NextBillion - A WDI Publication

In many low- and middle-income economies, access to credit is limited not by a lack of economic activity, but by how that activity is evaluated. Individuals and small enterprises engage in informal economies on a regular basis, producing income, managing cash flows and fulfilling financial responsibilities. Nonetheless, they are excluded from formal credit systems, since these systems rely on data that is frequently missing in informal situations.

This situation has contributed to a global financing gap for micro, small and medium-sized enterprises (MSMEs) that is estimated in the trillions (as of 2018), rising significantly when informal enterprises are included — numbers that highlight the scale at which economically active but undocumented businesses remain underserved by traditional credit systems. The problem is not that these borrowers exhibit poor financial behaviors, and it’s not even that there’s a lack of data about their financial activities. It’s that credit assessment systems aren’t designed to capture and interpret the data that’s generated by their financial decisions.

using alternative data to boost financial inclusion

A common misperception about financial inclusion is that underprivileged groups don’t generate useful data. In practice, informal economies generate a diverse set of economic signals: These include mobile money transactions, airtime and utility payments, merchant activities, and platform-based transactions (i.e., e-commerce and social commerce). According to research conducted by the Global Partnership for Financial Inclusion, alternative data from cellphone usage, transaction histories and other elements of a customer’s digital footprint can increase loan access when integrated into risk assessment models. However, it can be challenging for lenders to acquire access to this data — and to establish how it fits into their existing credit system. They typically need to establish (often-costly) partnerships with third party providers to obtain this data, and these providers still may lack data on many of the customers they’re hoping to serve.

Additionally, most credit systems are built around structured, standardized data inputs such as formal credit bureau records, bank transaction histories, documented income and collateral documentation. Informal market data, in contrast, is typically spread over multiple lending, payment and e-commerce platforms, varying in format and frequency, difficult to verify, and only loosely linked to formal identity systems. As a result, even when data is available, it is challenging to employ in traditional credit models.

Fintech innovation has attempted to overcome this gap by incorporating alternative data based on behavioral and transactional criteria into credit assessments, and in some cases, this has allowed first-time borrowers to acquire credit. However, these approaches are often layered onto traditional credit models rather than replacing them. The underlying systems built around fixed credit scores, formal identity verification and stable income assumptions remain unchanged. As a result, while these models improve visibility into borrower behavior, they do not fundamentally change how creditworthiness is assessed. Without rethinking how systems interpret fragmented and irregular data, access improves at the margins, but it won’t completely solve the problem of getting credit to previously excluded borrowers.

Building a Credit Assessment System for Informal Markets

While working as a product manager at Qore, a banking-as-a-platform solution provider that enables lenders to reach excluded borrowers, I was involved in building Recova, a credit assessment and decisioning system designed to support lending in markets with limited formal financial data. This challenge is particularly evident in markets like Nigeria, where a large share of adults remain excluded from formal credit despite active economic participation. According to World Bank research, a key barrier to reaching these customers is their lack of reliable credit histories and formal financial records, which makes is difficult for borrowers to be assessed using traditional credit models.

Recova had the same limitation as other credit assessment systems, as many of its borrowers didn’t have standard credit histories. But their financial behaviors were easy to see in their mobile money transactions and payment activity. To understand and assess these behaviors, the system added: transaction-based indications from digital payment activity; rule-based risk frameworks that could function with insufficient data; and flexible scoring models that let lenders change risk limits depending on the situation.

However, in building Recova, we learned that for many lenders, the key challenges in serving these borrowers did not involve data input, but rather system architecture. We realized that this architecture needed to address three key issues:

• Identity Fragmentation: Borrowers often operated across multiple financial channels — such as cash, mobile money and informal networks — without a formal identity. This made it difficult to consolidate this financial behavior into a borrower’s profile.
• Data Reliability: Available data was often incomplete or inconsistent, and lending systems could not assume its accuracy, which required decision models that could function despite this uncertainty.
• Risk Interpretation: Traditional lending models interpret income variability as risk, but this variability is common in informal economies. As a result, the system needed to distinguish between actual financial instability and expected fluctuations in income and behavior.

Recova did not totally address these issues, because they reflect structural constraints that cannot be resolved within the scope of the product. However, it continues to successfully serve lenders across Africa by enabling credit decisioning in informal markets through flexible, context-aware algorithms that take these borrowers’ circumstances into account, rather than simply feeding their data into the lender’s established credit scoring models.

