What is an excise duty?

An excise duty is similar to a VAT as its value is based on the price of the service. However, unlike VAT, excise duty is applied on the amount sold by an operator, as opposed to being applied to the retail price.

This difference is important because from the perspective of the consumer, the tax is not distinguishable. Operators will be required to pay this tax directly based on the revenue generated by the transactions.

How could this impact the market?

As the operator pays the tax directly, the operator could decide to take the tax hit completely, leaving retail prices unchanged, and to absorb this extra cost entirely, but this would significantly harm its profitability and in the medium term prices would have to rise. Alternatively, the operator could pass part of this cost through, and the burden of the tax will be reflected in a proportional increase of the price of mobile money services for the customer.

Therefore, in general, excise tax either:

• Makes the market less profitable for a provider which would decrease their willingness to invest and develop, e.g., products and distribution networks
• Makes the market prices higher, increasing the barrier for customers to use the service.

Is the transposition of the excise duty on the consumer legal?

The Kenyan Finance Minister Njeru Githae has declared that the duty will not affect the consumer, adding that it will be paid by the service providers.  He went further to warn operators that the government will monitor the mobile operators to ensure they don’t make any secret adjustments to transpose the cost to the customer.

While I am sure that the Ministry will check collusive pricing behaviour, the Minister shall be aware that it would be perfectly rational and economic for operators to increase retail prices by the same amount of the new tax, as it is a legitimate cost increase they are facing. This transposition would be legal.

What’s next?

Mobile money transfer services in Kenya are an integral part of the country economy, with almost 20 million Kenyans using this service. The Central Bank of Kenya estimates that in the first half of 2012 KES 726.23 billion (approximately USD 8.6bn) were transferred through the mobile money platform. It’s hard to understand the rationale of this new tax on a service that is greatly benefiting the whole Kenyan society. Voice services in Kenya are taxed at a rate of 26% ( 16% VAT + 10% excise Duty). Adding this new 10% tax on mobile money transactions will stifle a nascent and important driver of commerce and socio-economic development. Unfortunately there was no consultation or engagement with stakeholders before the amendment was introduced.

The President must assent to it for the tax to become law. Our hope is that he will deny his assent.

It’s in Kenya that Mr. Sayem Ahmed, Chairman of the Executive Committee of the Board of DBBL, and Mr. Abul Kashem Md. Shirin found the solution. Eighteen months ago, they embarked on a trip to East Africa to learn from the world’s most famous mobile money service, M-PESA. When they came back to Bangladesh, Mr. Shirin presented the idea of a mobile money service for the unbanked to the board along with a compelling business case.

When MNOs enter the mobile money business, they see a business case of direct revenues from revenues and indirect benefits of ARPU lift, churn reduction, and/or savings on airtime distribution. However, a bank’s rationale for entering the mobile money business will necessarily be different as they don’t have the indirect benefits.  In DBBL’s case their aim was to increase the balance sheet, growing the number of deposits taken in over mobile banking and on lending those funds.  On this basis, they estimated that at an interest spread of 5%, a deposit of 5 billion taka (USD 62.5 million) would create an annual income of 250 million taka (USD 3.125 million), which is equivalent to the maximum annual expense of the project. Mr. Abul Kashem Md. Shirin was hoping to achieve the above targeted deposit and income by the end of 2013.

Following the Central Bank’s “Guidelines on Mobile Financial Services (MFS) for the Banks” (of 22 September 2011, as amended on 20 December 2011), it was clear to both Bangladeshi banks and MNOs that mobile money would only be through partnership models led by banks.  DBBL launched its mobile money service on 31 May 2011 in partnership with Citycell and Banglalink, the number 4 and number 2 MNOs in the country, respectively. The service which is powered by Sybase 365, a division of SAP, has 338,000 unbanked customers (as on 18 July, 2012).

Unbanked people subscribed to any mobile network operator (MNO) in Bangladesh can register to the service. With a DBBL mobile banking account – which is different from a standard DBBL bank account – customers can perform P2P transfers, merchant payments, utility payments, foreign remittance transfers, and receive salary and government allowance disbursements. The most popular transaction in terms of number of transactions performed is airtime top up, while it is cash-in, in terms of value.

