Cash Aid via Mobile Payment in Kenya - An Evaluation

Posted by KatrinVerclas on Mar 30, 2009

In early 2008 violence errupted in Kenya after the most recent elections there the previous December.  Post-election tribal warfare resulted in the death of 1,200 people, internally displaced 400,000 to 600,000 people, and destroyed more than 41,000 properties.  The economic cost of the crisis has been estimated at more than KSh 100 billion (approx US $ 1.5 billion), with more than half a milion jobs lost. The World Bank noted that over 2 million Kenyans may have been driven into poverty as a result of the violence.  Food security also declined with farmers unable to cultivate and harvest their farms in early 2008.  

The Kerio Valley on the north side in the scenic Rift Valley was especially hard hit by the violence. It is an area that even in the best of times is drought-prone and lacking electricity and other infrastructure.  Tribes in the valley rely on income from their livestocks and some agriculture. The Kenyan Budget Survey shows that over 60% of the population there is "food poor" or, in the words of Concern, an Irish aid agency, "less  than 40% of the population has the resources necessary to provide their basic energy requirements of 2,250 kcal per adult equivalent per day."

It was in this context that Concern Worldwide, an Irish aid organization, started an experimental program to distribute cash aid via mobile phone payments in the Kerio Valley to a select number of beneficiaries to supplement income lost in the violence and prevent famine amongst those receiving aid.  Concern now has published an extensive evaluation of the programme (PDF)

What follows is a summary of the findings and a discussion of the value of distributing cash aid via mobile money transfers in humanitarian emergencies. 

Concern also produced a short video of the prgram that, even though it it more promotional, nonetheless demonstrates how the program worked.

Cash Versus Food Aid in Emergencies

There has been much discussion of the pros and cons of cash versus food aid in emergencies.  Cash is increasingly becoming interesting to aid agencies because it enables the beneficiaries to buy largely what they need and it does not disrupt local markets the way large infusions of food would. Cash also has multiplier effects in a community, benefiting non-recipient households. For aid agencies, transporting cash is a lot less costly than transporting large quantities of food.  Transporting cash can be less secure, however, so mobile payments offer an alternative that is more secure for beneficiaries and organizations alike, as well as takes advantage of the benefits of cash instead of food aid. 

Concern's Mobile Cash Aid Program in Kenya

Concern wanted to offer temporary relief to households affected by the violence and test the efficiency of the M-Pesa mobile payment system "as a means of making cash safety net transfers to poor people living in marginal rural areas."  Concern partnered with Safaricom to to pilot the use of M-Pesa for this purpose -- the first time that M-pesa, or any mobile-phone based payment system, was used for transferring cash aid.

M-Pesa is a joint venture between Vodaphone and Safaricom that enables users to send cash from one M-Pesa user to another over the Safaricom mobile phone network.

The Concern cash aid program targeted 51 households, comprising about  3700 individuals.  These individuals were chosen by community leaders directly (rather than by Concern staff) according to a certain set of needs that the leaders agreed upon.  Beneficiaries, for example, had to have been directly affected by the post election violence and the most vulnerable groups were given priority.

Using the original food distribution lists as a starting point, a second list was drawn up ranking households in order of vulnerability. The list also identified the name of the matriarch and how many members were in each household. Community members were given the opportunity to comment on the leaders rankings in a public meeting. There was apparently little disagreement on the ranking of the households, but a fair amount of discussion on the size of households, as some had counted members who were living elsewhere. 

The matriarch of each household was registered to receive the cash and issued with an M-pesa-enabled SIM card. Concern notes, that "targeting women, combined with the relatively small size of the transfer, was effective in ensuring that women retained control of the cash. Qualitative evidence indicates that about 70% of the transfer was spent on food, with the remaining 30% on transport and other non-food essentials. No incidents of misuse of the cash were found during the evaluation."

The first distribution took place by using ‘clusters’ – beneficiaries were clustered into groups of ten and their combined transfer was sent to the phone number of the cluster leader who then collected the cash and distributed it accordingly amongst the members of the group. For the second transfer each beneficiary was issued with a SIM card and had to travel to the distribution site personally to collect their cash.  Both distributions were facilitated by M-pesa agents who travelled down to the
distribution site - for security reasons Kinyach police station - with the necessary cash.
As with a food distribution, beneficiaries usually spent the best part of a day travelling to and from the distribution and collecting their transfer. Some less mobile beneficiaries found it difficult to make the return journey to the distribution site, as they would have with a food distribution, strengthening the argument for using the ‘cluster’ system under certain circumstances

Concern calculated that each household member would require about Ksh 320 to buy their ‘basic food needs’ for two weeks. 

