The leaky bucket: Can digitization of social welfare programs reduce leakages?

Globally, developing and transition economies spend an average of 1.5 percent of GDP on social assistance. While social assistance has helped reduce poverty, there has been a clear failure in targeting the poorest populations. 

Social assistance schemes are non-contributory interventions (i.e. the government or other providers pay the full amount of the assistance) designed to help individuals and households cope with chronic poverty, destitution, and vulnerability. Examples include unconditional and conditional cash transfers, non-contributory social pensions, food and in-kind transfers, school feeding programs, public works, and school fee waivers. Across developing countries, estimates from available household survey data show that on average 7 percent of people escape absolute extreme poverty because of receiving social assistance transfers. 

But the most vulnerable are often left behind. According to the World Bank’s ASPIRE database, only 22 percent of the poorest households in sub-Saharan Africa receive any form of social assistance, and only 24 percent of all expenditure on social assistance reaches the poorest quintile on the continent (see Figures 1 and 2). This implies that 78 percent of the poorest people in sub-Saharan Africa receive no form of social assistance from the government. Further, the poorest and richest quintiles receive similar proportions of social assistance transfers making social assistance programs in sub-Saharan Africa the least progressive compared to other regions (Figures 1 and 2).

Source: Parekh & Bandiera, 2020

Source: Parekh & Bandiera, 2020

Accurate targeting is fundamental to successfully implementing social assistance programs. The benefits of social assistance programs are often aimed to reach the poorest people of the country--as typically measured through a means-test or an income-test--but in reality programs that effectively find these beneficiaries can be extremely hard to design and implement. When people are involved in informal or agricultural jobs, they often lack documentation of their wealth making it difficult to accurately measure or verify the income of potential recipients. This challenge is intensified by the absence of advanced social security or tax records. This lack of information can cause some undeserving beneficiaries to enter the program (what we refer to as “inclusion error”) or leave other vulnerable households outside the safety net (known as "exclusion error”). The inadvertent denial of benefits to legitimate participants can lead to potentially dire consequences for these individuals and can also undermine public support for such programs, as well as other state-led income redistribution systems. To what extent and how can biometric IDs and digital financial services (DFS) address both inclusion and exclusion errors?

Improved data & identification

With limited access to banks and formal financial institutions, administrative data (i.e. routine information collected, used, and stored by governments and other organizations primarily for operational, rather than research purposes) on the wealth and income of households remains lacking for poor populations. The introduction of digital identification (ID) systems and DFS provides an alternate and, possibly more precise, solution to this identification barrier. 

Digital ID systems can potentially alleviate this lack of accurate income data by facilitating improved record-keeping and generating administrative data on individuals in the country. Digital financial services enables documentation of a person’s financial transactions that can be used to create indices for eligibility. Further, digital government-to-person (G2P) payments could also provide a trove of data that would be linked through identifying factors to create a more robust dataset on beneficiaries. This would allow programs to leverage data collected by other programs. For example, in Togo a digital cash grant targeted to households in the informal sector (Novissi) used the voter ID database—from February 2020 presidential elections—which contained precise location and occupation information, to identify beneficiaries. Of course, centralized detailed datasets could also lead to state surveillance and misuse of data if adequate security measures are not in place. 

To support the claim that digital IDs improve targeting, a difference-in-differences study in India found that the introduction of a new technology that allowed for the direct deposit of transfers into a beneficiary’s account in a government transfer program reduced leakages to ghost-beneficiaries (i.e. beneficiaries who did not really exist).

Improved monitoring systems

Further, digitization of payments can lead to more effective process monitoring, which enables implementers to make real-time decisions based on incoming data. This is particularly crucial during unexpected shocks, such as the on-going COVID-19 pandemic. Currently, governments are using innovative digital solutions such as satellite imagery and alternative data sources to reach those most affected by the pandemic. The existence of a mature digital payment and ID system could have greatly helped governments respond to the crisis as seen in the above example from Togo. In a context where there is a reliable identification system in place, digital data collection can make policy decision making easier and more responsive to the needs of the people on the ground. 

