Bank-led or Mobile-led Financial Inclusion?

A few weeks back Bill Maurer (of IMTFI) and I wrote a short piece for the PYMNTS.com website’s end of year round-up of issues in the payment space. In it we predicted domestic payments in Africa was a neglected topic, likely to come up more often in the future. As evidence, we cited new Gallup data that showed that a month before the survey over 124 million people transacted in the 8 countries we analyzed, mainly in cash, indicating a large and underserved market (at that point we didn’t have all the data, we now find 134m in the 11 countries mentioned in this post).[1]

I believe that payments are an optimal gateway product for financially underserved households. Unlike credit, insurance, and savings, payments do not require trust by either party. Providers don’t have to screen clients either (as with savings) because anyone willing to sign up is profitable for providers. Payments are needed by a large number of households (over 50% of the adults in the 11 countries did one or more transactions in the past 30 days) and represent a significant pain point. Willingness to try and pay are high on the client side as well (out of all transactions, roughly 50% were in cash, many in person – highly inefficient ways to move money.)

Payments are a good way to get large numbers of previously unbanked people on the system in a hurry, without slowing down to figure out who is a good credit or insurance risk, who is going to save above a certain amount, or who to trust with hard earned savings money. But which approach should we take in developing the market for payments services? There are different paths, and it looks like they are not all equal.

One of the things the data shows is just how different one market is from the next, both in the level of activity but also in the channels through which money is sent, reflecting the different options on supply from the financial sector in each country.[2] And it turns out that not all channels are equal when it comes to serving the poor. Besides revealing a wide diversity of markets, another interesting finding is that even in the markets where banks have the most outreach (evidenced by having a larger share of transfers) they are still not reaching the poor, whereas in Kenya, Uganda, and Tanzania, where mobile money systems are at scale or scaling up, a much higher percentage of mobile transfers are initiated by poor people.

First, let’s look at how transaction volumes vary by country in the figure below. Kenya is clearly an outlier, with 46% of adults reporting sending a domestic remittance in the last 30 days. Then there is a group of reasonably active countries with 18-23% of adults report sending (Uganda, Sierra Leone, Tanzania, Botswana, Nigeria, South Africa) and a second group (Zambia, DRC, Rwanda, Mali) with roughly half that level of activity at 7-15%.

Now let’s look at how they move money.

Most of the time, people bring money in person. 21% of adults in the 11 countries had sent or carried money to someone else in the past 30 days and of those nearly 2/3rds (13 percentage points) had carried money in person in cash. Carrying cash is a common way to get money around and reflects the lack of better options for the poor as well as the non poor. It’s not clear how many of these trips were initiated just to carry money (a very costly way to move, given time and transport costs) versus how many were more opportunistic, bringing money along while traveling for some other reason. Opportunistic money sending can also be very inefficient since waiting for another reason to take a trip could imply a significant delay and extra risk of robbery.

In general, among those who said they sent money rather than carrying it in person, cash again was the most popular channel (43% of senders). About one-quarter (26%) of respondents transferred money through banks or financial institutions. Two in ten (21%) sent money by mobile phone and one in 10 (10%) used money transfer services such as Western Union.

Yet as the figure below shows, the usage of these channels also varied widely across the 11 countries. More than eight in 10 of those respondents who sent money domestically in Mali (89%), Rwanda (83%) and Sierra Leone (83%) used cash. While just 7% of Kenyans sent cash.

In fact, there appears to be roughly three types of markets. “Bank-led” payments markets (not to be confused with a bank-led regulatory model) have middle levels of activity and a significant bank presence. The Bank-led markets include South Africa and Botswana – 50% and 47% of senders respectively used banks – and to a lesser extent Nigeria – Nigeria has just over 50% of cash-based but also 44% using bank transfers. “Mobile-led” payments markets with middle to higher levels of activity, dominated by mobile transfers including Kenya – over 90% of senders used mobile – Uganda (68%), and Tanzania (60%). And “limited” markets – those dominated by cash, and characterized by low levels of activity including DRC, Mali, and Rwanda. Zambia probably belongs in this last group, in that it has some bank activity but there is also a nascent mobile presence and a strong over-the-counter presence implying Zambia is not purely cash, bank, or mobile based.[3]

So, to expand financial inclusion, which type of market should we steer toward – the bank-led or the mobile-led models?

If our initial goal is leverage the power of payments-as-gateway (discussed above) to bring large numbers of poor people into the system, should we focus policies to expand the supply of bank-based services, or the supply of mobile-based? We can get a picture of what might happen if we grew mobile or banking sector by looking at those markets where either sector is already developed.[4]

In the mobile-led markets 21% of those sending mobile based remittances were poor people. Here poor is defined as the lowest two income quintiles. Hence, we would expect 40% if the poor were represented equal to their share in the population. In the three bank-led markets, only 8% of those sending bank-based remittances were poor. Thus if we were to grow the banking sectors in the more limited markets to the level of the bank-led markets, it seems likely the market would skew toward serving the rich much more than if we were to grow a vibrant mobile sector in each of these markets.

[1] The questions in the survey referred to payments involving distant counterparties (i.e., those not in the same village or neighborhood) within the past 30 days, and respondents were asked to specify the channels they used to send and receive payments: via mobile phone, bank transfers, a money transfer or bill pay service, in cash via a friend or courier, or even traveling with the money in person.

[2] The data and graphs in this post come from Kendall, Godoy, Tortora, and Sonnenschein (still in progress).

[3] Sierra Leone is an exception to this taxonomy, with relatively high levels of activity, but nearly all of it in cash. Sierra Leone may be an outlier in that it is much smaller than the others both geographically and population-wise, and has a history of recent civil war and a strong mining industry and other factors causing significant internal and external migration.

[4] This is not to say any of these markets is fully developed or at potential, just that we should look at ones with some level of activity in a given sector to make any projections.