The Way M-Pesa Works for Cash Aid

M-Pesa is simple to use and wide-spread in Kenya, with more than 5 million users. To send money with M-Pesa, a customer needs to be registered, have M-pesa ‘activated’ on the SIM card, have the M-pesa program installed on the phone.  The sender then goes to one of 2,500 M-pesa agents throughout Kenya, deposits cash with an agent, and after a confirmation SMS then can send the money to the phone number of the recipient using the special menu on their phone. The recipient will receive an SMS alerting them that funds have arrived. If the recipient is already a registered M-pesa user, she can withdraw the cash or keep it in her account to send on to someone else or buy services or goods with it. If the recipient is not registered, she must withdraw the cash at a registered M-pesa agent.

Concern notes that,

While approximately 93 million shillings is sent through M-pesa every day, the vast bulk of this traffic takes place between urban centres which offer agents good cell phone coverage, banking facilities, a reasonable level of security and a literate clientele which both sends and receives money ensuring that a certain level of liquidity is maintained. By attempting to use M-pesa to supply cash to often illiterate beneficiaries with little or no exposure to financial or telephone services, living in areas which offer few of the preconditions for a successful M-pesa agency, Concern was taking a bold step and had to support or subsidise a number of elements of the process to ensure that the programme had a reasonable chance of success.

The first obstacle was the low level of phone ownership amongst the community-selected beneficiaries.  Concern notes that of the "571 targeted households just 225 (39%) reported owning or having once owned a phone at the time of registration on the programme."

How It Worked

Concern provided 45 phones to groups of ten households, as well as solar chargers and adapters. Local youth were deployed as 'clerks' who helped heads of households how to operate the phones and were on hand on cash distribution days to assist beneficiaries with the process of receiving the M-pesa text message.

Beneficiaries were issued the M-Pesa registered SIM cards.  Concer's evaluator writes that,

Although it was planned that each beneficiary was issued with their own SIM card from the outset, for technical reasons the first round of cash was distributed through clusters whereby groups of 10 households shared a SIM card and funds for all ten households were sent onto the one card and collected and divided up by the group member to whom the card was registered. The technical problems were eventually resolved; for the second transfer each beneficiary was given a SIM card and registered as an M-pesa user, however, in most cases they still had to wait their turn to use a shared handset at the distribution point so they could insert their card to access the M-pesa message.
Concern needed to be sure that it was sending money to the right people – one wrong digit in a phone number would result in the beneficiary not receiving his or her transfer.  To help ensure the right people received the money, Safaricom staff facilitated access to a database which allowed names to be linked to phone numbers. Before each distribution, Concern would provide Safaricom with a list of each number to which it wished to transfer funds. This list enabled the generation of a list of names which were then manually checked against Concern’s list of beneficiaries by two different people. When it was clear that all beneficiaries were associated with a phone number the payment was made.

What were the results?

The results of the evaluation are very convincing and show that cash aid transmitted via mobile phones can be an effective (for beneficiaries) and cost-effective and secure way (for aid organizations) to aid populations affected by humanitarian emergencies.  The evaluation showed that the cash was used for the purposes intended --

at least 70% of the total cash distributed was spent on food items, nearly half of which comprised the staple maize. Other food items purchased included beans, oil, tea, sugar, fruit and vegetables. The beneficiaries interviewed reported using an average of 17% of their transfer on costs related to transport. In a few instances these costs comprised fares for motorised transport but, as most journeys were made on foot, they largely consisted of payments for labour to carry food back from the market, or food purchases for children accompanying them on the journey. Around 12% of the transfer was used for other purposes including milling, paraffin, loans or gifts to friends and relatives or savings. 

