An evaluation across 79 markets changed what merchants and consumers believe about digital payments, but not how they pay. The economics of the cash economy explain why.

Mobile money is everywhere in Tanzania. More than 60 percent of adult Tanzanians have mobile wallets, distributed across four major providers and many send and receive transfers regularly. Demand for cross-network payments led the Bank of Tanzania to build the Tanzania Instant Payment System (TIPS), which improved the rails for interoperable digital payments and pushed providers to expand merchant acquisition. By 2025, almost 2.8 million merchant digital payment accounts had been registered. 

Yet, despite the growth of mobile money and the expansion of digital payments infrastructure, many still reach for cash to pay for everyday purchases. This pattern is not unique to Tanzania. Across sub-Saharan Africa, consumers use mobile money for remittances, but return to cash for retail payments. What explains this? 

Why Person-to-Merchant Payments Are the Harder to Coordinate

There is a tendency to see digital person-to-person (P2P) payments and person-to-merchant (P2M) payments as two sides of the same coin: solve one side and the other follows. Once the technological infrastructure is in place, both hinge on coordination between senders and receivers. But they sit within different networks and social structures, and that changes how hard the coordination problem is to solve.

P2P is one-to-one. It connects pairs of senders and receivers who often already have social ties such as a daughter and mother. They can both easily communicate with each other, observe how the other uses mobile money, and benefit directly and immediately when both adopt it.

P2M operates differently; it depends on many-to-many coordination. Even if one merchant-consumer pair transacts digitally, this does not ensure that other merchants and consumers in a market will also start using the technology—especially when virtually everyone else still uses cash. With imperfect information about others’ preferences, both sides default to the payment method they are certain the other side will accept.

Survey evidence we collected across 300 markets in Tanzania in 2025 show how cross-sided misperceptions contribute to the cash equilibrium. Eighty-two percent of merchants said they would use Lipa kwa Simu (LKS) more often, but believe their customers prefer cash. Seventy-five percent of consumers said the same, blaming merchants for not using digital payments.

 Can Two-Sided Activation Break the Cash Equilibrium? 

We tested this idea in a randomized evaluation (RCT) with Innovations for Poverty Action (IPA) across 79 markets in six regions of Tanzania. The logic was straightforward: could incentivizing a small group of merchants and consumers within a local market to use digital payments lead others to follow? In program markets, a small group of merchants and consumers received incentives to use LKS over eight weeks. In half of those markets, the same group also served as LKS Ambassadors, sharing promotional materials and encouraging others to try digital payments. Our main outcomes measured whether this shifted the beliefs and behavior of the other merchants and consumers who were not part of the program.

The evaluation changed beliefs, but not behavior. The intervention worked—up to a point.

Program participants used LKS, earned incentives, and distributed promotional materials, increasing the visibility of digital payments in local markets. Even merchants who did not participate in the program came to believe that digital payments were much more common. Compared with similar merchants in markets without the program, they estimated that about 75 percent more of their peers accepted digital payments. These changes in perception also translated into action: non-participating merchants became much more likely to ask customers to pay with LKS, a practice that was rarely observed in markets without the program.

But consumer behavior change did not follow. This was true  both for participating consumers—only 18 percent earned the maximum bonus, compared with 47 percent of participant merchants—and for  consumers who did not participate in the program. The upshot was that the rate of digital payments in markets with and without the program remained essentially the same.

Why Didn’t Consumers Switch to Digital Payments? 

We argue that the answer lies in the economics of the informal markets in which retail, and now digital, payments are embedded. 

Even in markets chosen because digital payments were most likely to take off, where at least 25 percent of merchants had Lipa Namba accounts in early 2025, 95 percent of consumers in our sample reported earning their income in physical cash: from trading and selling, from services such as transport, hairdressing, and food vending, or from piece work and casual labor. Thus, the money they earn—daily in small irregular payments—arrives in their hands rather than in their mobile wallets.

The way consumers earn then shapes the way they pay for goods and services. Merchants in the same markets as these consumers report that the median consumer-merchant transaction is TZS 1,000 (less than USD 0.50), with three-quarters of transactions below TZS 2,000. Under these circumstances, few payment technologies can rival cash. It arrives in the same form in which it is spent, is accepted by virtually every merchant, and can be used immediately at no additional cost. 

While nearly all of consumers’ earnings come from cash, they also receive a sizable share of inflows from P2P remittances.¹ Thus, many consumers have—at least for a few days, before they cash out—a wallet balance that could be used for LKS. So why doesn’t this serve as a springboard for digital payments? 

One fundamental reason is that mobile money services were designed to support the cash economy, not to replace it. Built and optimized for liquidating transfers, wallets have functioned primarily as pass-throughs for cashing out remittances. Ninety-six percent of consumers in our sample live within a few minutes’ walk of an agent to do exactly that.

Two additional factors reinforce this pattern. With Mobile Network Operators’ (MNO) revenue streams highly dependent on withdrawal fees, they have no incentive to encourage consumers to skip the cash out, keep money on their phones, and pay with LKS. And regressive pricing makes it worse: MNOs impose the highest fees on the digital payments consumers make most often—under TZS 20,000.

Implications

The result is a self-reinforcing cycle. Consumers pay in cash, putting hard currency in the hands of merchants, who, in turn, use it to pay their own suppliers, reproducing the cash economy.

Breaking this cycle is difficult given its roots in informal economies and mobile money's origins as a cash-delivery system. Supply-side investments in payment rails, merchant acquisition, and QR codes remain important, but our findings suggest demand-side factors matter just as much. The transition to a digital equilibrium will remain out of reach until low-value digital payments become affordable and providers do more to promote paying digitally over cashing out. 

¹Consumers in our sample receive, on average, three mobile money transfers per month, totaling 75,000 TSh (~$30).