After the launch of its mobile money service last year, DBBL opened 400 small offices in the country’s rural areas. A total of 10,423 agents have been appointed to perform cash-ins and cash-outs for customers. These agents are normally small shop owners or retailers of various MNOs. Agents perform on average 3.3 transactions (cash-in and cash-out) per day.

DBBL has now partnerships with Citycell, Banglalink, Airtel and GrameenPhone. As per the partnership agreement, MNOs provide USSD connectivity between the DBBL server system and agents/customers who are using their mobile phones. They also engage their retailers to work as DBBL agent points. In return, MNOs get around 25% of the transaction fees paid by customers[1].

Launching the service proved more difficult than expected. In particular, the main challenge faced by DBBL was customer education. But 14 months after the launch of its mobile money service, DBBL is on the way to reach its financial targets.

When I asked Mr Abul Kashem Md. Shirin what advices he would give to other banks willing to launch mobile money service for the unbanked, this is what he told me:

• Don’t be afraid to put a lot of money on the table: mobile money requires heavy initial investments;
• Manage expectations: mobile money will be profitable in the long-term, but don’t adopt a short-term view.

M-PESA continues to make headlines, increasing revealing insights on the economics of mobile money. Recently released financial statements from Safaricom indicate that M-PESA contributed 15.8% of Safaricom’s total revenue (compared to 12.5% the previous year), making it the company’s second largest revenue stream after voice. Most strikingly, M-PESA revenues increased 43% with only a 6% growth in registered customers from the previous year (now at 14.9M). This suggests that most of the revenue growth was driven by a large increase in the number of fee-based transactions (e.g., peer-to-peer transfers) per customer.

In addition to fee-based revenues, M-PESA generates cost savings for Safaricom’s airtime business: 29% of Safaricom’s pre-paid airtime top-ups are now conducted electronically through M-PESA rather than through paper-based scratch cards, which are more costly to distribute. M-PESA also appears to help reduce churn on Safaricom’s core airtime business: Safaricom’s customers increased by 11% to 19.1M from 17.2M recorded in FY2011, despite raising its airtime tariffs above its competitors in early 2012.

Though we can glean M-PESA’s revenue contributions and indirect benefits derived, direct profit contributions are undisclosed and sensitive to cost assumptions. Large costs associated with building and maintaining a mobile money deployment include agent network management, marketing, and platform licensing fees. Remarkably, M-PESA’s agent network grew by 46% (to 36,700 agents) over the last fiscal year. At this scale, commissions paid to agents for each cash-in/cash-out transaction they conduct, as well as for signing up new customers, is substantial.

Nevertheless, M-PESA’s continued revenue growth in Kenya is good news for the industry, as it may encourage other providers to invest more heavily in building out mobile money systems that serve poor and rural customers. While Safaricom’s M-PESA has been an anomaly to date, accounting for nearly 70% of all mobile money transactions globally, according to MMU’s Global Mobile Money Adoption Survey, we are now seeing mobile money gain traction in other markets. Notably, M-PESA now accounts for 8.5% of Vodacom’s service revenue in Tanzania, compared to 2.8% the previous year.

Despite our optimism regarding the growing importance of mobile money as a business driver for mobile operators, we recognize that the business case for mobile money in developing markets is undermined by the high on-going cost of converting cash to electronic value (and vice-versa). This underscores the pressing need to enhance the range of use-cases (beyond p2p transfers) that can be performed electronically. Increased experimentation on revenue models through fee structure changes and new product offerings may help to boost the proportion of electronic transactions conducted relative to cash-based transactions. Particularly from a financial inclusion perspective, getting the economics right is critical to enable new partnership models that can yield a range of affordable mobile-enabled financial services for low-income segments.

These presentations are available in English and Spanish.

[Add film clips]

These presentations are available in English and Spanish

1. Mobile Money Pricing – (presented by Paul Leishman)

2. Precios del Dinero Movil – Part 1 and 2 (presented by Camilo Tellez)

In part, this problem is a function of our industry’s stage of development: for most operators, mobile money is a new (and unfamiliar) business, so it hasn’t been completely clear which metrics they should monitor and manage. But the other part of the problem, particularly for operators who do know the metrics they want to monitor and manage, is that technically it can be hard to get the necessary data. While most operational metrics that an operator would need to track can, in theory, be gleaned from the m-wallet transactional database, in practice we find that the data doesn’t always make its way onto the m-wallet reporting dashboard in an easily digestible way.

