3 common challenges of delivering digital social protection

Jenni Henderson has been working in consulting – primarily in the field of international development – since 2013. In 2019, she started an Mphil in Public Policy at the University of Cambridge. As part of her studies, she spent the summer of 2020 on a work placement with Public Digital. Here, she worked with the international development team and researched the digitalisation of social protection programs in light of the global pandemic.

This post is a summary of that research.

There are excellent examples of digital teams across the world who have rolled out a digital response to coronavirus at pace. For example, more than 800 social protection measures were planned, introduced, or adopted across 171 countries in the first 4 months of 2020. And more continue to unfold in response to – and in spite of – the pandemic.

Many of these countries had strong foundations of digital infrastructure which they were able to scale-up, adapt or build on to deliver wide-reaching, fast, and transparent social protection measures.

However, having spent the last several years working in development finance and based predominantly in low-income or emerging economies, I suspected that for many policymakers taking social protection programs to the digital realm would be no easy feat.

So, for the last stretch of my MPhil I spent the summer researching digital social protection in lower-middle income countries (LMICs).

Here are a few stories I came across in my research, which highlight common pitfalls of digitising social protection. I draw on LMIC examples, but the lessons from them are relevant to us all.

Attempting to scale without starting small

In 2012, Ti Manman Cheri (TMC) became Haiti’s first government-led conditional cash transfer program. It was also Haiti’s first attempt at digital social protection. Its aim was to provide cash to recipients using mobile money through TchoTcho Mobile – a limited purpose, financial/mobile instrument, in partnership with a private sector mobile network operator, Digicel.

Political urgency led to the program moving to whole-of-country implementation only shortly after its launch and before adequate user research and testing.

As a result neither the government nor the private sector partners were equipped to roll out the program as promised. The various stakeholders faced many challenges (mapped out extensively by Consultative Group to Assist the Poor in this assessment), but here are the key takeaways:

Firstly, the payment service provider (PSP) lacked a sufficient network of customer-facing teams and locations to implement e-transfers nationwide. As a result, a second PSP was assigned. Digicel’s TchoTcho Mobile allowed citizens to hold their received government payments for extended periods of time and also use the product for other transactions. The second provider, however, did not have the infrastructure to link payments with an account for users to store-value, so recipients tended to withdraw their full amount at one time. This meant that where cash could be withdrawn, customer-facing teams were often overwhelmed.

Bureaucratic processes within Fonds d’Assistance Economique et Sociale to issue payments involved several layers of stakeholders’ approval and proved inconsistent. At the same time, Digicel’s inability to schedule payments on its mobile platform meant staff had to upload payments manually for real-time cash transfers.

Poor integration between the government’s TMC program and private-sector payment service providers (PSP) created challenges with data management. These generated issues in recipient registration, verification, release of payments, and monitoring. Common data issues stemmed from incorrect or different phone numbers, ID numbers, and conflicting names between TMC beneficiaries and linked mobile users on PSP platforms.

Cumbersome manual reconciliation processes resulted in a backlog of unpaid recipients rejected by Digicel’s payment platform. Being a small team without any historical experience on implementing mobile-based cash transfers and limited operational training, the government’s TMC program staff struggled to keep up.

Giving a small, inexperienced team the chance to start small, test with real users, and identify & address challenges iteratively, might have helped.

Buying new IT before investing in people

Ghana’s e-Zwich payment system was introduced in 2014 with the objectives to improve public sector transparency and accountability and extend financial inclusion to vulnerable people.

Yet, despite great excitement over the sophistication of its underlying technology— the “world’s first biometric money” which allowed digital payments to be processed without reliable internet— its initial results were mixed.

It was expected to reach more people and remove substantial leakages from public financial flows and payrolls. However, as discussed in this analysis, take-up was slow, and the technology encountered opposition, particularly from Ghana’s public sector unions.

Why? Analysis from the International Monetary Fund (IMF) concludes that “in retrospect, it would have been better to have separated out the unique identification system from a particular financial technology.”

The ‘technology-first’ approach did not match users’ needs or expectations (both citizens, and bureaucrats.)

It is one thing to build digital infrastructure, but another all together to build people’s capabilities or desire to use it.

The Haiti example, discussed above, also made this error. Staff reported they underestimated the need for recipient education on TMC payments.

As is common for many digital social protection programs, it can be recipients' first time using a mobile money product. In Haiti, it was the first time any recipients had registered for a mobile-phone-based G2P program. As a result many did not provide their phone number to receive payments or provided an incorrect name for matching with the Digicel system.

