Why 1.3 Billion Adults Still Can't Open a Bank Account

There’s a comfortable story about financial exclusion: the unbanked are very poor, and as economies grow they’ll be brought into the system. It’s partly true and mostly beside the point. The more useful question is why a bank account, a database row and some messaging is uneconomic to provide to roughly a fifth of the world’s adults.

How many adults are unbanked, and what the number actually measures

The authoritative figure comes from the World Bank’s Global Findex, a household survey run roughly every three years. The 2025 edition based on interviews with more than 145,000 adults across 141 economies during 2024, puts the number of adults without an account at a bank, credit union, microfinance institution or mobile money provider at approximately 1.3 billion. Account ownership now stands at 79% of adults globally.

Three things are worth noting before that number gets used as a rhetorical prop.

First, it’s falling, and quickly. It was around 2.5 billion in 2011, 1.7 billion in 2017 and 1.4 billion in 2021. Whatever is happening, it is not stagnation, and any honest account has to explain the improvement as well as the shortfall.

Second, it’s concentrated. More than half the world’s unbanked adults roughly 650 million people live in eight economies: Bangladesh, China, Egypt, India, Indonesia, Mexico, Nigeria and Pakistan. That is a meaningfully different problem from a billion people thinly distributed everywhere, and it changes what a serious intervention would look like.

Third, “having an account” is a low bar. It counts a dormant account with a zero balance. Findex separately tracks inactive accounts, and in some countries a substantial share of accounts see no deposit or withdrawal in a year. The population that is formally banked but functionally excluded is considerably larger than the headline suggests.

The reasons people give

The survey asks unbanked adults why they don’t have an account, and lets them give more than one reason. The answers cluster consistently:

• Not enough money to make an account worthwhile which was by far the most common answer
• Cost fees, minimum balances
• Distance the nearest branch is too far
• Documentation no ID, proof of address, or employment record of the kind required
• Trust in the institution, or in the system generally
• Someone else in the family has one so the household is served through a relative

Look at that list as a product designer rather than as an economist. “Not enough money” is not a statement about poverty in the abstract. It’s a statement that at this balance, the account is not worth what it costs me. That’s a pricing failure. So is distance, so is cost, and documentation is a compliance requirement rather than a technical one.

Why the account itself is expensive

Here’s the part that rarely gets said plainly: a bank account has a substantial fixed cost per customer per year, and that cost barely moves with the balance.

Onboarding requires identity verification against documents many people don’t hold. The account is then subject to ongoing anti-money-laundering monitoring, sanctions screening, dormancy handling, statementing, dispute handling and regulatory reporting. Branch or agent networks have to be maintained to serve cash-based customers. Card issuance and replacement cost real money.

None of that scales down. Serving a customer with a $40 average balance costs a bank something not wildly different from serving one with $40,000 but only one of them generates enough net interest income and fee revenue to cover it.

So banks do the rational thing: minimum balances, monthly fees, or simply no product for that segment. The customer experiences this as exclusion. The bank experiences it as arithmetic. Neither party is behaving irrationally, which is precisely why exhortation hasn’t fixed it.

What mobile money actually proved

The clearest counter-example is mobile money, and it’s worth being specific about what it demonstrated.

By stripping the product down, a phone number as the account identifier, an agent network instead of branches, no card, no credit, providers got the per-customer cost low enough that a low-balance customer became viable. Account ownership in Sub-Saharan Africa rose substantially on the back of it, and mobile money accounts are the main reason the global unbanked figure fell as fast as it did.

That’s a genuine achievement and it settles the argument about whether the excluded population “wants” financial services. It plainly does, when the product fits.

But mobile money has two structural limits. It is custodial: your balance is a claim on a provider, which can be frozen, and which depends on that provider’s solvency and the regulator’s posture. And it is national: systems are licensed per country and largely don’t interoperate across borders, which is exactly where the need is sharpest.

Cross-border is where the cost is worst

Remittances are the cleanest illustration of what the current system charges for the least sophisticated service it offers.

The World Bank’s Remittance Prices Worldwide tracks the average cost of sending $200 across borders. It stood at 6.36% globally in Q3 2025, more than double the UN Sustainable Development Goal target of 3%, and the spread underneath that average is wide. Banks are the most expensive channel by a distance at around 15%, while digital services average 4.6% against 7.3% for non-digital ones. Sub-Saharan Africa remains the most expensive receiving region: of the thirteen corridors that cost more than 20%, nine originate there.

More than six per cent to move a number from one ledger to another, on a route that hundreds of thousands of people travel every month. That the digital channel already runs well below the average is itself the point: the cost is a property of the plumbing, not of the distance. The cost isn’t computational. It’s the chain of correspondent banks, foreign-exchange spreads, compliance duplicated at each hop, and cash-out agent margin seach layer taking a cut for a service that, technically, is a message.

What changes if the account has no marginal cost

Now consider the same problem with a different architecture.

A self-custodial wallet has no per-customer cost to the provider, because there is no customer account to maintain. There is no balance held on anyone’s behalf, no minimum, no dormancy, no statementing. The user holds their own keys; the network settles.

That single change removes most of what made the low-balance customer uneconomic. There is no threshold below which serving them stops making sense, because there is nothing being served. The software is the same software whether the balance is $4 or $4 million.

Settlement cost stops scaling with amount, too. On Solana, a transfer costs a fraction of a cent regardless of whether it moves $10 or $10,000, and confirms in well under a second. Sending $200 across a border for a fraction of a cent rather than six and a half per cent isn’t an incremental improvement to remittances; it removes the category.

What a non-custodial wallet actually is

Who holds the keys, what the provider can and cannot do, and where the real trade-offs sit.

→

The honest caveats

This is where crypto arguments usually overreach, so let’s be careful.

Self-custody moves the burden rather than removing it. No provider can freeze your funds, and no provider can help you if you lose access. Historically that meant a seed phrase which, for precisely the population under discussion, is a worse failure mode than a bank’s paperwork. Any serious claim about financial inclusion has to come with a recovery model that doesn’t assume permanent custody of twelve words.

Volatility is real. Someone earning daily wages cannot absorb a 20% drawdown in their savings. Holding an asset that moves against you is not obviously better than holding an inflating local currency, and often worse. Stable value is a prerequisite, not a detail.

On- and off-ramps are still the bottleneck. Getting cash into and out of the system remains a licensed, local, agent-dependent business the same problem mobile money solved with feet on the ground. Cheap settlement doesn’t help if the first and last mile are still expensive.

Regulation is not an obstacle to route around. Anti-money-laundering rules exist for reasons, and a system that serves excluded populations at scale will have to engage with them rather than pretend they don’t apply.

Where this leaves things

The unbanked are not a demographic waiting to be uplifted into a system that already works. They are evidence that the system has a cost floor, and that a large fraction of humanity lives below it.

Every intervention that has genuinely moved the number, mobile money above all worked by attacking that floor rather than by extending goodwill. The interesting property of self-custodial settlement is that it takes the floor to roughly zero: no account to maintain, no balance minimum, no per-customer compliance overhead, and a transfer fee that doesn’t care about the amount or the border.

That’s necessary. It is not close to sufficient. Getting from cheap settlement to actual inclusion means solving recovery for people who cannot manage a seed phrase, stable value for people who cannot absorb volatility, and cash access for people who are paid in it.

That’s the harder half of the problem, and it’s the half we’re working on.