The Future of Financial Inclusion: Building a World Where Money Works for Everyone

When I first started working in financial data strategy at JOYFUL CAPITAL, I thought financial inclusion was mostly a policy topic — something governments and development banks worried about. Then I spent three weeks in a rural county in Guizhou, watching a vegetable farmer named Auntie Liu try to get a small loan to buy a greenhouse. She had a smartphone, a clean repayment history with a village credit cooperative, and a stable income stream. What she didn't have was a formal credit score, so the bank's system simply couldn't see her. She wasn't "unbanked" in the classic sense. She was invisible to the algorithm.

That trip changed how I think about my job. Financial inclusion is not charity, and it is not just a regulatory checkbox. It is the discipline of making financial systems actually see people — their real behavior, their real cash flows, their real constraints — and then serving them profitably and responsibly. That is an engineering problem, a data problem, and a design problem all at once.

Today, roughly 1.4 billion adults worldwide remain outside the formal financial system, according to the World Bank's Global Findex. Another billion are "formally served" but effectively excluded, because the products they can access don't fit their lives — overdraft fees they can't avoid, minimum balances they can't maintain, loan sizes that don't match their seasonal income. The future of financial inclusion is about closing both gaps: getting the unbanked in, and making sure those who are already in are genuinely served.

What follows is my attempt to map where this is heading, from the perspective of someone who builds credit models and data pipelines for a living. Some of it is technical. Some of it is philosophical. All of it comes from real work, real mistakes, and real conversations with people like Auntie Liu.

Data Becomes the New Collateral

For most of modern banking history, collateral was the price of admission. If you had land, a house, or a factory, you could borrow. If you had none of those things, you were out. This is why traditional lending failed the poor so consistently: their wealth was held in things that banks couldn't easily seize or value — livestock, standing crops, informal receivables, social capital, future wage income.

Digital data is dissolving that constraint. Mobile money transaction histories, utility payments, supply-chain records, and even phone metadata now serve as functional collateral in many markets. In Kenya, M-PESA transaction trails have been used by lenders to underwrite microloans with default rates that sometimes rival those of conventional retail banks. In India, the Account Aggregator framework lets individuals share their financial data with lenders in a consent-driven, standardized way, turning a messy pile of bank statements into a machine-readable credit profile.

At JOYFUL CAPITAL, one of our portfolio companies runs a merchant cash advance product for small vendors in Southeast Asia. The underwriting model doesn't look at tax returns. It looks at point-of-sale frequency, basket size volatility, supplier payment regularity, and the time of day the shop opens. Vendors who consistently open at 6 a.m. and whose weekday sales pattern is stable are statistically good borrowers — not because someone wrote a rule, but because the model learned it from hundreds of thousands of observations. Behavior is the new balance sheet.

The catch, of course, is that data-driven underwriting can also reproduce historical exclusion. If a model is trained on past lending decisions, and those decisions were shaped by redlining, gender bias, or geographic discrimination, then the model will simply automate yesterday's unfairness. I've seen this firsthand. A model we were reviewing flagged a category of applicants as "high risk" — turns out the pattern was essentially "lives in a neighborhood that was historically underserved." We caught it. But the fact that it was there at all was a reminder: fairness in data-driven finance is not automatic. It has to be designed, audited, and re-audited.

There's also a subtler problem: data visibility isn't evenly distributed. A woman who runs a home-based business may have almost no digital footprint, while her male counterpart who runs a shop has POS receipts and supplier invoices. If we treat "data availability" as a neutral signal of creditworthiness, we systematically favor the already-connected. The future of financial inclusion is not just about collecting more data — it's about choosing data that doesn't merely reflect existing power structures.

I think the next big breakthrough here is what people in the industry are calling "thin-file modeling" combined with alternative data — not just transactional data, but psychometric data, mobile wallet patterns, and community references. Some of these approaches raise privacy concerns, and rightly so. But if we refuse to use any proxy for creditworthiness that isn't traditional, we've effectively decided that the excluded stay excluded. That's not a neutral choice either.

Mobile Money and the Agent Network

You can't build financial inclusion if people can't reach the financial system. In many rural areas of Africa, Asia, and Latin America, the nearest bank branch is hours away. But the nearest mobile money agent might be a five-minute walk. This is why agent networks and mobile wallets have done more for inclusion than any policy mandate in the last twenty years.

