
When Devex visited the center’s head office, Telford and her colleagues were hosting four members of the World Food Programme’s Mobile Vulnerability Analysis and Mapping, or mVAM, unit, whose data on food security, among other things, is shared on HDX. Over the course of about five hours, the conversation ranged from how to protect the identities of people surveyed in conflict areas, to the difficulties of teaming with the private sector, to the politics of when to declare a food emergency. Relying on detailed data to make this decision, it emerged, can help insulate WFP country directors from the criticism of displeased governments.
The meeting also showcased the impressive detail of WFP’s data on subjects such as commodity prices and weather patterns, as well as their user-friendly graphics. The aim of the mVAM unit is to create a system to detect early warning signs of a food emergency.
But they aren’t the only ones. Action Against Hunger and the World Bank are also trying to use data analysis to identify patterns that may indicate malnutrition or signal a coming famine. In the past, that has sometimes lead to more competition than collaboration, though WFP is now working with both agencies on how to better share data and technology.
The Centre for Humanitarian Data is aware of the risk of duplication in its work, too. A report on the concept for the center, delivered in March 2016 by the consultancy Dalberg, attempted to map the growing humanitarian data ecosystem. It pointed to groups such as Global Pulse, the Global Humanitarian Lab, the Global Alliance for Humanitarian Innovation, Global Partnership for Sustainable Development Data, the UN Data Innovation Lab, the Development Data Hub, and the Digital Impact Alliance — all of which have aims similar to aspects of the center’s mission.
“The data space needs investment, and it needs focus, and it needs this senior level buy-in and support.”— Sarah Telford, lead for the Centre for Humanitarian Data
Still, the consultants found an unmet need for training and capacity building to increase the understanding and application of data, including coordination to help innovations scale and avoid duplication, as well as better data standards. They wrote that they “clearly see the opportunity for a coalition, led by OCHA and the Netherlands, to intervene in the ecosystem and drive improvements across humanitarian data.”
“In Silicon Valley, you have angel investors who come in and put money into things and some of them work and some of them don’t,” Telford said. “We don’t really have the same discipline in the humanitarian community. A lot of times, things are created and they go on too long, even if they are not delivering value.” One of the things she said she took from the World Humanitarian Summit in Istanbul, Turkey, in 2016, was the need to focus more on results and, if these aren’t forthcoming, to change and do something different.
At the same time, Telford welcomed the growing number of data initiatives in development and humanitarian work. Last month, the U.N. High Commissioner for Refugees and the World Bank established a “Joint Data Centre” on forced displacement. “A few people wrote to me and were kind of like ‘oh, another data center.’ But I feel differently,” Telford said. “The data space needs investment, and it needs focus, and it needs this senior level buy-in and support. The more that we have that, the better it is for us.”
As chief executive officer of the Global Partnership for Sustainable Development Data — a network of 280 governments, companies, civil society groups, and others working to promote better data as a means to meeting the Sustainable Development Goals — Claire Melamed also has a clear view of the suite of new data initiatives.
HDX plays a “hugely important role” by offering a unique product, she said, praising the center’s part in “building those relationships of trust that are so important in this field.”
At the same time, Melamed said some overlap with other organizations’ work was inevitable. “As long as we are responsible about it, communicate well, keep track of what each other are doing, and do the best we can to build the sector as a whole, a small amount of duplication might be a price worth paying for the huge amounts of innovation that we are seeing at the moment.”
“Creating efficiencies in making data available and interoperable, and then building services that make it easier for agencies to do what they need to do is a clear role for the center.”— Stuart Campo, a senior fellow at the Centre for Humanitarian Data
In this brokering role, Telford said the center is uniquely placed. “We’re part of OCHA, and OCHA’s mandate is to coordinate. You don’t have that same role in the development space,” she said. “There isn’t a coordinator of all the development actors.”
Stuart Campo, a senior fellow on Telford’s team and researcher at the Harvard Humanitarian Initiative, told Devex it is precisely the chance to prevent duplication that represents one of the center’s greatest potential strengths.
“If three different agencies are expressing interest in the same data or related analytical tools and services, then the center certainly could have a role to play,” Campo said. “Part of the challenge is that every agency needs — or feels it needs — slightly different analyses, even if it’s the same methods and data. So, there are limits to the degree you can centralize that. But certainly, creating efficiencies in making data available and interoperable, and then building services that make it easier for agencies to do what they need to do is a clear role for the center.”
Campo, who is coordinating the inaugural class of the Data Fellows program this summer, was attracted to the center by its clear objectives, something other initiatives sometimes lack, he said. But he sees sustained demand and engagement within the humanitarian community as critical to the center’s success. Its offer of faster data, better tools and services, and enhanced coordination is highly valuable, “but if it doesn’t get understood and taken advantage of, then that’s a big risk.”
As a result, the team spends a lot of time on outreach efforts, including through blogs, leaflets, films, and impact stories. Ensuring that the small team has a presence in as many locations as possible is part of that outreach effort, Telford said, adding that they are hoping to add another group of team members in the Asia region this year.
“We have seen an increase in awareness and the legitimacy of our work since the U.N. Secretary-General opened the center last year,” she said. “We are also encouraged when we hear the head of OCHA, Mark Lowcock, or senior officials at ... the World Bank talking about our work. If we can get leadership on board with the value of data sharing and the use of shared standards and infrastructure, then it will make our work a hundred times easier.”