Is it OK to make a profit serving the extreme poor?

De-regulated Britain has a nasty reputation for taxing poverty – the poor pay much more for electricity, get charged crippling interest rates, and get ripped off when they buy things (*see example below); it’s been called a ‘poverty premium’. It seems repellent but it makes commercial sense and it’s all quite legal. The argument for allowing this predatory behaviour is that this is the only way for companies to serve this demographic profitably; the problems stem from the way companies market to people whose maths isn’t good enough to work out they’re being fleeced, or who feel they have no choices.

Those of us working in the mobile for development (M4D) sector in poorer parts of the world realise we have to grapple with the same basic issue: serving poor people is risky and if it doesn’t cover its costs it won’t happen once donor support is withdrawn. But if companies are allowed to profit unsupervised from providing services to quite naive consumers then you risk questionable business practices. But can the economics ever support such services in the long run?

I’m working on an innovative Ethiopian mobile voucher scheme, and it seems perfectly logical to us to hope that lending money to farmers will allow them to buy fertiliser to grow a bigger crop to sell for more money to reinvest in a better future. This virtuous cycle depends however on a few things working out: it needs to rain, the fertiliser needs to make a bigger harvest, the price of corn needs to stay up despite there being more of it, the money farmers earn needs to support a loan interest rate that is more than the lender’s wholesale borrowing rate, and the farmer needs to repay the loan.

But what if I imagine adding another service on top, that might make the farmer an extra $20 on top of all this – if I charge less than $20 for this imaginary new service but still make a profit, then surely everyone’s a winner? This is a poverty premium question: what if I have something that I am pretty sure farmers will benefit financially from, but which I need to provide on a commercial basis – what are the limits to the economics of a service that targets poor farmers in Africa and Asia with a useful value-add which needs to make a profit?

In Ethiopia, where farmers can get agricultural advice for free on their phones, the Agricultural Transformation Agency (ATA) [note: I advise them but not about this agri advice service] has had an amazing take-up, getting over half a million registered users in less than 6 months and over 3.5 million calls. But nobody can run a free service forever unless there’s some profitable model underlying it. In this article I’m going to ask when it’s fair to make a profit, and see if there are price points at which it’s possible to improve lives and make money, and how to charge those prices.

Let’s look again at a previous blog post where I compared this successful Ethiopian agri info service with a similar one in Mali. Orange Sènèkèla is a value added mobile telecom agricultural service (agri VAS) priced at a discounted $0.10 a minute whose call centre had reached ‘few users’ and whose USSD service ‘had not acquired any customers at the time’ of GSMA’s own baseline survey. Yet Sènèkèla provides accurate market pricing data that is known to be a pain point for farmers in Mali, so it seems strange that people aren’t calling it. Although 95% of focus group respondents found Sènèkèla to be too expensive, surely a farmer selling $500 of produce should want to spend $0.30 on a three minute call that might make an extra few $; but apparently not.

Clearly price is part of the problem given that the free service in Ethiopia has a massive call volume. Let’s make a very unscientific guess at a fair price: just 5% of the focus group in Mali presumably didn’t think Sènèkèla was too expensive, meaning that the price is way out right now; if we flipped these numbers and asked what price you’d have to drop to until just 5% of farmers said it was too expensive, then perhaps $0.03 would work. The interesting thing about the Ethiopian service is that the 500k registered users is about 5% of the total farmers in the target area – having worked in mobile services for over 20 years I regard 5% as being close to saturation of a target mobile consumer base. My gut feel therefore is that a price between zero and $0.03 would probably work for this demographic, whose income let’s not forget is typically less than $1 a day if earning an agricultural wage, i.e. employed and paid regularly, rather than the even more precarious income of a small holder farmer who gets nothing if the rains fail. In Ethiopia, there were 1.63 million calls to the service 16 Sep – 2 Dec 2014, an average of 21,158 a day. If ATA had charged $0.03 for this service the income would have been $230,000 annualised.