M-PESA in Kenya is the famous example: launched in 2007, it now processes transactions equivalent to a significant share of Kenya's GDP, and it brought millions of people into the formal financial system through something as simple as sending airtime and cash via SMS. But the less-discussed innovation was the agent network — small shopkeepers and kiosk owners who became the human interface for digital money. The agents earned commissions, the users got convenience, and the system got scale.

What I find interesting is that this model is now being replicated and adapted in very different contexts. In Bangladesh, bKash did something similar. In Brazil, Pix — the central bank's instant payment system — didn't rely on agents as much, but it achieved explosive adoption because it was free for individuals and interoperable by design. In India, UPI did the same thing at an even larger scale. The common thread is low friction and high trust.

But mobile money has its own inclusion challenges. Smartphone penetration is still limited in some markets. Agent networks can be exploitative if commissions are structured badly. And digital literacy gaps mean that the most vulnerable people are often the ones who need the most hand-holding — which is expensive. I remember a project in the Philippines where we trained agents to use a tablet app for onboarding. The technology worked fine. The problem was that many agents had never used a tablet before, and the training program was three hours long. We ended up redesigning the onboarding process to be voice-guided and icon-based. That redesign was more impactful than any model improvement we made that quarter.

The future here is not "everyone uses a smartphone app." It's a multi-channel, multi-interface financial system — agents, USSD, voice, cards, wallets, and yes, apps — where the channel adapts to the user, not the other way around. That's harder to build than a single polished app, but it's the only way to reach the last mile.

AI and the Credit Decision Engine

Artificial intelligence is often described as either the savior or the destroyer of financial inclusion. The truth is that it's a tool, and like any tool, its impact depends on how it's used.

On the positive side, AI is making it possible to underwrite people who were previously unscoreable. Machine learning models can ingest hundreds of variables — transaction timing, bill payment consistency, even the linguistic patterns in a loan application — and find signal where traditional models found only noise. At JOYFUL CAPITAL, we've seen models that improved approval rates for first-time borrowers by double digits while actually reducing default rates, simply because they captured behavioral patterns that a logistic regression with five variables couldn't.

On the negative side, AI can amplify bias, obscure decision logic, and create feedback loops that entrench exclusion. If a model denies a loan to someone because "people like them" historically defaulted, that person never gets the chance to prove the model wrong. The model's prediction becomes self-fulfilling. This is sometimes called the exclusion feedback loop, and it's one of the most serious risks in AI-driven finance.

The industry's response has been a mix of regulation and technical innovation. The EU's AI Act classifies credit scoring as high-risk, requiring transparency and human oversight. In the US, the Equal Credit Opportunity Act requires lenders to provide adverse action notices — a reason for denial — which is harder when the model is a deep neural network. Researchers are working on explainable AI techniques that can surface the "why" behind a decision without revealing proprietary model weights.

My personal take: explainability is not just a compliance requirement. It's a product feature. If a borrower is denied, and they understand why, they can change their behavior and reapply. If they just get a "no" with no explanation, they're more likely to give up or turn to predatory lenders. I've seen this in focus groups. The difference between "your application was declined" and "your application was declined because your income volatility is high; here are three things that could help" is enormous — not just for the borrower's experience, but for their financial trajectory.

Another area where AI is quietly making a difference is collections and customer service. Chatbots that speak local languages, that can explain a repayment schedule in simple terms, that can reschedule a payment without a human agent — these are not glamorous, but they reduce the friction that causes people to fall out of the financial system. And they do it at a cost that makes serving low-balance customers economically viable.

The future of AI in inclusion is not "replace human judgment." It's "augment human judgment at a scale that was never possible before." The banks that figure this out will have a real competitive advantage — not because they're being nice, but because they're reaching a market that their competitors can't see.

Digital ID and the Identity Layer

You can't include someone in the financial system if you don't know who they are. This sounds obvious, but roughly 850 million people worldwide lack any form of official identification. Without ID, you can't open a bank account, register a SIM card, or pass a basic KYC check. You are, in the eyes of the financial system, a ghost.

Digital identity systems are changing this. India's Aadhaar is the largest biometric ID program in the world, covering over a billion people. It has enabled e-KYC processes that reduced the cost of customer onboarding from dollars to cents, and it's a big reason why India's financial inclusion rates have jumped so dramatically. Estonia's e-Residency program, while different in purpose, shows how digital identity can be portable across borders.

