Fitting The Pieces of The Liquidity Management Puzzle: Nancy Kiarie, Ian Odongo, and Vera Bersudskaya

Fitting the Pieces of the
Liquidity Management Puzzle

Nancy Kiarie, Ian Odongo, and Vera Bersudskaya
www.helix-institute.com
Fitting the Pieces of the Liquidity Management Puzzle
This paper is part of a series of synthesis papers that summarise data on agent networks. We have collected the
data over four and a half years from nine countries in Africa and Asia through the Agent Network Accelerator
(ANA) project1. MicroSave’s Helix Institute of Digital Finance implemented the ANA project, with funding from
the Bill & Melinda Gates Foundation, the United Nations Capital Development Fund (UNCDF), Financial Sector
Deepening – Uganda (FSDU), and Karandaaz Pakistan. This paper synthesises knowledge and data on liquidity
management approaches to ANA research markets and beyond.
The past decade has seen the proliferation of Digital Financial Services (DFS)
The past decade
deployments across the globe. Stakeholders have different objectives for engaging in
has seen the
DFS, from extending financial inclusion for the unserved or underserved populations
to cost-cutting and diversification of revenue streams. Agent networks have become
proliferation of
the channel of choice for delivering financial services to customers, as they present Digital Financial
a low-cost alternative to brick-and-mortar solutions and use existing bank branch Services (DFS)
networks or Mobile Network Operator (MNO) distribution networks, or both. deployments
across the globe.
A key challenge to the credibility and sustainability of agent networks is ensuring
adequate liquidity – in the form of sufficient cash and e-float to facilitate transactions.
While this is difficult, DFS providers, master agents, and agents have devised innovative ways to support liquidity
management. This paper draws on Agent Network Accelerator Surveys across Pakistan, India, Indonesia, Bangladesh,
Kenya, Uganda, Tanzania, Zambia, and Senegal, which used both quantitative and qualitative instruments. It discusses
what liquidity management entails, the implications of poor liquidity management, the existing models of liquidity
management and their successes and shortcomings.
1. Defining Liquidity Management

This paper defines liquidity management as all the activities involved
in maintaining sufficient e-float and physical cash at the agent outlet,
to perform cash-in/cash-out transactions. The distinction between
cash management and e-float management is important.
E-float refers to the balance of electronic money (e-value) present

in an agent’s wallet or account that is used to process customer
transactions. When a customer makes a deposit (cash-in), the agent
transfers e-value from their wallet or account to that of the customer
in exchange for an equivalent amount of physical cash. When a
customer withdraws money (cash-out), the agent issues physical cash
to the customer in exchange for an equivalent amount of e-value.
E-float management is the process of ensuring that the amount of

e-value present in the agent’s wallet (on the agent till) is sufficient
to process customer deposits. Cash management refers to the steps
involved in obtaining physical cash, either in exchange for e-value
or as an amount dedicated to the agent business, used to perform
customer withdrawals. This is commonly referred to as ‘re-balancing’.
1. See Appendix A for further detail on the data and countries covered.
2 www.helix-institute.com
2. Describing Liquidity Management in ANA Countries
The challenges that agents encounter in liquidity management are similar across the markets studied. Figure 1 suggests
that agents struggle most with unpredictable fluctuations in client-demand, time spent on rebalancing activities (when
delivery of cash or e-value is not an option), and lack of capital.
Figure 1. Number of ANA countries (out of nine) reporting as top three barriers to liquidity management
Unpredictable fluctuation in client demand 7
Long wait at rebalancing point 6
Lack of resources to buy enough float 4
Need to shut store to rebalance 2
In East Africa, longer travel times to rebalancing locations are statistically associated with less frequent rebalancing. This
is because agents, or their staff, may have to shut shop and incur travel expenses, rebalancing fees, and tips during the
rebalancing process. The rebalancing costs in proportion to the monthly revenues of agents are the highest in Indonesia
and Pakistan (Figure 2). This is primarily due to the relatively low revenues of agents.
Figure 2. Rebalancing costs as a percentage of agents’ monthly revenues
Indonesia 9.0%
Pakistan 6.9%
Zambia 5.4%
Tanzania 2.0%
Bangladesh 0.8%
Kenya 0.2%
Uganda 0.1%
Senegal 0.1%
Agents in Zambia, Kenya, Senegal, and Indonesia primarily use banks to rebalance. In contrast, agents in Tanzania,
Uganda, Bangladesh, and Pakistan also use master agents or aggregators as well as fellow agents to rebalance. In the
former countries, agents generally rebalance every third day, compared to every other day in the latter.
DFS practitioners often assume that agents need more cash than e-float in rural areas on a presumption of higher
demand for withdrawals among rural residents who receive remittances. However, Figures 3 and 4 demonstrate that
with the exception of Indonesia, the demand for e-float is higher in both urban and rural areas – although it is easier
to rebalance in urban areas due to ease of access to rebalancing facilities. Higher rates of agent non-dedication in rural
areas in Zambia and Kenya help explain why fewer agents there struggle to get cash – as agents can dip into the cash
generated by the parallel business when needed. Agent liquidity management-needs appear similar across locations,
with agents needing assistance to convert cash into e-float.
Figure 3. Percentage of rural agents, by type of liquidity they usually need when rebalancing
E-float
43% Cash
Pakistan 15%
42% Both
44%
Bangladesh 34%
22%
26%
Tanzania 24%
50%
34%
Uganda 14%
52%
64%
Kenya 12%
25%
41%
Zambia 24%
35%
46%
Senegal 19%
35%
22%
Indonesia 44%
34%
Figure 4. Percentage of urban agents, by type of liquidity they usually need when rebalancing
E-float
41%
Pakistan 15% Cash
44%
Both
54%
Bangladesh 24%
22%
24%
Tanzania 23%
53%
27%
Uganda 16%
57%
64%
Kenya 7%
29%
45%
Zambia 24%
31%
39%
Senegal 21%
39%
59%
Indonesia 17%
24%
Agent non-exclusivity further complicates the agents’ task of maintaining sufficient levels of e-float and physical cash for
the agent business (Figure 5). We explore the implications of managing float across providers in section 4.
Figure 5. Non-exclusivity rates across ANA countries
Pakistan ‘17 78%
Tanzania ‘15 70%
Senegal ‘15 66%
Uganda ‘15 64%
Bangladesh ‘16 56%
Kenya ‘14 13%
Zambia ‘15 9%
Indonesia ‘17 4%
3. Impact of Failed Liquidity Management