What does it cost to build one of these services? GSMA granted $400k to each of its four mFarmer Initiative winners to ‘develop and scale commercially viable’ Agri VAS ‘for an implementation period of 2 years’, so let’s call it $200k a year. Round numbers it seems that the economics might support a profitable Agri VAS so long as farmers call it in huge numbers and it’s $0.01 a minute to call and calls last just 3 minutes.

This analysis is very superficial – a much bigger problem about pricing this kind of service is that as a poor consumer you need to try before you buy. If the service I call doesn’t tell me anything I don’t know, it’s a waste of my time and money, but I don’t know this before I call. Sènèkèla costs me $0.10 a minute even if I hear exactly what my neighbour told me just the day before. This implies that the Ethiopian service is a success because farmers lose just a few minutes of their time but no money to call it up;  they call it back again because they find it useful. This leads me to the conclusion that offering free services and then adding paid value added services on top is a viable approach.

But how much can I charge for this secondary value-added service? If the farmers trust the service, and know it’s valuable, what’s the right price point? Can I genuinely make a decent profit out of it, once the farmers trust the service I provide? If so, then we can develop economic models that overcome the ‘poverty premium’.

Let’s calibrate this by looking at farmer incomes. In Ethiopia, a farmer with 1 hectare of land may spend $100-150 on inputs (seeds, fertiliser and chemicals). This might yield a harvest of 25 quintals of teff which he can sell for 1,600 Birr per quintal, or $2,000 for the whole crop. Sure, some of this crop is for the family’s own consumption, but a 1% increase in the yield is “worth” $20. If the farmer was also an economist, he’d be willing to spend $19 on value added services so long as he could be sure it would boost his yield by 1%. Even if we discount this by 90%, he should still be willing to spend $2 for every 1% increase in yield. That’s three hours of phone calls to the agricultural advice line (at our presumed $0.03/three minute call rate).

What I learned from this exercise is the following:

  1. There’s a chicken-and-egg problem with value added services: if you don’t know it’s worth calling, you wont call it if it costs money. Free services overcome this problem
  2. A free service plus a paid value-added service looks like it ought to be sustainable if together they genuinely add value to people’s lives
  3. Even at this subsistence end of the food chain, you can provide services at a profit to the poorest people. There is a win-win to be found in providing basic mobile phone value added services to the poorest


* Poverty premium example: UK, weekly payment plan for washing machine, Dec 2014

Brightpoint total payable Hoover washing machineBHS Direct prive for Hoover washing machine

Total payable to Brightpoint if I pay £11 per week for the Hoover DYN 10166P8 washing machine is £1,716. If I buy today from BHS it’s £444. I pay more to Brightpoint in each of three whole years than I pay to BHS just once by buying now.

Are mobile operators irredeemably incapable of offering services that are meant to support the poor?

The other day in Addis Ababa I saw some astonishing numbers: since 1st July 2014 there have been over 3 million calls to agricultural value-added service Semanya Haya Simint in Ethiopia. Big numbers for sure, but for how long have Ethiopian smallholder farmers been using this service? It would take years to build a service volume like that, right?

Wrong; the service launched on 1st July this year. in just 150 days from service launch there have been three million calls to a hotline that gives information about on cereal, horticulture, and pulse/oil seed crops, as well as a wide range of agriculture-related activities. 20141028 – IVR-SMS Fact Sheet

If you’ve been to Ethiopia recently, you’ll have grappled with a monopoly called Ethio Telecom that provides one of the most frustrating mobile phone services on Earth. Coverage is patchy; a “3G SIM”, whatever that is, was unobtainable for much of Feb-June this year. To get mobile data these days you need to manually create an Access Point, which is buried deep in your Android settings; I’m not even sure I could find them on an iPhone. Even with your Access Point set up (your APN is, and the MCC is 636, in case you’re wondering), data is fitful. The SIM I bought at the official Ethio Telecom store belonged to a number range that hadn’t been provisioned on the network yet. When will my SIM work?, I asked. Maybe in a fortnight, came the unconcerned reply.