However, the risk of technology failures leading to exclusion of eligible beneficiaries remains if access to biometric IDs is not universal or systems are not well implemented or managed. This could especially be a problem if more vulnerable populations are left out of the ID system. Further, issues around privacy and data misuse remains a major concern when data is collected and held centrally. 

What about stateless groups?

Some populations, including refugees, migrants, and other ‘stateless’ groups, may be particularly challenging to reach or may inherently be excluded from specific ID systems on the basis of nationality/citizenship. In such cases, complementary ID systems may be necessary to reach these populations, in order to facilitate access to social services and increase economic integration. 

For example, in Kenya, UNHCR has established a biometric database of refugees and asylum seekers that runs parallel to their national registration system. Elsewhere, efforts are in place to introduce regional ID systems that extend beyond national identity. In West Africa, for example, the West Africa Unique Identification for Regional Integration and Inclusion (WURI) program is being piloted in Côte d’Ivoire and Guinea to provide identification to citizens and non-citizens alike, with the goal of reaching vulnerable populations and facilitating access to services at both the country and regional level.

A word of caution

The integration of ID systems with payments is not, however, a fool-proof solution to targeting problems. Collusion to siphon a portion of the transfers could occur due to imbalanced power dynamics. For example, local bureaucrats may establish sharing agreements where a cash transfer is approved in return for a proportion of the transfer. 

Collusion could also occur at the central level through state capture or conscious efforts by administrators or the state to exclude beneficiaries from certain groups, particularly if the ID system is linked to voting rights. Empirical evidence of detecting such collusions is limited and there is tremendous scope to explore and document its existence, as well as possible solutions. 

Moreover, if beneficiaries are unable to navigate digital G2P payment systems, they may not be able to access their grants even when eligible, as seen in Jharkhand, India. Here, the use of Aadhaar—a digital biometric identification card—as a digital tool to authenticate beneficiaries, reduced the benefits for those who had not registered for an ID. Given that more vulnerable populations are often also those who have less access to knowledge on technology, this could mean the most vulnerable are particularly susceptible to exclusion when shifting to a digital payment mechanism. 

What’s next?

The Digital Identification and Finance Initiative (DigiFI) is currently funding projects on various aspects of targeting using digital systems. 

• In Malawi, researchers are exploring whether digital IDs can reduce poverty;
• In Kenya, a study is evaluating the inclusion and exclusion effects of linking digital IDs to social protection programs; and
• In Ghana and Kenya, researchers are studying how digital social transfers can help citizens during the COVID-19 pandemic. 

You can read more about our research agenda and existing projects. Yet there is an urgent need for more evidence in sub-Saharan Africa on whether digital systems help achieve accurate targeting and efficiency in public programs and whether such systems ensure more equality for hard to reach and/or marginalized citizens. If you are interested in pursuing research on these topics, please reach out to our team at [email protected].

Author’s note: This is the fourth blog in the DigiFI series on the various aspects of their research and policy priorities. The next blogs will explore to what extent digitization can help tackle corruption and other leakages, better incentivize public administrators, and improve take-up.

As of December 2018, two-thirds of total global mobile money transactions were driven by users in sub-Saharan Africa (SSA), with values exceeding US$25 billion. In 2017, the region recorded roughly 135 distinct mobile money implementers, and 338 million accounts. Mobile money has been touted as a revolutionary tool for expanding access to financial services in low resource environments. With just a mobile phone, users are able to quickly transfer money at low cost and, typically, without needing access to an existing banking account. 

Low resource settings—where families and friends financially support each other, even when geographically distant—are particularly well-placed to realize the benefits of this technology. The amount remitted by migrants from SSA reached US$48 billion in 2018. In Nigeria, remittances reached US$25 billion in 2018—almost four times more than foreign direct investment and official development assistance combined. In Lesotho, remittances amounted to 16 percent of the country’s gross domestic product. Prior to Kenya’s recent growth in mobile money services, household members had to find creative ways to send money to loved ones who lived across the country. These informal processes—such as sending physical cash with friends or bus drivers—were very expensive, caused delays, and created uncertainty that the money would reach the anticipated destination both from a logistical and security perspective. Mobile money offered a solution with direct person-to-person transfers. 