Liquidity management is a challenge for providers and agents alike, in new as well as established deployments, across
markets. Ineffective liquidity management has a broader impact on the sustainability of agent deployments. The
following results are from the ANA research reports, and are further corroborated in MicroSave’s briefing notes – Agent
Dormancy: Impact on Customers and Agent Dormancy: Reasons and Remedial Measures.
3.1 Denial of Transactions and Reputational Risk
Agents who fail to manage liquidity effectively are forced to deny transactions when they do not have either e-float for
customer deposits or physical cash for withdrawals or both. In Tanzania, agents report denying as many as 20% of the
daily transactions conducted (Figure 6). At times, rather than admit that they are not able to facilitate a transaction
for want of liquidity, agents tell customers that the provider systems are down, which poses a reputational risk to the
provider.
Figure 6. Percentage of daily transactions denied due to lack of float
Tanzania 20%
Uganda 13%
Zambia 12%
Indonesia 11%
Senegal 8%
Pakistan 7%
Bangladesh 5%
Kenya 3%
As alluded to above, non-exclusive agents have a harder time managing

liquidity across the different providers they serve. Anecdotal evidence
suggests that non-exclusive agents may allocate more money to transacting
with their favourite provider, with provider loyalty usually driven by
commissions earned and the support services provided. If this general
wisdom were true, non-exclusive agents would be expected to deny, on
average, a greater share of their transactions. However, the data in Figure 7
shows that exclusive and non-exclusive agents across the research markets
generally deny a similar percentage of transactions for a given provider. One
explanation for this could be the e-float management arrangements of some
master agents and aggregators that manage non-exclusive agents are able to
move e-float between providers at the agent-level.
Access to provider data would enable further analysis of liquidity

management. However, data alone may not resolve the issue of lack of
e-float. This is because typically, providers, master agents, and aggregators
monitor e-float levels on a real-time basis and provide SMS notifications
to agents when float balances are low. However, these monitoring systems
can only capture data on transactions rejected through insufficient liquidity,
not the potential transactions that the agent rejects before the transaction
is attempted. The data in figure 7 is derived from quantitative analysis of
information that agents have provided.
Figure 7. Percentage of daily transactions denied due to lack of e-float, by exclusivity
Exclusive
4% Non-Exclusive
Kenya
1%
6%
Bangladesh
4%
Pakistan 9%
7%
7%
Senegal
9%
Indonesia 11%
11%
Zambia 12%
12%
Uganda 13%
13%
Tanzania 21%
20%
3.2 Reduced Agent Profitability

A paper from MicroSave’s Helix Institute, Successful Agent Networks, highlights sustainability as one of the key
dimensions of the success of an agent network. Sustainability is determined by how profitable the business is to the agent
and their motivation to remain an agent. Across virtually all ANA countries, more frequent rebalancing is associated with
higher agent business turnover.2 Similarly, in most markets, agents who deny a higher percentage of their transactions
are statistically less profitable3. Because transaction volumes drive agent profitability, liquidity outages have a negative
impact on the sustainability of individual agent businesses and the agent network as a whole. The more transactions an
agent denies, the less profit they receive, which in turn undermine the agent value proposition.
3.3 Loss of Customer Trust

When an agent who does not have enough cash or e-value turns away customers, it undermines the customers’ trust in
the system. Building trust in DFS is a major hurdle that financial service providers have to overcome to drive adoption
and usage. This is particularly true for developing countries where smartphone prevalence is low and customers are
most comfortable with face-to-face interactions. Repeated denials lead customers to avoid ill-reputed, illiquid agents,
who then fall dormant due to insufficient demand.
3.4 Agent Workarounds