But digital ID is not a simple good. It raises serious privacy concerns. A centralized biometric database is a tempting target for authoritarian abuse. Exclusion can happen at the enrollment stage — if a person's fingerprints don't scan, or if their name is spelled differently across documents, they can be locked out of the very system meant to include them. And in some countries, digital ID has been used to surveil and control populations, not empower them.

From a financial data perspective, the key insight is that identity is not a binary. A person can be identifiable enough for a small mobile wallet but not for a large loan. The future of inclusion is tiered identity — a graduated system where people can access basic services with basic ID, and unlock more as they build a verifiable history. This already exists in some markets through tiered KYC rules, but it's inconsistent and often poorly communicated.

I've also become convinced that decentralized identity — self-sovereign identity — is worth watching. Instead of a government or a bank holding your identity, you hold credentials on your phone and present them selectively. A lender could verify that you earn between X and Y without seeing your actual salary. This is technically feasible with zero-knowledge proofs, and it could be transformative. But it's early, and the user experience is still rough. The average person doesn't want to manage cryptographic keys. They want a wallet that works.

In the meantime, the practical work is less glamorous: making sure that ID systems have grievance mechanisms, that enrollment is accessible to people with disabilities, that names with apostrophes and diacritics don't break the system. I've spent hours on data cleanup for exactly these reasons. It's not the future — it's the necessary plumbing of the present.

Embedded Finance and Everyday Life

One of the most important shifts in financial inclusion is that finance is no longer a destination. It's becoming a feature embedded in other activities — buying seeds, paying for a ride, ordering food, receiving a salary.

Embedded finance means that a farmer doesn't have to go to a bank to get a loan. She can get one at the point of buying fertilizer, or at the moment she delivers her harvest to a buyer. A gig worker doesn't have to wait for a paycheck; he can get an advance on the income he's already earned. A small shop owner doesn't need a separate accounting system; her POS terminal automatically offers her a working capital line based on her sales.

This is powerful because it meets people in their existing workflows. The biggest barrier to financial inclusion is often not cost or trust — it's friction. If getting a loan requires a trip to a branch, a stack of documents, and a week of waiting, most people won't do it. If it requires tapping a button in an app they're already using, they will.

I saw this clearly in a project with an agricultural supply chain platform. Farmers were selling to a cooperative that paid them digitally. We embedded a credit product directly into the payment flow: after three successful delivery cycles, the farmer would see an offer for an input loan. The uptake was astonishing — not because the loan was cheap (it wasn't especially), but because it appeared exactly when the farmer needed it, with no extra effort.

But embedded finance also raises questions about consumer protection. When a loan is offered at the moment of a purchase, is the borrower really making a free choice? Or is the design exploiting a moment of need? This is where behavioral economics meets ethics. I don't have a clean answer. But I do think that transparency and cooling-off periods matter more in embedded contexts, not less.

The future of financial inclusion is not just about more products. It's about better-timed, better-contextualized products — financial services that show up when they're needed, in a form that makes sense, with terms that are clear. That's a design challenge as much as a data challenge.

Regulation as an Enabler, Not a Blocker

In my early years in this industry, I thought of regulation as the enemy of innovation. Now I think that's backwards. Good regulation is what makes inclusive finance possible at scale.

Consider open banking. The UK's Open Banking initiative, the EU's PSD2, India's Account Aggregator framework, Brazil's Open Finance — these are regulatory interventions that force banks to share customer data with third parties at the customer's request. Without them, fintech lenders would be stuck with whatever data they could scrape or buy. With them, they can build real underwriting models based on comprehensive financial histories.

Consider sandboxes. Regulatory sandboxes allow fintech companies to test new products with real customers under relaxed rules. This is how many mobile money and microinsurance products got their start. The regulator learns about the technology, the company learns about compliance, and the consumer gets access to something that might not have existed otherwise.

Consider tiered KYC. Instead of requiring full documentation for every account, regulators can allow simplified due diligence for low-value accounts. This lowers the barrier to entry while still managing money-laundering risk. It's a pragmatic compromise that has worked well in several African and Asian markets.

Of course, regulation can also be a blocker. Overly restrictive data protection rules can prevent beneficial data sharing. Inconsistent cross-border rules make it hard to serve migrant populations. And in some cases, regulators simply don't have the technical capacity to oversee new business models. I've seen projects stall for months because no one at the regulator understood what a "model risk" was.