But does this technology live up to its promises? What are the impacts on the households who use mobile money?

Mobile money can generate financial resilience

A growing body of research is emerging with a consistent finding: households are able to better respond to unforeseen difficulties when they have access to mobile money. When an unexpected negative event—such as a flood or a child falling ill—occurs, households with mobile money are able to rely on the easy and affordable transfer of money from family and friends even if they are living far away.

In Kenya, a study that uses a difference-in-difference technique has shown that the impact of mobile money on a household’s resilience can be sizable. Households who had access to M-PESA (Kenya’s mobile money platform) during a negative event—such as unexpected extreme weather or illness in the family—were better able to respond and did not have to reduce their spending on food and related goods in response to the event. This is because mobile money helps risk-sharing among the community or family irrespective of location, strengthening these informal insurance networks.

In related work that also uses a difference-in-difference methodology, researchers looked at the effects of a health shock and found households who used M-PESA were able to spend more on health-related expenses while keeping up with other household payments. Households who didn’t use M-PESA financed their health-related expenses by cutting down on non-food expenses, including withdrawing their children from school. Similar results have been found in other contexts using randomized control trials in Mozambique and Uganda,  as well as in Tanzania and Bangladesh where difference-in-difference and instrumental variable estimation strategies were used respectively. 

Mobile money can facilitate higher savings for households

Mobile money can also be helpful for financial resilience as it facilitates saving. In Kenya, mobile money accounts for women were used to easily (and securely) allocate and label money for savings. The researchers found that labeling an account, such as encouraging women to use the account for emergency expenses and savings, along with a one-on-one activity eliciting savings goals and weekly SMS reminders on the savings goals especially increased the amount saved. This amount accrued helped women respond to unplanned expenses and made them less reliant on existing networks for support. 

Also in Kenya, researchers used instrumental variables to find that mobile money use led households to save more. Households with mobile money accounts were 16–22 percent more likely to save and their average household savings increased by 15–21 percent, the equivalent of US$2.7 to US$3.7 per month.

Mobile money facilitates transparency and formalization

Mobile money electronically records all transactions, which improves the security of payments as well as their transparency, the consequences of which could be far-reaching on the economy. Greater transparency of earnings, transactions, and remittances could greatly improve tax collection. A mature mobile money system could also foster formalization of the economy, integrating informal sector users into formal banking and insurance, and allowing for stronger links to the government through social protection schemes, tax collection, and other government programs.

In the long run, mobile money can affect occupation decisions 

Mobile money and the ability to easily and safely receive money from social networks at a distance has been found to change the way that individuals within households make decisions around their occupation. In Kenya, researchers used a difference-in-difference methodology to examine the long-run effects of M-PESA usage and found that higher access to mobile money changed occupation choices, especially among women. The study estimated that approximately 185,000 women moved from agriculture to small-scale retail as a result of access to M-PESA. 

Similarly, access to mobile money led to a shift from farm-based work to self-employment in Uganda and migration from rural to urban areas where the income is higher in Mozambique. The latter was because mobile money increased individuals’ trust that they could easily and safely remit money to their families in the rural areas. 

Long-term effects on poverty and women

Non-RCT research from Kenya found that access to mobile money increased per capita household consumption and savings, and therefore reduced the rate of poverty. This increased consumption translated to a movement of 196,000 households out of extreme poverty—equivalent to 2 percent of all households in Kenya. Long-term effects are examined in this study by running a follow-up survey in 2014 for households that saw relatively large increases in agent access between 2008 and 2010. The effects were largest in female-headed households, which highlights how impact can be amplified when technology is given to female household leaders. The researchers posit that mobile money could give women in male-headed households, who are also usually secondary income earners, more financial independence. Agent density—the number of agents operating in a specific area—played a key role. Increased agent density was linked to 3 percent of women in both female- and male-headed households taking up business or retail occupations over-farming.