The level of consumer (user) education that is required when both a new technology and new ways of delivering a service coalesce cannot be underestimated. This makes good service design all the more important.

Similarly, the design and roll out of digital social protection, particularly in LMICs, takes disrupting and upskilling bureaucracy.

A common pitfall in several case studies was outsourcing the ‘digital’ side of a new social protection program to the private sector entirely. This led to public sector program managers falling out of step with the program they were meant to have a handle on, while the private sector pushed ahead in designing mobile-money or tech platforms inaccessible to the type of individuals the social program was intended for.

Underestimating socio-cultural complexities around identity

I spoke to an anthropologist who has spent years living in Tanzania among Massai tribes. An anecdote she shared of obtaining visas for a Massai choir group before touring North America illustrates some of the socio-cultural challenges of establishing identification, which is a prerequisite for digital social protection.

“The visa form required the names and birthdates of each parent and step-parent and every sibling, half-sibling and step-sibling—no small thing for a polygamous community where a man can have anywhere up to ten wives and one hundred children (and then some).

The Maasai do not use calendars, events are chronologised in relation to other big events. Thus Matteo’s birth was determined to have occurred around the times of the last big rains, when the grass was particularly green. Whether this was before or after the birth of his half-brother was up for debate (his brother was either a year older, or a year younger than him, depending on whom you asked).

The real difficulty arose, however, when we asked Petro for his mother’s name.

‘Mama Petro’, he promptly replied.

‘Yes, but what was her name before you were born, Petro?’

‘Ah, Mama Nemburis.’

‘Right… But before she had Nemburis or any children, what was her name then?’

‘Oh. Wife of Isaac.’

‘Ok, but in the beginning, when she was first born, what was she called by her father?’

‘Daughter of Matteo.’

‘Didn’t she have any other name?’

By this point, we had amassed quite a crowd and a heated debate ensued regarding this worthy woman’s name. Eventually, it was declared that of course she had some other name, but since no one present could remember what it was, it would be best to ask her herself.

So we all herded over to Mama Petro’s house for a mug of chai and a chat. She, of course, remembered what her own name was, prior to the birth of her illustrious children: ‘Wife of Isaac’. Prior to that, she couldn’t quite recall…

At the end of the day, we decided that ‘Neema’ was a lovely name. We entered it into the Grand Maasai Database, a sizable excel spreadsheet of all the details we had helped people make up for their birth certificates, passports and visa applications over the years.

Some of our friends now have more than one legal name and several birthdates, depending on which item of ID you refer to. They often don’t remember these details themselves, of course. That’s what Facebook is for.”

Nearly one billion people worldwide are estimated to lack ‘acceptable’ proof of identity. 29% of adults in sub-Saharan Africa have no way of identifying themselves. This percentage is much higher among women, youth, and the very poor.

The Malawi Government, for example, invested US$52 million to register over 9.1 million citizens in 2017. It required a large-scale campaign divided into 5 phases each covering 4-6 of Malawi’s 28 districts, 2,000 customized biometric registration kits (BRKs), and a lot of workers.

Even this, however, encountered challenges in cultural, and social norms. Religious beliefs in certain areas around the capture of fingerprints resulted in a few incidents in which villagers attacked and chased away registration agents, requiring police intervention.

The underpinning technologies for digital social protection are digital identification, social registries, and payment systems. My point? For a digital social protection system to genuinely support the vulnerable or hard to reach, it will take time, investment, effort and genuine desire to understand human needs.

It is an easy pitfall for bureaucracies or donors to underestimate the socio-cultural complexities of translating people’s identities to data.

Earlier this year Claire Bedoui shared several success stories of different governments’ use of tech in response to coronavirus.

One thing that struck me during my own research was how quickly governments could successfully roll out digital social protection programs when they had pre-established digital infrastructure and capable teams.

Peru, for example, leveraged existing G2P systems/processes and quickly included private banks and mobile money providers as service providers to increase its reach. Meanwhile, it’s civil registration agency, RENIEC began using digital technology in 2006 and had already achieved near universal registration. (To note, the agency prioritised social inclusion and took steps to ensure that forms and information were available in major language groups and followed the naming practices of each ethnic group.)

Chile and the Philippines have also been celebrated for rapid digital social protection measures. Both were able to make use of well-established G2P payment ecosystems.

For many LMICs the challenges of making social protection programs digital are tangible. Building the back-end of bureaucracy; balancing investment in technology with investment in citizen education; and putting in the hard-yards of identifying and understanding who citizens are and what they need, takes time and investment.

Considering some of the success stories, the best time to start putting in this effort and investment was probably years ago.

The second best time is now.