So today we’re pleased to present a solution to this problem: the GSMA Mobile Money Metrics Dashboard. This Dashboard presents practitioners with an easily digestible summary of their operational metrics that matter most.

And this tool is more than an excel spreadsheet: for existing customers of Comviva, Fundamo, Sybase 365, or Utiba, the Dashboard can be integrated as a reporting feature free of charge. Moreover, these vendors have agreed to include the Dashboard as a free, pre-installed reporting feature for every future m-wallet customer. Thus, all existing and future customers of Comviva, Fundamo, Sybase 365, and Utiba will be able to access this Dashboard at the click of a button. If you’re a customer of one of these vendors, send an email to mmu@gsm.org to get access to the Dashboard. Other stakeholders should feel free to request a copy as well.

So what, exactly, is in the Dashboard? That is, what are the operational metrics that an MNO should monitor and manage? After consultation with leading mobile money practitioners and industry experts, we have selected five key sections for inclusion in this tool:

1. Aggregate volume and value – These graphs present a practitioner with the bottom line in terms of aggregate growth. The Dashboard also presents the monthly growth for each individual service (i.e. money transfer, bill payments, airtime top-up) to help operators identify whether top-line aggregate growth is masking important product-level trends.
2. Customer base – In this section, the Dashboard presents the growth of active and inactive customers. Separating and comparing these two figures can help illuminate whether an operator has an issue with customer inactivity.
3. Customer engagement – Beyond growth in the number of active customers, a healthy mobile money service is one in which customers are adequately engaged with the service. This section of the Dashboard measures the number of transactions per customer, float held per customer and the distribution of transaction frequency across segments.
4. Agent engagement – A healthy mobile money service doesn’t just have engaged customers, it will also have engaged agents. To this end, the Dashboard presents the number of cash-in and cash-out transactions per agent, as well as the average balance of e-money agents keep on hand (a good indicator of whether they’re seeing value in the service).
5. Customer and agent balance – This section of the Dashboard presents the number of active customers per agent, a ratio practitioners must closely manage, and the growth in total number of agents.

Profitability Series

1.      Is there really any money in mobile money?

2.      How much must an MNO invest in mobile money before turning a profit?

3.      How significant are airtime distribution savings to profitability?

4.      How significant are churn reduction benefits to profitability?

5.      How significant is ARPU uplift to profitability?

6.      How significant are direct revenues to profitability?

7.      How can an MNO manage costs to achieve profitability?

8.      How can MNOs ensure their tariff and commission models are well designed?

9.      What metrics should an MNO monitor and manage?

For a mobile money service to scale and achieve profitability, it’s critical to have well designed customer tariff and agent commission models. This is a topic we’ve written about extensively already: Neil Davidson and I shared some insights on how MNOs can best incentivise agents in the “How to Incentivise a Network of Mobile Money Agents” section of our recent agent network handbook; in August I wrote a blog post detailing four common pitfalls that MNOs should avoid when designing their agent commission models; and any reader who’s used the GSMA Financial Model will know that pricing and commissions are one of the most sensitive variables at play. So how can MNOs ensure their tariff and commission models are well designed? Here again, MTN Uganda’s MobileMoney reveals some key insights.

If the MTN Uganda MobileMoney customer tariff model looks familiar to you, that’s probably because you’ve seen it in action before: in structure, it’s similar to the one used by Safaricom’s M-PESA. And as Ignacio Mas noted in the 2009 Mobile Money for the Unbanked Annual Report, this tariff structure (and the way it’s taken to market) works for a few reasons: it’s simple and transparent, customers are not bound by minimum balance requirements or prohibitive deposit fees, and it offers customers an ability to send money to non-customers.[1] It’s inevitable that MNOs will innovate and trial new models, but the design features listed above can be considered prerequisites for an effective tariff model in any environment.