Farmers have phones

Even if the mobile phone network is a bit patchy, farmers really do have phones

Of course, the farmers in Ethiopia aren’t using smartphones to call 8028 (which is what Semanya Haya Simint means), they’re using basic phones whose selling point is they’re dustproof and go a week without recharging. But despite the shortcomings of Ethio Telecom, they’re calling 8028 in droves. There are over 530,000 registered users. That’s a customer acquisition rate of 3,500 farmers per day, on average. There are more registered users of the 8028 service than there are people in any city in Ethiopia apart from Addis.

One of the reasons I was struck by the sheer size of this Ethiopian service is that I’d recently looked at the four mAgri value added services (VAS) that each received of $400,000 grants through GSMA’s mAgri mFarmer initiative. The one that had caught my eye was Sènèkèla in Mali, because the GSMA’s report showed that Sènèkèla has done a great job of gathering a source of farmer information that addressed a specific problem: farm gate prices are volatile and so farmers can benefit greatly from knowing that the price at a particular place is higher than somewhere else. You can read the full report on GSMA’s website and at their partner RONGEAD’s site, but in brief the International Institute for Communication and Development (IICD) and RONGEAD created a network of agents to collect market price information from within 5km of markets in the main towns and the surrounding areas then develop market price advisories using some local analysis and some forecasting done in France.

Here is another way farmers can get information about best practice

Here is another way farmers can get information about best practice

When farmers in Mali use the service they either get to speak to a call centre staffed by experts, or get market prices by USSD. And how has this service fared? The GSMA is coy about the results, but their own data states that the potential market is 1.1 million farmers, with 86,000 as the target for the funding period. Although the data isn’t clearly presented, GSMA states that the “% of target market reached” is 0.01%. If this refers to the 86,000 figure, then that’s 8.6 farmers – let’s round it up to nine. And if it’s the 1.1 million farmers, it’s still only 110 farmers. This data was at March 2014, and Mali is a poor country and is suffering from political instability. But it’s hard to dress these numbers up as anything other than disappointing.

We should look for systemic problems that may explain the difference between the results in Mali and those in Ethiopia. Ethiopia is politically stable, and has a bigger population – something like 75% of Ethiopia’s 94 million people live on the land, while Mali’s population is just 15 million. Wealth and literacy indicators are similar: Mali’s population hovers around 16% female and 33% male literacy (2003 UN data), while Ethiopia is slightly ahead at 25% female, 40% male literacy (2011 data from Save the Children); GDP per capita seems to be $715 Mali vs $500 for Ethiopia. We can also state that Mali’s phone network would have to be truly awful to be worse than that of Ethiopia.

Perhaps the quality of the data being provided makes a difference? Perhaps it is because the people in Mali place greater reliance on personal connections and trust oral information from their peers more than some remote call centre of experts who perhaps only speak French, while Mali has 20 common languages? These aspects are hard to measure, and may be valid reasons, but Ethiopia too has a plethora of incompatible languages and scripts – my office colleagues who speak Amharic can’t read what the Tigrayans write, even though it’s in the same script; and the Oromffiya have transliterated their language into the Latin script, although the words are similar to Amharic.

What we need is some hard data to unpick what’s going on. Fortunately the GSMA has run a market study in Mali to try to figure out the reasons for the low take-up, and the results are both profoundly sad and utterly predictable: “95% found the service too expensive at 10 cents/ minute“. OK, so now we know that the service depended on Orange Mali making some money out of it. And you need to know that the Ethiopian service is free to call.

Shall we spend our last 30 cents on a phone call just now?

Shall we spend our last 30 cents on a phone call just now?