While mobile money holds promise, it is not a silver bullet

As described above, the long-run effects of mobile money on poverty reduction, and the consistent cross-context findings of its ability to mitigate the effects of negative events, are impressive. However, there are a number of pre-conditions required for these impacts to be achieved. These factors include: 

• Developing a strong profitable network of agents to deliver the mobile money system, including cash-out services;
• Making the registration/use of the technology simple and trouble-free for the consumer;
• Investing in infrastructure to scale from the start; and
• Creating a strong regulatory environment once the technology has been developed. 

At a more fundamental level, individuals must have access to mobile devices and they must understand how to use the mobile money product. For example, in the roll-out of mobile money in Mozambique, information was disseminated to individuals and agents in an intense mobilization process. This is seen as an important part of the success of the program implementation. 

With all the promise that mobile money holds, it is not without its challenges and risks. 

• Infrastructure challenges: Mobile money requires telecom infrastructure which can require significant up-front investment. A lack of broadband infrastructure on the continent—25 percent coverage as of 2018—means there will be limited integration with other digital financial products. Mobile money is often reliant on SMS-based technology which makes it difficult to link mobile money with internet-based digital financial products. 
• Operational risks: Evidence also suggests that mobile money opens avenues for fraud. Research shows many mobile money companies are vulnerable to data breaches due to improperly encrypted communications, potentially allowing an attacker to steal money. 
• Regulatory challenges: Finally, many policymakers and private sector companies still struggle with how to regulate mobile money to protect against these operational risks, encourage its use, and build a network of agents. There are also unanswered questions around how to design mobile money platforms and networks, link these to digital identification systems, and leverage the spread of mobile money to create stronger financial markets.

Open questions

So far, the use of mobile money has been concentrated in the person-to-person (P2P) payments space as well as in certain regions/countries. Person-to-business payments, business-to-business, and government-to-business payments are rarely transferred through mobile money. In addition, even for countries with a high mobile money penetration in the P2P space, there is limited evidence on whether it has transformed financial markets or the use of physical cash. 

Although there has been significant work on the impacts of mobile money on households, and some work on the impacts on microenterprises, questions remain around how to increase the take-up of mobile money in countries with low penetration. The possible benefits for governments to use mobile money for person-to-government (P2G) payments, for example, as a tool for tax collection, also remains largely unexplored. While there is a large potential to use mobile money for P2G payments, this still needs to be rigorously explored. You can read more about this and many more unexplored policy and research questions in the Digital Identification and Finance Initiative’s (DigiFI)’s framing paper.

Authors Note: This is the third blog in the DigiFI series on the various aspects of their research and policy priorities. The next blog will focus on one of the barriers to public service delivery, namely targeting.  

The COVID-19 pandemic and the associated economic restrictions are likely to result in an increase in poverty for the first time in the past 20 years. This has catalyzed a worldwide increase in social protection measures—such as government cash grants and other public measures to ensure a basic level of welfare. As of May 2020, 190 countries/territories had planned, launched, or modified their social protection safety net programs, often cash-based, in response to the pandemic. These reforms have highlighted a growing need for the expansion or adoption of digital government social protection payments, also known as government-to-person (G2P) payments, to better serve their constituents. Decision makers have looked to digital payment tools during the pandemic to decrease the implementation challenges created through in-person physical cash distribution. 

Digitization allows governments to send money without having to transfer physical cash and has proven particularly useful during the pandemic to reduce in-person contact and limit the spread of disease, while maintaining critical economic relief measures. Many countries in Africa had started digitizing government transfers prior to the pandemic. South Africa was one of the first countries to do so by introducing a biometric identification (ID) and debit card grant payment disbursement system. With the spread of mobile money, particularly in East Africa, governments, and large nonprofits such as GiveDirectly have been using mobile money to distribute cash grants to beneficiaries. In addition, many sub-Saharan Africa countries have or are considering rolling out digital ID systems and coupling them with social protection programs to uniquely identify individuals and provide a system for authentication of the beneficiaries. 