It also merits note that MTN Uganda’s customer tariff model grants customers minimal leeway to defraud the operator of prospective direct revenues. That is, given that the p2p transfer fee typically accounts for less than half of the total end-to-end cost of sending money using the service, customers have little incentive to perform a direct deposit. Moreover, MTN has structured its tariff tiers in such a way that there is no opportunity for a customer to reduce their fees by splitting a cash-in or cash-out into multiple smaller tranches.

But it’s not just MTN’s customer tariff model that merits attention. Their agent commission model has been thoughtfully designed, too. The article Neil Davidson and I wrote for the 2010 Mobile Money for the Unbanked Annual Report details most of our thinking on agent incentives, but it’s worth briefly noting here how MTN espouses some key principles.

First, MTN pays MobileMoney agents a commission for every activity that they perform, even though MTN may not charge customers a fee directly for each one. For instance, even though MTN doesn’t charge customers a fee to cash-in, they do provide agents a commission for providing this service in recognition of the time and cost involved. Of course, while MTN take a temporarily hit by subsidising cash-in, the fees collected from an end-to-end money transfer (which includes a cash-in, a transfer, and a cash-out) do exceed the corresponding commissions paid. All told, the margin MTN earns for a typical end-to-end P2P transfer (excluding variable technology fees) to a registered customer is just north of 50%.

Second, while MTN may pay agents for both cash-in and cash-out, they deliberately pay a higher commission to agents for facilitating cash-out than they do for cash-in. This stems from the simple fact that ‘cash-out’ agents have a higher cost of restocking their inventory of physical cash than cash-in’ agents do for restocking their inventory of e-money. As such, ‘cash-out’ agents must be compensated accordingly.

Third, MTN recognized that to keep agents engaged in the period following launch when transaction volumes are typically low, it would be important to provide them with a different source of revenue. To this end, they have provided agents with a commission for every customer that they register.[2] Thus, in the early days following launch, MobileMoney agents earned money by registering customers; as the service scaled they increasingly earned their money from facilitating cash-in and cash-out transactions for customers.

Profitability Series

1.      Is there really any money in mobile money?

2.      How much must an MNO invest in mobile money before turning a profit?

3.      How significant are airtime distribution savings to profitability?

4.      How significant are churn reduction benefits to profitability?

5.      How significant is ARPU uplift to profitability?

6.      How significant are direct revenues to profitability?

7.      How can an MNO manage costs to achieve profitability?

8.      How can MNOs ensure their tariff and commission models are well designed?

9.      What metrics should an MNO monitor and manage?

[1] Ignacio Mas: Good Service Design Features of M-PESA’s Money Transfer Service. 2009 MMU Annual Report.

[2] This decision also helped drive customer growth.

But before we begin our evaluation process, let’s first briefly take stock of the costs (and there are many) that are involved in launching a mobile money service. Before launch, MNOs incur a series of technology costs, including investing in an m-wallet platform, upgrading their SIM or USSD access gateway (in most cases), and deciding whether to embed their application on all new SIMs – and in most cases consequently upgrade to a larger card (while this isn’t a cash outlay at first, its a decision of major financial significance). The next tranche of costs are go-to-market related, and include recruiting and paying for management and back-office staff, training and merchandising a network of agents, and designing and launching above and below-the-line marketing campaigns. Most of the costs identified thus far carry on after the service has been launched, but the day a service goes live, a third set of costs come into play: ongoing costs. These typically include cash-in/cash-out agent commissions for agents, SIM cards, starter packs and agent registration commissions for customer acquisition, and internal transfer fees for using SMS services or selling airtime at a discount. For a full breakdown, refer to Exhibit 6.

So which of these are strategic and which, if any, are discretionary? Unfortunately, answering this question is not as simple as sorting costs according to size. If we look at the drivers for MTN Uganda’s MobileMoney, the service we’re using as the focal point for our profitability discussion on this blog, we find that highly strategic operational activities – things like building and managing an agent network, or providing great customer care – are comparatively inexpensive. Since launch, 7% of MobileMoney’s total costs have been on building and managing their agent network[1], and 4% has been on back-office customer care.[2] And while it’s true that Safaricom spends somewhat more on these particular activities, and has benefited from an agent network of industry leading quality, the insight is still applicable: these activities are routinely touted as strategic imperatives for any successful mobile money service – but for MTN, they’ve cost a pittance compared to the amount spent on technology[3] (30%) or customer registration commissions[4] (12%) to date.