There was also some impact on gender relations, as Concern and its partner organization 

placed a strong emphasis on ‘branding’ the programme as being an intervention specifically designed for women to buy food for the household. Wherever possible (identity documents permitting) the matriarch of the household was registered as the recipient of the cash.  For many beneficiaries, this was the first time the importance of their productive and maintenance roles had been formally recognised. The intervention clearly stressed on the need for women and men to negotiate at the household level on expenditure priorities and, as the women ostensibly controlled the cash transfer, provided a rare opportunity for wives to discuss with their husbands matters that impact on all their lives.
On the negative side, however, this empowerment may have served to reinforce the woman’s normal role of tending to the daily upkeep of the household. As the women were normally registered to collect the cash it was them who were burdened with the long walk to collect it and purchase food from the market, although it should be added that the women interviewed for this evaluation were in no way resentful of this. While the project probably had little impact on deep seated gender roles, women’s role as efficient managers was made more visible. Women’s confidence and self esteem was also heightened as they showed they were able to use the resources given to them for the benefit of their entire families and strengthen inter-community ties by assisting other women with gifts of food or money.
The programme, with its combination of technology and cash, also had the effect of transforming beneficiaries from passive recipients of aid to participants in a programme in which they learned new skills and were empowered with a choice of how to use the cash. Basically, the process by which assistance was delivered, while not as important as the assistance itself, was of significant benefit to participants.  

For Concern, one of the main benefits of the partnership with Safaricom was that
Safaricom took responsibility the distribution and the security of the cash up until it was handed over to beneficiaries. As the evaluation notes, "essentially Concern was able to do what few cash distribution schemes have been able to do before – that is, contract out the distribution and security of cash to a reputable institution with national presence with the option of managing much of the payments process remotely from Nairobi."

The evaluator also "found no evidence nor heard any reports of beneficiaries having their cash stolen while travelling back to their villages from the distribution point. " In fact, the evaluation notes that "beneficiaries mentioned that they felt safer travelling with cash than they would have had they been transporting food as cash is much easier to conceal."

Shortfalls

As with any project there were some shortfalls that are noteworty and important for future transmission of cash aid via mobile phone. 

The distribution of the money took place in Spring 2008 and the evaluation notes that,

Given that the post election violence took place in December and January (interviewees reported a particularly bad spell of raiding in mid January) it could be argued that the cash arrived somewhat late. However, Concern and the DoE targeted the area with one emergency food distribution in March and the situation of the affected communities by the time they received the first transfer was as bad, if not worse, than in January: Although their was relative peace, food prices had spiralled and drought had severely impacted on their ability to produce their own food.

Concern's evaluator is candid about other logistical issues that are critically important when considering larger-scale cash aid payment via mobile, especially in very remote areas: 

Two main issues affected the prompt transfer of cash from Concern to beneficiaries providing important lessons for future programming. The first was beneficiaries losing their SIM cards, the second was beneficiaries being issued with inactivated cards. Normally when a SIM card is lost the owner takes the SIM certificate to an agent who can then activate another card within a matter of minutes. The problem that beneficiaries in Kerio Valley encountered was that they could only easily access an agent on the day of distribution so had to get the new card issued on the distribution day when they had access to the agent. Fortunately the cash distribution process for the bulk of beneficiaries was not held up as the agents delegated a staff member to deal with lost cards in a separate queue.  In some instances the beneficiary had lost the SIM certificate as well as the card, in which case an entire new card had to be issued, loaded with airtime, and registered for M-pesa, a process which could only realistically take place in Iten or Eldoret where agents have access to the M-pesa database. Beneficiaries who lost their SIM and SIM certificate faced a delay in receiving payments while their new card was processed. At the time of writing this report there were still some beneficiaries who had not received their total entitlement because of this reason; it is not clear how they will get their money, but Concern is considering sending the transfer to another beneficiary on their behalf.

A unusually high number of SIM cards -- in a stroke of bad in that particular batch distributed to the beneficiaries --were unable to be activated because default settings were not remdoved.  The report notes that "normally, when a new SIM card is loaded with airtime the default call-barring settings are automatically removed allowing text messages and calls to be received. In about 50% of cases (according to the Safaricom representative interviewed) beneficiaries were issued with cards which were not activated and therefore unable to receive the M-pesa SMS sent by Concern. These cards had to be taken to a customer care centre in Eldoret for activation."