What I've learned is that the most successful inclusion projects are ones where the regulator is a partner, not an adversary. That means early engagement, transparent communication, and a willingness to explain what the model does and how it's governed. It's slower. It's more work. But it produces durable results.

The future of regulation in this space is likely to be more principles-based and more outcomes-focused. Instead of prescribing exactly how a lender must assess creditworthiness, regulators will define what constitutes fair treatment and hold companies accountable for results. That's harder to enforce, but it's also more adaptable to rapid technological change.

Financial Health, Not Just Access

The final piece — and maybe the most important — is that access is not the same as health. You can have a bank account and still be financially fragile. You can have a loan and still be trapped in a debt cycle. You can have insurance and still be wiped out by a medical emergency.

Financial health is a broader concept: the ability to manage day-to-day finances, absorb shocks, and invest in the future. It's what financial inclusion is ultimately supposed to deliver. And by that measure, we have a long way to go.

The Center for Financial Services Innovation (now the Financial Health Network) has done pioneering work on measuring financial health. Their framework looks at spending, saving, borrowing, and planning. What they found is that a lot of people who are "included" are not healthy — they're just not excluded. They have accounts, but they're overdrafting. They have credit, but they're revolving. They have insurance, but they don't understand it.

This matters because it changes what we optimize for. If the goal is "more accounts opened," you get one set of behaviors. If the goal is "more people financially healthy," you get a different set. You start caring about product design, about financial coaching, about whether a loan actually improves someone's life or just shifts their problems around.

At JOYFUL CAPITAL, we've started incorporating financial health metrics into our portfolio monitoring. It's not easy. The data is noisy, the definitions are contested, and there's a real risk of measuring the wrong thing. But it's the right direction. Inclusion without health is just a bigger funnel into the same broken system.

The future of financial inclusion, then, is not just about who gets access. It's about what they get access to, and whether it actually helps them build a better life. That's a higher bar. But it's the bar that matters.

Conclusion: The Long Game

Financial inclusion is not a problem that will be "solved" in a single product cycle or a single policy window. It's a long, uneven process of making systems more visible, more accessible, and more fair. The tools are getting better — data, AI, mobile, digital ID, embedded finance — but tools alone don't determine outcomes. Choices do.

We can choose to build models that reproduce historical exclusion, or models that find new signals of creditworthiness. We can choose to design products that exploit moments of need, or products that genuinely help people manage their lives. We can choose to treat regulation as a burden, or as a scaffold for scale.

What excites me about this moment is that the technology is finally at a point where serving the underserved is not just a moral imperative — it's a viable business. The unit economics are improving. The data infrastructure is maturing. The regulatory frameworks are becoming more supportive. That combination doesn't come along often.

What worries me is that the window could close. If we build exclusionary systems now — biased models, exploitative products, surveillance-based identity — they'll be hard to unwind later. Path dependence is real. The choices we make in the next five years will shape the financial system for decades.

So the future of financial inclusion is not a prediction. It's a project. And it's one that people like me — people who sit at the intersection of data, AI, and capital — have a particular responsibility to get right. Not because we're the smartest people in the room, but because we're the ones building the invisible architecture that determines who gets seen and who doesn't.

Auntie Liu got her greenhouse, by the way. It took a while, and we had to build a custom model, and she still thinks I'm a bit strange for asking so many questions about her vegetable sales. But she's growing tomatoes now, and her daughter is in college. That's not a data point. That's a life. And it's the reason I do this work.

—

JOYFUL CAPITAL's Perspective on the Future of Financial Inclusion

At JOYFUL CAPITAL, we see financial inclusion not as a side project but as a core investment thesis. Our experience building data-driven credit models across emerging markets has taught us that inclusion and commercial viability are not opposites — they are complementary when designed correctly. We focus on three pillars: data that sees real behavior, AI that augments rather than automates judgment, and products that fit into people's existing lives rather than demanding they change their lives to fit the product. We believe the next decade will be defined by "inclusion infrastructure" — the identity, data-sharing, and risk-modeling layers that make it possible to serve the next billion profitably and fairly. We are actively investing in and building these layers, from alternative-data underwriting engines to embedded credit in agricultural supply chains. Our view is that the winners in this space will be those who can combine scale with empathy — who can build systems that are efficient enough to serve low-balance customers and transparent enough to earn their trust. Financial inclusion is not charity; it is the largest untapped market in the world. We are building for it, and we invite partners who share our conviction to join us.