So if these activities deliver such good value for money, why do some practitioners have a difficult time getting budget to do them properly? In many cases, this stems from the fact that highly strategic, financially insignificant costs often require a commitment to spend in advance of having any indication of whether the mobile money service will be a success. For instance, MTN had to commit to a fixed monthly contract with a field marketing agency ($623,000); pay for and train their dedicated call centre representatives ($440,000); and design and fund an above-the-line marketing campaign ($850,000) all prior to launching their service. Each of these activities has been instrumental in MTN Uganda’s success, and their decision to invest aggressively in them ultimately stemmed from their confidence that the service would become a hit.

But that’s not to say all of their spending has been strategic; some costs were discretionary, and potentially could have been substituted for less expensive, equally effective alternatives. For instance, MTN recently introduced an airtime bonus for customers who top-up using mobile money, an incentive many MNOs have used in an effort to encourage customers to top-up using their e-wallet. But this tactic was particularly costly since it negates a big portion of the savings realised from eliminating discounts paid to dealers.

Moreover, like they have in other markets, MTN has pursued a strategy of aggressively registering new customers in Uganda. In practice, this has meant registering more inactive customers (552,213) than active ones (421,254). And this strategy has been expensive: MTN has spent a total of $1.3 million on registration commissions and new SIM cards for customers that have not performed a single revenue-generating transaction (stay tuned to this blog for the findings of GSMA research into customer activation).

Profitability Series

1.      Is there really any money in mobile money?

2.      How much must an MNO invest in mobile money before turning a profit?

3.      How significant are airtime distribution savings to profitability?

4.      How significant are churn reduction benefits to profitability?

5.      How significant is ARPU uplift to profitability?

6.      How significant are direct revenues to profitability?

7.      How can an MNO manage costs to achieve profitability?

8.      How can MNOs ensure their tariff and commission models are well designed?

9.      What metrics should an MNO monitor and manage?

[1]Includes handset subsidies, agent POS merchandising, and field marketing agency costs

[2] Includes total cost of back-office staff

[3] Includes cost of m-wallet platform and monthly charges, SIM access gateway upgrade and monthly maintenance charge, and SMS communication fees

[4] Includes commissions paid from MTN to agents ($1.33 per registration).

For MTN Uganda, who currently offer domestic money transfer and mobile top-up services, direct revenues include fees to send money, and fees to withdraw money from an e-wallet. To date, these direct revenues, less commissions paid to agents, contribute 52% of total gross profit for the service. It’s clear then, that this is an area of the business case not to be neglected. So how can MNOs ensure they’re well positioned to fully capture this source of value? Well in the case of MTN Uganda’s MobileMoney, one decision has had more of an impact than any other: enabling p2p transfers to unregistered recipients.

Uganda is a fragmented mobile market: according to Wireless Intelligence at time of writing, MTN holds 44%, Zain, Warid and Uganda Telecom each hold roughly 18%, and Orange holds 3% market share. So it’s not surprising, then, that when MTN launched the service, they made sure customers could send funds to recipients on any network. To date, 38% of P2P transfers made using MobileMoney have been from a registered customer to an unregistered recipient; and this use case has generated 45% of total revenue (and even more in gross profit). Two things are striking about this data: first, the overall number of P2P transfers to unregistered users is quite high, which suggests that had MTN not offered this option, they likely would have left some revenue on the table. Second, P2P transfers to unregistered users are more lucrative for MTN than P2P transfers to registered users (i.e. 38% of transactions are generating 45% of revenue). This occurs because MTN charges customers a premium – 7% for low and 94% for highest value transfers – to make a transfer to an unregistered recipient, and the commission paid to agents remains the same. Thus, by enabling P2P transfers to unregistered recipients, MTN not only expands the base of potential users for their service, they also generate a significant amount of revenue.