Concern also notes that while cash is well-recived and useful for many of the recipients, some may have preferred direct food aid. Particularly older recipients noted in interviews that they would have preferred food, "presumably because it would have entailed
one journey to the distribution site rather than two journeys (one to the site and a second
to the market)." The evaluation notes that

Concern’s use of M-pesa allowed a food security intervention to be executed at minimal opportunity cost to themselves. Whereas a traditional food distribution would have meant that Concern would have had to source, procure, store, transport and distribute the food, under KVCTP [the name of the program] these functions were passed on to the beneficiary. While for many households, the time costs associated with collection of cash and purchase of food were acceptable, for a minority of less able bodied beneficiaries from households with limited labour availability, they were not. Clearly solutions have to be found to make the system more appropriate to the circumstances of the most vulnerable.

A more serious development that seriously affected the effectiveness of the cash aid scheme was the rapid inflation of food prices during the time of distribution that significantly eroded the purchasing power of the money transfer.

The problem was that the transfer was based on the prices in Arror market in early April but,  by the time beneficiaries received their first distribution in early June, prices had risen considerably. For example, maize was selling for 20/= per kilo in April but 30/= per kilo in early June – an increase of 50%, while the price of beans reportedly doubled over the same period.
Given the rapid inflation experienced, the fixed-size transfer was quickly reduced to the equivalent of significantly less than 50 percent of households’ minimum monthly food requirements - a finding corroborated by most beneficiaries interviewed, who reported that the food bought with each transfer lasted for just a few days, rather than the full week intended by Concern. It can therefore be concluded that the transfer size was not sufficient to provide half of targeted households’ food needs.
The survey found strong anecdotal evidence (corroborated by interviewees from two different villages) that the increased effective demand produced by the cash transfers had the effect of raising the price of maize in Arror market – the place where most of the beneficiaries spent their cash. The extent to which vendors raised prices as a result of increased demand is not fully clear – one group in Kinyach reported that the price of maize rose from 30/= to 35/= (16%) on the first distribution day, while in Ayatia interviewees claimed that prices were increased by 60% (from 50/= to 80/= per kilo). However, it is unlikely that the 60% increase reported is entirely attributable to dealers taking advantage of the surge in demand – while market inefficiencies certainly contributed to some of the inflation, the larger proportion of was a reflection of the global rise in grain prices over the project period.  

Concern found that for its operation, the overall projects costs where largely the same whether wholesale good is distributed or whether cash aid is dispensed via mobile payments.  The evaluator estimated that with a food distribution beneficiaries would have received an additional two days of minimum calorific requirements – a result of buying the food wholesale and the costs associated with equipping beneficiaries to receive funds through M-pesa.

The evaluator notes that

even when the difference between wholesale and local market prices is accounted for, there is a point (relating to the relative costs of cellphone equipment and food transport) at which transferring cash through M-pesa represents better value in terms of calories provided per shilling spent than a traditional food distribution. In the case of KVCTP, however, the savings possible through longer-term programming were never realised because of the short duration of the pilot.
In terms of providing calories for the amount of money available, KVCTP was marginally less effective than a traditional food transfer would have been. The two reasons for this are: under a food transfer, staples would have been bought wholesale while beneficiaries were buying retail from inefficient markets at a time of rapid inflation and the cost of equipping beneficiaries to receive M-pesa. This is not to say that using M-pesa for delivering cash transfers should be written off; when the costs of equipment are discounted the cash transfer option proved to be more effective than a food distribution in terms of delivering calories.

The most important recommendation for future projects, of course, is aid agencies deploying mobile cash aid transfer link payments to the price of a basket of staple foods. in the case of Concer, early warning data was not considered when the organization established the size of the cash transfers; the report notes that "there were indications as early as February that prices were set to rise, but these appear not to have been considered when setting the size of the transfer."

The entire evaluation of the Concern program in Kenya is here in PDF format. 

Photo: Screenshot courtesy of Concern Worldwide video/YouTube. 

AttachmentSize
KenyaCashTansferPilot-EvaluationReport-July08.pdf392.89 KB

Post new comment

The content of this field is kept private and will not be shown publicly.
  • Allowed HTML tags: <a> <em> <strong> <cite> <code> <ul> <ol> <li> <dl> <dt> <dd><p><br> <b><i><blockquote>
  • Lines and paragraphs break automatically.

More information about formatting options