How expensive is Sènèkèla? Or to put it another way, what is a rational price that would justify a farmer calling Sènèkèla, so he could decide which way to point his cart when he heads to market after the harvest? Farm workers in Ethiopia may get 19 Birr per day, which is just under 1 USD. Let’s look again at the per capita GDP: average yearly income per person in West Africa is $309, so allowing for days off and to keep the maths simple, a farmer in Mali might think 1 USD is the value of a day’s work. For a household living off the land, 3 USD is a significant amount of money. And how long is a call to the service – three minutes? We know that the service costs 50 FCFA (0.10 USD) per minute to access in the Koulikoro region of Mali, which is less than half the usual network rate (108 FCFA/0.21 USD per minute). So to call the service might cost 0.3 USD for a three minute call, or in the Sikasso region it’s XOF 75 (USD 0.16) per message over the USSD channel. For a household for which $3 is a lot, then 0.3 USD is a hell of a lot to pay to get information that might not tell you anything you didn’t already know from your neighbour – that prices are down, and that it’s going to be a struggle to feed the family over the planting season.

Given these price points, why would Orange Mali want to charge farmers to call the service? Using GSMA’s own data, I calculated that if 10,000 farmers called Sènèkèla, then 1,800 of them would be “repeat users” and the call volume would be 55,000 calls, yielding about $16,500 in VAS revenue to Orange Mali. Quite by coincidence, the $400k grant over two years is $16.5k a month.

Surf the net Surf the BT CellnetWhat does this tell us? Let me tell you a story from 2000-1, the early years of the boom. My company had developed the first mobile train information service in the UK, and we’d done it by raising venture capital and buying hardware and building a WAP service, that people could call (dial-up mobile internet access!) and get their train times on a basic black-and-white WAP browser. It was very cool; not terribly valuable, but interesting, and one of the few genuinely useful services on a clunky mobile internet phone. I remember taking this to O2, and suggesting they link up to it on their WAP portal (it was called “Genie” and had been part of the infamous Surf the Net. Surf the BT Cellnet advert that almost ruined the early market for mobile internet services in the UK.) Genie’s response to me was astonishing. They said, “Yes, we’ll connect to your service – for £20,000 a month”. They wanted me to give them £20k a month, and they’d charge their customers for accessing a service that was mildly useful. I was giving them something for nothing, that had cost us a fortune to develop, that nobody had paid us to build, that the rail companies weren’t going to promote, and instead of thanking us for helping to drive up their ARPU and give some credibility back to their internet services, they wanted me to pay £20k – a month.

Perhaps Orange Mali needs to show an income stream in order to satisfy the terms of their grant; perhaps they thought it would scale and make them some proper money. Perhaps they don’t care; perhaps they didn’t do the maths; maybe Mali is different. However, the facts are there: over half a million Ethiopian farmers have called 8028 more than once, many have called multiple times. It’s free today, it may not be free tomorrow; and it’s 100% dependent on Ethio Telecom. But Sènèkèla in Mali has a long way to go before it deserves the name “Value-Added Service”, and 95% of its users say it’s too expensive to call. Still, 95% of Sènèkèla’s users was less than 100 farmers … I really hope this changes, that Orange Mali has found a way since March 2014 to drive farmers to use and value the service, to exploit the data that Orange has spent so much energy gathering, and to use it to increase the income they get from farming in a really tough climate.

Why do mobile operators think like this? ‘Bill the customer’, ‘Information is valuable’, ‘Callers must pay’. Why not offer the service free for a year while you iron out the kinks, while you figure out what is the info that farmers really want to hear, and what can be dropped. Sènèkèla has invested in a great rural farming outputs price data bank whose analysis can be used for decades to help stabilise the rural economy, helping to drive up household incomes, which in turn will mobilise savings, which provides a platform on which to develop financial literacy, financial inclusion, tax revenues and the rest. Why would you risk killing all this off by charging people 10% of a days wages to call it?