Mobile Money's Idle-Account Problem: What GSMA's 2026 Data Says About the Limits of Access-Only Metrics

The GSMA’s State of the Industry Report on Mobile Money 2026 leads with a genuine milestone: more than $2 trillion flowed through mobile money wallets globally in 2025, double the $1 trillion recorded in 2021. The second trillion took four years; the first took twenty.

Underneath that headline sits a figure that has changed remarkably little in a decade of reporting. Registered accounts reached 2.3 billion in 2025, growing by 268 million. Accounts active on a 30-day basis rose 15% to 593 million. That puts monthly account activity at 25.7% — up half a percentage point on 2024, and, in the GSMA’s own framing, the highest it has been since 2021.

Roughly three quarters of registered mobile money accounts are not used in any given month. That ratio is the most policy-relevant number in the report, and it is the one that access-oriented indicators are structurally unable to see.

Why the registered-account figure travels further than it should

Registered accounts are an attractive indicator for the same reasons they are a weak one. They are cheap to collect, unambiguous to report, monotonically increasing, and they aggregate cleanly across providers and countries. A national digital financial inclusion strategy with an account-registration target produces a number that goes up and can be reported against.

The trouble is that registration is a provider-side event, not a behavioural one. An account can be opened during a promotional drive, opened to receive a single government transfer, opened as a condition of an agricultural input subsidy, or opened twice by the same person across two operators. Each of those increments the numerator. None of them establishes that a financial service is being used.

This is a familiar hazard in ICT4D measurement, and mobile money is a particularly clear instance of it: the field has a long record of indicators that measure the deployment of an intervention rather than its use, and an even longer record of those indicators being read as outcomes. Registered accounts sit alongside telecentres built, devices distributed, and portals launched — all real, all countable, none of them evidence that anything changed for a user.

What 30-day activity does and does not fix

The 30-day active metric is a substantial improvement and deserves credit as such. It is behavioural, it is standardised across the industry, and it makes the gap visible rather than hiding it. Much of the value of the GSMA series comes precisely from its willingness to publish both numbers side by side.

But a single-transaction-in-30-days threshold is a low bar, and it flattens distinctions that matter for development outcomes:

• A user receiving one monthly remittance and cashing it out entirely is active.
• A user whose wages, savings, merchant payments, and bill payments all move through the wallet is also active.
• A dormant account activated once by a government transfer is active in that month and dormant either side of it.

These are not the same relationship to a financial system, and only the second one plausibly supports the claims typically made for mobile money in development terms — resilience to shocks, savings accumulation, reduced cash-handling risk, access to credit histories.

The activity rate also tells us nothing about who is on which side of the line. Aggregate activity can rise while the composition of active users stays narrow. Without disaggregation by gender, rural or urban residence, income, and literacy, a rising activity rate is compatible with deepening use among an already-included group and no change at the margin.

Reading the 25.7% carefully

There is a reasonable objection to treating three-quarters inactivity as straightforward failure, and it should be stated rather than dismissed.

Some share of registered-but-inactive accounts are duplicates — multi-SIM and multi-provider behaviour is common in many mobile money markets, and a user genuinely active on one wallet may hold two dormant ones. Some are seasonal: agricultural income cycles produce legitimate months of non-use. Some are precautionary, in the sense that a person may value holding an account they use rarely.

The honest position is that the denominator is inflated by an unknown amount, so 25.7% understates the share of people who use mobile money, while the numerator’s low threshold overstates the depth of that use. The two biases run in opposite directions and neither is quantified in the public reporting.

What is defensible is the direction of travel. The activity rate has moved by fractions of a percentage point while registered accounts have grown by hundreds of millions. Whatever the correct level, growth in the headline access metric is not translating into proportional growth in use — and the reporting series has been consistent enough for long enough that this is a pattern rather than a one-year artefact.

Implications for research design

For researchers and for donor program officers assessing digital financial inclusion proposals, the practical consequences are fairly direct.

Treat registration targets as inputs, not outcomes. A proposal whose logframe terminates in accounts opened has not specified an outcome. Ask what the theory of change requires users to do, and whether the monitoring plan can observe it.

Ask for the activity denominator and its definition. Provider-reported activity rates vary in whether they use 30-day or 90-day windows, and whether they count balance enquiries as transactions. Both choices move the number materially, and neither is always stated.

Push for disaggregation as a condition of the data, not a supplementary analysis. The distributional question — whether growth in activity is reaching women, rural users, and low-income users, or concentrating among existing users — is not answerable after the fact from aggregate figures.

Distinguish transaction value growth from user-base growth. The $2 trillion figure is impressive and largely reflects higher-value flows, including merchant and business payments. It is a measure of the volume moving through the rails, not of how many people are standing on them. Both matter; they answer different questions.

The GSMA series remains the most useful longitudinal dataset the sector has, and the report’s own presentation of the inactivity figure is not evasive — it states plainly that almost 75% of accounts are inactive monthly. The problem is downstream, in how selectively the series gets quoted. The registered-account total appears in strategy documents and press coverage far more often than the activity rate does, and a decade of that selection has produced a public account of mobile money’s reach that the underlying data does not support.