85% Inactive: Ethiopia’s Digital Money Problem

Ethiopia’s digital finance numbers read like a success story: 157 million registered mobile money accounts, more than 35 non-bank digital financial service providers, and transaction volumes that grew more than 130-fold in six years, from 240 billion Br in 2020 to 33 trillion Br this fiscal year. But a figure the National Bank of Ethiopia disclosed alongside that growth cuts the story down to size: fewer than 15% of those accounts are actually active.

The gap is usually framed as an activation problem, something a better app or a cheaper fee structure fixes. The channel-by-channel breakdown suggests something narrower and harder to fix. Mobile banking, the channel that requires a smartphone, a bank relationship, and reliable connectivity, shows relatively higher usage. Mobile money and agent-based accounts, the channels built specifically to reach people without any of those things, show the lowest activation rates in the system. That is close to the inverse of what a financial inclusion strategy is supposed to produce.

It lines up with the other number the central bank put on the table: women in rural areas still travel an average of six hours to reach basic banking services. If agent networks and mobile money were closing that distance, activation on those channels should be climbing fastest, not lagging behind mobile banking. Instead, the pattern looks more consistent with an account base that skews toward people who already had some form of access, registered through KYC drives or salary and merchant disbursements, and quietly stopped using accounts that never solved a problem they had. The 157 million figure counts registrations, not reach.

That reframes what the transaction-volume growth is actually measuring. A 130-fold rise in six years is a real number, but concentrated activation means it likely reflects deepening use among an already-served, largely urban population rather than expansion of the financial system’s footprint. Growth in value moving through the system is not the same evidence of inclusion that growth in the number of people using it would be, and the NBE’s own data suggests the two have diverged.

The National Digital Payments Strategy 2.0 is explicit that the center of gravity is shifting from account growth to account use, and its 2030 Digital Public Infrastructure agenda, real-time payment rails, a national digital identity layer, and a data exchange layer, is the right infrastructure for that shift. But infrastructure does not travel the six hours on its own. None of the disclosed elements of Strategy 2.0, the open APIs, the fraud compensation fund, the digital currency, and tokenization work, directly targets agent density or rural liquidity, which is where the activation data suggests the actual failure sits. A national ID layer makes onboarding easier; it does not put a working agent within reach of a woman who currently walks half a day to find one.

The test for Strategy 2.0, then, is not whether the activation rate rises. It is whether it rises on the channels that were supposed to reach rural users in the first place. If mobile banking activation keeps climbing while agent-based accounts stay flat, the strategy will have improved the experience for people the financial system already served, and left the six-hour walk exactly where it was.

What Needs to Change

• Agent networks in underserved areas need working cash-in, cash-out liquidity, not just a physical presence; an agent who cannot pay out cash on demand trains users to distrust the channel after one bad visit.
• Commission structures built around new sign-ups reward registration, not use; shifting a share of agent compensation to verified transaction activity would push the incentive toward the number that actually matters.
• Deliberate agent recruitment in rural, women-served locations, rather than density targets measured at the zonal level, would address the six-hour figure directly instead of assuming national infrastructure trickles down.
• Strategy 2.0’s own KPIs should track activation rates by channel and region, not registration totals, since registration is the metric that has already proven it can rise for six years without denting the underlying problem.

source: EBR