The digitization of G2P payments can occur at different stages of the payment chain: selection of beneficiaries, such as through digital IDs, management of payment systems through digitized, high-frequency data collection, or money distribution, for example using mobile money or direct bank transfers. Many countries are aiming to digitize their economies in the hope that this will allow them to centralize systems, make them more efficient, and reach a wider range of beneficiaries, among other benefits. While there is some relevant evidence on the merits and weaknesses of digitization G2P payments, there is still a great deal left to be studied and understood. 

What are the possible effects of the digitization of G2P payments on beneficiaries? 

Digitization of social protection programs could potentially transform the relationship between citizens and the state, if it allowed for more effective distribution, including: 

Reduced cost to collect social protection payment: The digitization of payments can reduce the costs of payment collection for the recipient. These costs include not having to wait in line to collect the money, along with a saving on any transport cost. In addition, there may be tangible impacts of the time/money saved for other individuals within the household. A study in Niger found that households who received mobile transfers (as opposed to traditional cash transfers) had greater diet diversity and that their children ate more food. These impacts are likely due to reduced time and travel costs to obtain their transfers, as well as increased decision-making power for women. 

Reduced delays and uncertainty in payments: When the existing infrastructure is strong, digital payments can reduce delays and uncertainty of G2P payments. Physical cash is difficult to transport over long distances. There may be delays in the delivery of the money due to weather or road infrastructure, especially in the rainy season. Not only is the digital transfer beneficial to reduce the stress of not knowing if the next payment will arrive, but the improved predictability and reliability of payments could allow individuals and households to plan and invest with a longer-term view.

Improved access to financial products and services: Digitized G2P payments could lead to greater financial inclusion through improved access to other financial products, such as loans and savings products. It is argued that once a bank or mobile money account is opened for the G2P payments, there is a higher likelihood that individuals would have access to other financial products. G2P payments are thought to be a gateway for financial inclusion, however more research would need to be conducted to support this claim.

More flexible payment modalities: Digitization allows for increased flexibility of G2P payment modalities in terms of the value of the payment and the timing. Digitized payments can more easily be made in lump sums or incrementally over time. Payment timing of when the money arrives can be adjusted more easily in digital payment systems compared with a physical cash-based form. Individuals may have a preference for smaller payments over time to limit their spending, or to have it arrive as a lump sum on a certain date to coincide with large payments, such as school fees. In addition, different schemes can be applied on top of the G2P payment to encourage healthy financial behaviors. For example, the labeling of accounts, which is easier to implement digitally than when in physical form, has shown to have positive effects on long-term saving behavior as shown through this study.

However, technology is but a tool and, if not used properly, could instead exacerbate the existing problems that beneficiaries face, or create new ones:

Exclusion of beneficiaries: If beneficiaries are not able to navigate the system of digital G2P payments, they may not be able to access their grants even when eligible, as seen in Jharkhand, India. In India, the use of Aadhaar—a digital biometric identification card—as a digital tool to authenticate beneficiaries, reduced the benefits for those who had not registered for an ID. Given that more vulnerable people are often also those who have less access to knowledge on technology, this could mean the most vulnerable are particularly susceptible to exclusion when shifting to a digital payment mechanism. 

Predatory systems that leverage lack of digital (financial) literacy: The digital systems could undermine the intention of the cash transfer if poorly structured or have predatory financial inclusion elements. G2P payments can facilitate connection with other financial products; however, without adequate digital financial literacy, beneficiaries could be led into unwittingly connecting their accounts to financial products they do not require. As with the exclusion concern (above), this issue is particularly important as those who are the most vulnerable groups of people are also more likely to not have access to digital financial training. For example, in South Africa, Torkelson notes under Net 1’s biometric system, social protection beneficiaries had to place their fingers on multiple scanners without understanding why. This process, along with broader dubious practices and mass public outcry, led Net 1 to be charged at the Constitutional Court.  