Credit Access in Informal Economies Is a System Design Challenge

Building Recova brought to light the fundamental problem with the way current lending systems are structured: Making it easier for people in informal economies to get credit isn’t just about getting more data; it’s also about changing how credit assessment systems obtain and understand the transactional data generated by informal borrowers. To tackle this, the industry needs to stop thinking about lending as a simple data problem and start thinking of it as a layered system design problem, where several constraints must be handled by the various layers of the credit assessment system at the same time. These constraints include:

• The Data Layer: The challenges facing this layer aren’t defined by a lack of resources, but by fragmentation and inconsistency. Credit assessment systems in informal economies leverage financial signals from mobile wallets, bank accounts, informal savings clubs and transactions with merchants, but these signals don’t always match the system’s requirements in terms of data structure or frequency. This makes it difficult for traditional systems, which rely on clean and standardized inputs, to interpret them effectively. As a result, simply aggregating more data does not improve decision-making, unless the system can reconstruct meaningful patterns in a borrower’s financial behavior from incomplete and irregular inputs. What is required is a shift toward systems that: actively interpret data to identify borrowers’ behavioral consistency; adapt to changing inputs by incorporating alternative data rather than just relying on a fixed data set; and attempt to make sense of partial financial histories rather than rejecting these applicants due to lack of data.
• The Identification Layer: This layer presents a greater challenge than the data layer, since financial behavior cannot be effectively utilized for credit assessment without a trustworthy method of linking this data to individuals or businesses. In many informal markets, individuals use various IDs, like phone numbers, wallets or unofficial business names, which are not consistently connected. This fragmentation hampers the creation of relevant credit histories and leads to the repeated reassessment of the same individuals as if they were new entries. To tackle this issue, we need systems that move beyond static identity verification and instead gradually establish identity by associating behavioral patterns and transaction histories with individual borrowers over time. This would allow them to develop credible financial identities through usage, rather than solely through documentation.
• The Trust Layer: In informal economies, trust is established in a fundamentally different way than in formal markets. Instead of being verified by formal institutions, it is built through relationships, frequent interactions and community validation. This difference is often missed by traditional credit systems, since they focus only on signals that can be legally verified and acknowledged by institutions. As a result, they’re overlooking a large group of potential borrowers whose dependability is often showcased through consistent conduct in small-scale transactions, engagement within local networks, and informal recommendations. To integrate these alternative markers of trust into credit systems, it is necessary to transform social and behavioral trust into structured inputs without oversimplifying the process, thus effectively connecting relational credibility with formal financial evaluation.
• The Risk Layer: Conventional lending models struggle to adjust their risk assessment processes to include informal borrowers. The basis of traditional credit systems is reliability, which borrowers demonstrate through consistent financial activities and stable earnings. In contrast, borrowers in informal economies exhibit greater instability, with income sources that fluctuate due to external disturbances, market trends and seasonal changes. While this variability is often viewed as a marker of significant risk, it can also suggest resilience and flexibility. A more effective approach is to substitute fixed stability measures with dynamic assessments that examine how individuals adapt to changes over time, recognizing patterns of long-term reliability, consistency and resilience despite short-term fluctuations.
• The Incentive Layer: This layer ultimately determines whether any of these systems work in practice. For credit systems to be effective in informal markets, all players must perceive obvious benefits in operating and utilizing these systems. Borrowers need confirmation that supplying their personal financial information will lead to significant access to credit; lenders must believe that the signals they get from borrowers are trustworthy; and intermediaries like fintech platforms and data providers/aggregators need to have a reason to offer accurate and consistent information. When these incentives don’t line up, the quality of the data deteriorates and adoption slows, putting the whole system at risk. Building an effective credit assessment infrastructure requires lenders to incorporate real advantages for all stakeholders into the system, and to be open to feedback on how well these incentives are working. If they don’t, people are less likely to keep using the system over time.

Understanding the challenges inherent to each of these layers leads to a critical insight: Credit systems in informal economies cannot be designed as linear pipelines. They must operate as adaptive systems, in which data informs identification, identification enhances trust, trust determines risk, and risk decisions reinforce incentives. This interwoven design reflects the real economic activity in informal markets.

Some aspects of this design strategy are already evident in digital lending and mobile financial ecosystems. However, they are frequently deployed as isolated data upgrades without identity resolution, or updated risk models without trust integration. The creation of a more functional approach depends on greater system-level coherence: creating infrastructure that allows all the layers of a credit assessment model to grow together.

Why Financial Inclusion Requires More than Data

The next step in financial inclusion will be based less on how easy it is to get more borrower data, and more on how well systems are built to understand this data. There is plenty of financial activity and signaling in informal economies, but the data this activity produces is still not in sync with credit systems that were created with formality, uniformity and traditional assessment standards in mind. If these systems are not adapted, large groups of economically active people will continue to be excluded from access to formal credit — not because they are untrustworthy, but because the system can’t properly assess their trustworthiness.

To design a better credit assessment system for these markets, providers need to go beyond small fixes and take a more integrated approach, in which data, identity, trust, risk and incentives are all seen as interconnected layers of a single system. These layers must be brought into line with the realities of informal economic life before credit systems can start to accurately identify and support the people they have long excluded.

Ayokunmi Sodamola is a technology professional working at the intersection of AI, human-centered design and system architecture.