Your Long Term Mobile Money Fees As a Mobile Payments User in Ghana
In 2018, mobile money users in Ghana transacted GHS 94 on average according to BoG’s figures. Mobile money interoperability was launched within the same year, which I wrote about here.
The new BOG figures got me thinking along the lines of: If as a mobile money user, you’re transferring about GHS95 on average in mobile payments, how much are you most likely to be paying in fees over time? If your thinking is that it’ll depend on the Telco network used by both the sender and the receiver, you’re right.
As a mobile money user in Ghana, an answer to this is important, because it gives you and I a sense of which network platforms currently allows you to transact on the platform in an affordable way and how much you’ll likely pay in fees over the long term.
This allows you to say something along the lines of: “Because on average I transfer GHS95 using Voda Cash to a receiver on AirtelTigo Money, I expect both of us to pay between . . . and . . . in combined fees over the long run.” The fee in this context is the combined cost of making transfers and cashing out.
In this article, I will first share my findings from a consumers perspective. If you’re interested in the technical details of this analysis you can read beyond my findings.
Before you proceed, please keep in mind that 2018 mobile money tariffs were used here, not 2019 tariffs.
Same Network rules them all. . . if you’re on MTN Momo
The chart above is the shows the average fees and the uncertainty or certainty around these averages.
MTN Momo users (MTN Momo to MTN Momo) are highly likely to pay 2.2% in fees on average if transacting at an average of GHS 95. That’s the lowest by far among every possible mobile money alternative to make transfers.
More broadly, this is interesting given MTNs current position as market leaders in Ghana where you’d expect them to instead, have higher long term fees even at low transfer amounts. This is generally good for Ghana’s mobile money eco-system since MTN Ghana’s platform is currently the biggest with 93.9% of all mobile money floats on their platform alone, as at 2018, therefore most transactions on the platform will be on MTN Momo.
However, it becomes clear that MTN Momo generally isn’t the most affordable if you typically transact with other mobile money platforms, relatively speaking. An example is MTM Momo to AirtelTigo Money which most likely averages at 3.5% in fees.
This goes without saying that interoperability has lowered the barriers to cross-network transactions compared to the previous ‘ token ‘ system which would probably have been way more expensive due to the friction the process introduced.
Therefore interoperability has brought clear benefits in terms of efficient transfers and lowered fees.
Bests parts of an interoperable mobile money system
The good thing about this way of generating long term fees is that we can mimic two users (a sender and receiver) who for every transaction they make, it is done using the platforms that offered the most affordable combined fee.
Of course, in reality, this may not happen but it gives us a good benchmark for other fees since this benchmark is a minimized fee. It basically combines the best parts of mobile money interoperability.
The fee-optimised channel comes out at about 2.14% on average, lower than most of Mobile Money Service alternatives if your transactions average at GHS 95.
MTN Momo have in essence figured out how to make their platform competitive in pricing terms for low-transfer users because their 2.2% compares well vrs an optimised fee of 2.14%. By simply doing MTN to MTN, over the long term you’ll likely be paying fees close to a fee-minimizing service. It may not be true for higher average amounts (which we’ll see later).
At the lower end of the table, it shows there’s still some way to go for platforms such as AirtelTigo and Vodafone to reduce long term fees for users who use their platform. This will definitely be beneficial to low-income users, who are a core customer segment Telcos are looking to include.
According to the World Bank, the global average cost of sending $200 in remittance was as high as 7%, in the first quarter of 2019. They add that in many African countries it’s over 10%. So for combined mobile money fees in Ghana to be below 4%, perhaps it highlights the progress we’ve made in this direction.
BUT… What happens to fees paid when the average Mobile Money transaction increases?
I extended this analysis by looking at situations whereby the average transfer increased from GHS 95 all the way to GHS 500.
Turns out, generally fees reduced as average transfers increased.
AirtelTigo Money and Voda Cash’s fees are more responsive to changes in the average amount compared to MTN Momo’s. This is clear from the steeper lines of say, AirtelTigo Money to AirtelTigo Money. MTN Momo’s is generally flat, meaning as amounts increase, fees paid are roughly still the same. Maybe MTN Momo’s percentage pricing is responsible for this.
What does this mean for users?
In all, over the long term as trust and adoption of seamless mobile payments increases, value transacted may increase too. If that is the case, it’s good to know there are alternative platforms beyond the market leader which are responsive to these dynamics. There’s still some work to do to reduce fees for the user who typically send low amounts.
Beyond all this, it’ll be interesting to include various forms of mobile money taxation from Zimbabwe, Uganda, Gabon and Kenya to understand how it changes these long term averages for users.
I consult on related research and data analytics projects like these. If you want to collaborate on a project, please email me using dave[my surname] at gmail dot com.
The Technicalities: What went into this analysis.
Motivation: I’ve been experimenting around what an estimate of fees you’ll typically be paying as a mobile money user looks like. The quest to answer this motivated most of this work which ended up being far more computationally intensive than I had anticipated. Therefore, most of my time was spent finding innovative ways around memory constraints to scale this analysis.
To scale, most of this had to be done in the cloud. I hope to write about how I combined R + Cloud + Parallel computation in the future.
Priors: The prior was a beta distribution of simulated data. This distribution had an average of 95 and skewed to the right because that’s how I imagine a users transaction averaging GHS 95, to be, hence an approximation to reality.
10,000,000 simulated transactions were used in all. GHS 95 was used as an average for the simulations since, according to figures put out by the Bank of Ghana, that’s the average value transacted on Ghana’s mobile money platform in 2018.
Simulation: Simulation was used because beyond producing an average, it gives the uncertainty around the averages.
The simulation takes each transaction, runs it through the algorithm and computes the amount left after transaction and cash-out fees are subtracted. The simulation is run 1000 times for 10, 000 simulated transactions. 1000 simulations are possibly not enough but this was done to reduce cloud computing time which gets costly quickly.
The same set of simulated transactions are run through each mobile money service combination in order to better understand how each platforms fees changes under similar payment circumstances.
Fees Data: The transfer fees (both same & cross-network) and cash-out fees used in the computations were publicly available as at October 2018.
Hope you found this as interesting as I did putting it together. Thanks for reading!
Find the data used in this analysis here.
I’m always open to meeting interesting people. If you have any question or want to discuss a possible collaboration, feel free to get in touch!
Originally published here.