Privacy concerns: There may be privacy concerns when digitally collecting beneficiary data through the G2P program. Once the G2P program moves towards a digital distribution, the ease of surveillance is far greater. Beneficiaries who receive G2P payments may not fully comprehend the value of their data and the ways in which it could be used in a digital format. There are ethical questions around informed consent, linking of databases with other departments, and the ability for governments to discriminate based on the digital information.  In countries where the program has to roll out quickly, before a strong privacy framework is available and enforceable, is a particular concern.

What are the effects of digitization of G2P payments on Government institutions? 

Individuals who receive digital G2P payments are not the only group to benefit, governments who implement the systems could also realize positive gains from these reforms. Digital payments and IDs may help overcome systemic challenges, along with other implementation benefits through moving to a digital platform. 

Reduction in leakages: In many low- and middle-income countries, pervasive corruption and tax evasion can undermine the provision of public programs, causing leakages and other service delivery delays and failures, and reduce the fiscal capacity of government agencies. For example, a 1996 Public Expenditure Tracking Surveys (PETS) survey of public schools in Uganda showed a leakage rate of 87 percent in Uganda’s education spending, which fell to 20 percent after the publication of the findings: in Tanzania, a 1997 PETS found 57 percent leakage in education and 41 percent in health spending; and in Ghana, a 2000 PETS found 50 percent leakage in education and 80 percent leakage in health. Evidence from India shows that digitization has the potential to reduce discretion and increase communication, thereby reducing corruption and leakages in program delivery.

Digital G2P payments could also help reduce quantity fraud. This type of fraud is slightly more difficult to monitor, but involves the beneficiary receiving a smaller quantity than the true eligible amount. Digital payments can assist with this type of fraud as the money moves directly to the beneficiary's account and limits the ability for middle-men to siphon off part of the payment. 

Reduced implementation costs: Manual cash payments are cumbersome and costly to distribute. In some contexts, cash-based social protection programs are delivered by mobile vehicles moving between villages. These benefits can be realized if there is a strong existing digital infrastructure in the country. The study in Niger considers the cost of implementing a mobile transfer. Although the initial cost of the mobile transfer program was higher, due to the purchase of mobile phones, its pre-transfer cost was approximately 20 percent lower than that of physical cash distribution. This suggests that, once phones and mobile money agents are in place, mobile transfers could be a simple and low-cost way to deliver payments.

More effective monitoring: Further, digitization of payments can lead to more effective process monitoring, which allows implementers the ability to make real-time decisions based on incoming data. In a context where there is a reliable identification system in place, digital data collection can make policy decision making easier and more responsive to the needs of the people on the ground.

In addition, the improved monitoring and data collected on the beneficiaries could assist with improved targeting. G2P payments with a strong digital ID could also help to identify who is eligible from an objective perspective. A digital footprint can assist with linking individual’s information across departments, which could allow for a better picture of the individuals, informing who is the most vulnerable and in need of assistance. 

Although there are many possible benefits of a digitized G2P system, there are also potential political economy challenges or pitfalls that governments may experience: While there is limited evidence on the impacts of digital G2P payments on governments, a study in India highlights how a direct bank transfer of a subsidy (in place for an in-kind subsidy) led to less diversion of the subsidized goods to the black market, reducing leakages. Despite these positive outcomes, this reform was abruptly terminated after political lobbying leading to the elections, highlighting the importance of political economy challenges with such policy reforms. Politically connected agents may lobby against, and may even succeed in subverting policy reforms when their discretionary (and often illegitimate) benefits are threatened. Further, if the digital payment system is badly implemented and leads to exclusion of beneficiaries or mistrust in the data collection and use, this could harm the long-term credibility of the government and associated institutions. 

Open questions

While there has been much progress on the spread of digital payments across Africa, we still have little evidence about how to effectively design policies to maximize benefits from digital G2P payments. The Digital Identification and Finance Initiative in Africa (DigiFI)—hosted by J-PAL Africa at the University of Cape Town—aims to fill this evidence gap by generating cutting-edge policy research projects focussed on the study of innovative government payment and ID systems. We lay out our research agenda in our framing paper. 

Authors Note: This is the second blog in the DigiFI series on the various aspects of their research and policy priorities. The next blog will focus on mobile money.