But not every MNO allows P2P transfers to unregistered recipients: some reason that by doing so, they are forfeiting potential net new mobile revenue from recipients who, if they want to receive money, have no choice but to activate a SIM from the MNO in question (and then, as the theory goes, start to use this new SIM for mobile services, too). But this walled garden logic is risky: mobile money is a service that is predicated on network effects, and particularly in countries with fragmented mobile market share, the ‘closed model’ presents an insurmountable customer experience barrier to adoption, ultimately making it difficult to scale the mobile money service. And if a mobile money service cannot scale, its sustainability becomes questionable – so in the end, any benefits of net new revenue will be short lived.

Its clear, then, that direct revenues are a significant value source, and mobile network operators have an opportunity to maximize them by enabling P2P transfers to unregistered recipients – a feature that, coincidentally, is just what customers in Kenya, Uganda and other successful mobile money countries have demonstrated that they want.

Profitability Series

1.      Is there really any money in mobile money?

2.      How much must an MNO invest in mobile money before turning a profit?

3.      How significant are airtime distribution savings to profitability?

4.      How significant are churn reduction benefits to profitability?

5.      How significant is ARPU uplift to profitability?

6.      How significant are direct revenues to profitability?

7.      How can an MNO manage costs to achieve profitability?

8.      How can MNOs ensure their tariff and commission models are well designed?

9.      What metrics should an MNO monitor and manage?

To gauge whether mobile money actually causes customers to spend more, ‘uplift in ARPU’ would need to be measured over time. But this particular type of analysis is tricky. First, the average selling price for airtime and SMS, and therefore ARPU, in a country varies for any number of reasons on a month-to-month basis, so it’s impossible to simply attribute any change solely to mobile money. Second, in many cases ARPU figures will already include revenue generated from mobile money – so taking credit again would be inaccurate.

It’s clear, then, that ‘uplift in ARPU’ isn’t a perfect metric. But what, if anything, is? We propose that the less catchy, but somewhat more accurate, phrase of “increased share of wallet for voice and SMS” is the more relevant metric. By measuring ‘minutes of use’ and ‘billable SMS events’, an MNO can isolate changes in customer behaviour, something that’s not possible with an ‘uplift in ARPU’ calculation. Additionally, “increased share of wallet” accurately describes just why a mobile customer might consume more mobile services on their mobile money SIM; that is, it’s easier to imagine a customer who carries two SIM cards, each month spending $3 on one, and $2 on the other, shifting some of her spending to the stickier of her two SIMs.  So if we accept “increased share of wallet for voice and SMS” as a good metric, the question still remains: is it a significant driver of profitability?

Unfortunately, our findings in this department are inconclusive. From a survey conducted in 2009 by McKinsey & Co., CGAP and GSMA, we know that in the Philippines 44% of mobile money users carry more than one SIM, and 68% report using their mobile money SIM as their ‘primary SIM’; this is encouraging, but not conclusive evidence that this benefit is real. In the case of MTN Uganda’s MobileMoney, active customers do consume slightly more voice and SMS than non-mobile money customers, but drawing a solid conclusion here would be incredibly challenging from a data-mining perspective.

While we haven’t conclusively pinpointed the impact of “increased share of wallet for voice and SMS” in a financial model, it’s plain to see that the potential to reap benefits is massive – and there are some steps MNOs can take to position themselves to do so.

Beyond executing well to ensure customers do indeed have an incentive to keep their mobile money SIM in the phone more often than not (a subject I discussed in the previous post in this series), promoting mobile money as a method of topping up is also important. In particular, MNOs have found success by promoting mobile money as an option for topping up in small increments, and topping up after hours when scratch cards may be unavailable. For instance, WING, a Cambodian mobile money service, has enjoyed success with their mobile top-up feature, and found that 33% of top-ups on their system occur outside typical store hours, and 70% occur at the US$1 price point, a level at which scratch-cards are a particularly expensive as a distribution option.

Profitability Series

1.      Is there really any money in mobile money?

2.      How much must an MNO invest in mobile money before turning a profit?

3.      How significant are airtime distribution savings to profitability?

4.      How significant are churn reduction benefits to profitability?

5.      How significant is ARPU uplift to profitability?

6.      How significant are direct revenues to profitability?

7.      How can an MNO manage costs to achieve profitability?

8.      How can MNOs ensure their tariff and commission models are well designed?

9.      What metrics should an MNO monitor and manage?