Where Banking Still Means Walking In
There is a country where the number of bank branches per capita grew more than 500% in a single decade. It’s not a post-conflict reconstruction story or a newly wealthy petrostate. It’s Ethiopia, where bank branch density rose from 2.9 per 100,000 adults in 2012 to 14.5 in 2023 — still one of the lowest densities on Earth, but expanding faster than anywhere else.
Meanwhile, Latvia lost 62% of its branches in the same period. Germany cut its network in half. The United States has been closing branches at a steady clip for a decade.
These opposite trajectories raise a question that most fintech commentary skips over: when does a country actually stop needing bank branches? The answer, the data suggests, is: later than you’d think, and only after several other things fall into place first.
The branch as bottleneck
In countries with low banking density, the branch isn’t just one channel among many — it’s the only channel. When Kenya has just 4.5 bank branches per 100,000 adults and Nigeria has 4.4, those physical locations are doing work that branches in Japan (33.7 per 100,000) or France (31.3) share with ATMs, mobile apps, and online banking.
The math is unforgiving. In a country with 5 branches per 100,000 adults, each branch serves roughly 22,000 adults. In Japan, each branch serves about 3,000. That difference means longer travel times, longer queues, and fewer services available per visit in the low-density country.
For context on how these access gaps connect to broader financial inclusion, see our state of financial inclusion report.
The sequence matters
Looking at countries that have successfully transitioned away from branch-dependent banking, a pattern emerges. The shift doesn’t happen all at once. It follows a rough sequence:
Stage 1: Build branches. Ethiopia is here. Branch density is climbing from near-zero, and account ownership is just 48.8%. Most adults still have no formal financial relationship at all. Physical branches are how you build the initial network.
Stage 2: Add ATMs. India illustrates this stage. With 14.5 branches and 25 ATMs per 100,000 adults, the ATM network is nearly twice as dense as the branch network — giving people cash access without requiring a teller. But digital payment adoption is only 48.5%, meaning most transactions still need physical infrastructure.
Stage 3: Digital substitution begins. Brazil is in early stage three. Branch density is 16.0 per 100,000, but ATM density has exploded to 111.0 — nearly seven ATMs for every branch. Mobile money has reached 58.2% of adults, and digital payments are at 77.4%. The branch is no longer the primary access point.
Stage 4: Branches become optional. Kenya has jumped ahead. With only 4.5 branches per 100,000 adults but 90.1% account ownership and 89.3% digital payment adoption, physical branches are almost irrelevant for most financial transactions. Mobile money at 87.5% has become the de facto banking infrastructure. Kenya is actually closing ATMs too — down 25% in recent years — because even cash withdrawal is moving to phones.
For a deeper look at how mobile money replaces physical infrastructure, see our global mobile money report and top 10 mobile money countries.
Why branches don’t just disappear
The assumption in fintech is that digital always wins. But the data shows branches persisting even in wealthy, highly digital economies. Japan has 95.8% digital payment adoption and still maintains 33.7 branches per 100,000 adults — the 12th-highest density in the world. Cash is still king in Japan, and the cultural preference for in-person banking keeps branches viable.
Switzerland ranks 10th globally with 34.1 branches per 100,000 adults, despite being one of the world’s most digitally advanced economies. High-income doesn’t automatically mean branchless.
The countries that are closing branches fastest tend to be mid-income European economies where digital banking arrived quickly and branches were expensive to maintain: Latvia, Netherlands, Sweden, and Belgium have all cut branch networks by 30–62% in a decade. For more on the debit infrastructure that replaced branches in these countries, see the debit card divide.
The ATM as transitional technology
ATMs occupy an interesting middle ground. They’re physical infrastructure, but they serve a narrower function than branches — mostly cash withdrawal. And the data shows ATM networks growing even as branches shrink, suggesting ATMs serve as a bridge between branch-dependent and digital banking.
Canada has 188.5 ATMs per 100,000 adults — the 4th highest in the world — while its branch density is a modest 19.0. That’s nearly 10 ATMs for every branch. Portugal has 161.2 ATMs per 100,000 adults against 29.2 branches — a 5.5-to-1 ratio. Brazil maintains 111 ATMs per 100,000 adults despite having only 16 branches.
But in countries where digital payments have truly taken off, even ATMs are declining. Kenya ATM density fell 25% as M-Pesa made cash withdrawal less necessary. Sweden cut ATMs by nearly 25% as Swish and card payments replaced cash. See our comparison of credit cards vs. mobile money for how different payment instruments serve different markets.
The access gap in numbers
The gap between branch-rich and branch-poor countries isn’t just about convenience — it shapes who can participate in the financial system. Consider:
| Country | Branches per 100k | Account ownership | Digital payments |
|---|---|---|---|
| Ethiopia | 14.5 | 48.8% | 20.7% |
| Nigeria | 4.4 | 63.3% | 54.5% |
| India | 14.5 | 89.0% | 48.5% |
| Kenya | 4.5 | 90.1% | 89.3% |
| Brazil | 16.0 | 86.4% | 77.4% |
| Japan | 33.7 | 98.5% | 95.8% |
| France | 31.3 | — | — |
Kenya breaks the pattern: low branch density but high inclusion, because mobile money substituted for physical infrastructure. Ethiopia shows the more typical case — low branches, low inclusion, low digital adoption. Without branches or digital alternatives, most adults are simply outside the financial system.
For more on how countries compare on banking access, see most-banked, least-banked.
What building-out countries should watch
For countries currently expanding their branch networks — Ethiopia, Nepal, Bolivia, Uzbekistan — the data from countries further along the path offers both encouragement and warnings:
Encouragement: Physical branches work. Bolivia expanded from ~36 to 62.6 branches per 100,000 adults and now ranks 3rd globally. The investment in physical access built a foundation for broader financial inclusion.
Warning: Branches are expensive to maintain and quick to become stranded assets. Germany went from 14.6 to 6.6 branches per 100,000 in nine years. Banks that invested in branches in the 2000s are now paying to close them. The countries building today need to plan for the eventual shift to digital.
The Kenya model: The most efficient path may be to build branches just enough to establish trust and basic access, then invest heavily in digital infrastructure. Kenya never built a dense branch network — it built M-Pesa instead, and achieved 90% account ownership with fewer than 5 branches per 100,000 adults. For more on how digital wages are accelerating this transition, see where salaries are paid digitally.
The bottom line
Physical banking infrastructure isn’t dying everywhere — it’s dying somewhere. In 110 of 172 countries with sufficient data, branch density is declining. In 32, it’s growing. The countries still building branches aren’t behind the curve; they’re on a different part of it.
The question isn’t whether branches will eventually become unnecessary. It’s whether countries can build enough access — physical or digital — before the window for traditional branch banking closes. For Ethiopia, with 48.8% of adults still unbanked, the answer may determine whether the next billion adults enter the financial system through a door or through a screen.
See our full global rankings of bank branches and ATMs per capita for the complete data.
Sources & method
Bank branch and ATM density data comes from the World Bank World Development Indicators, measured per 100,000 adults. Account ownership and digital payment adoption figures come from the World Bank Global Findex Database and mobile money data from the GSMA Mobile Money dataset, as compiled in FinStatGlobe’s derived datasets. Data years vary by country and indicator — most branch and ATM data is from 2022–2023, while Findex data is from 2024. Trend analysis uses five-year averages to smooth year-to-year volatility. See our methodology page for full details on data processing and quality checks.