In five ANA markets, we found that agents who conduct higher value transactions rebalance more often. When faced
with liquidity hurdles, agents devise workarounds to maximise the number of transactions conducted. In Kenya and
Tanzania, agents who deny transactions do more transactions on an average than agents who do not deny any – so
agents who are busier may have greater problems managing liquidity. Some agents resort to transaction revenue honing
and transaction pooling. Agents practice transaction revenue honing where commissions are tiered into bands. They
encourage customers to transact at the lower end of each commission-based tier – thus maximising their commissions.
Transaction pooling is a practice where agents will pool among themselves and provide each other either float or cash,
or both, when required.
3.5 Compromised Digital Innovation

A lack of liquidity can compromise digital innovation through a downward spiral of trust, where few transactions lead
agents to reduce liquidity, which leads to still fewer transactions and less liquidity. This then reduces transactions and
income, and can lead providers to under-invest in the systems to support DFS innovation and management, when in fact
they actually need to invest more in managing liquidity, downtime, and unauthorised fees, to encourage usage. Graham
A.N. Wright has noted this in his blog ‘Can Digital Financial Services Turbocharge Financial Inclusion?’.
3.6 Reduced Agent Investment

More sophisticated DFS platforms allow providers to support agents with information on when they will need to rebalance
and, in some cases, facilitate the delivery of liquidity to the agent outlet. In the absence of these sophisticated liquidity
support structures from the provider, agents have to maintain more float – thus requiring significant investment and
effort to manage liquidity in order to be transaction-ready for customers. This, in turn, reduces the return on investment
from the DFS business. When the returns on these investments are not very attractive, agents will divest to other more
lucrative ventures, with lower operational risks and cost. In ANA markets, agents who have been in operation longer
in general hold significantly more float, perhaps either because those who have not grown the business have reduced
investment and eventually churned out, or because those with greater float enjoy higher transaction volumes and profits,
or both. This can be seen in Figure 8.
2. With the exception of Indonesia.

3. With the exception of Kenya and Senegal, where the relationship is not statistically significant.
Figure 8. Average amount of e-float held by ANA agents (USD)
Senegal 1,517
1,031
1,473
Zambia
1,080
Kenya 1,325
770
Uganda 885
592
Tanzania 910
567 More Experienced (4 yrs)
Less Experienced (1 yr)
Qualitative observations through ANA research reveals that Asia (India, Pakistan, and Bangladesh) has agents who
make more commissions because the businesses are mostly operated by the owners themselves who are able to invest
more, without fear of staff stealing the money.
4. Why is Liquidity Management so Difficult?

An effective liquid network offers rebalancing that is: (i) cheap,
(ii) fast, (iii) frequent, and (iv) available on the basis of predicted
client demand.
This section describes factors that make it challenging for

providers and agents alike to achieve optimal liquidity
management.
4.1 Whose Responsibility is It?

One of the reasons liquidity management remains a challenge
is the lack of clarity on the responsibility of managing liquidity.
Mobile money and agent banking guidelines that govern
agent network operations provide no instructions on liquidity
management for DFS stakeholders. Instead, regulators focus
heavily on reducing risk through agent screening criteria and
processes, and on consumer protection through disclosure and
process. In Kenya, current policy limits the options to provide
liquidity management assistance through agent banking – a
financial institution is unable to transport cash directly without an armoured vehicle, which prevents liquidity delivery
to agent locations4. However, the delivery of cash is popular in Bangladesh and Pakistan and has been increasing in
other countries. Figure 9 shows the growing trend in liquidity delivery.
4. An order by the then Internal Security Minister in Kenya, John Michuki, following a spate of robberies on cash in transit vans as reported in the Daily Nation, 9th January, 2007.
Figure 9. Liquidity delivery to ANA agents
Rebalance at a Bank
Kenya ‘14 78% Liquidity Delivered
2%
76%
India ‘14
6%
31%
Tanzania ‘15 40%
Uganda ‘15 27%

52%
6%
Pakistan ‘17 93%
Bangladesh ‘16 1%
99%
Beyond the providers, agents in different countries draw from different sources of float or cash. This can be through
‘super-agents’ who manage liquidity locally for a transaction fee, or through ‘master agents’ who manage chains of
agents for a share of commissions.
In Uganda, and increasingly in Tanzania and India, master agents deliver cash to high-performing agents in urban
centres. Note that the terminology, ‘master agent’ and ‘super-agent’ is not used consistently from country to country.
In this paper, ‘master agents’ refers to third-parties who manage chains of agents and ‘super-agents’ refers to agents who
have liquidity and float that other agents may purchase (see 4.5 Liquidity Sources).
Nor are the roles of master agents standardised across markets. A blog by Jacqueline Jumah, titled ‘De-mystifying the
Role of Master Agents’ explores these interesting distinctions. Under this model, the responsibility for float management
falls to the master agent, who often deliver cash to their best performing agents. While this allows agents to concentrate
on conducting transactions, they pay for this service through commission share. Note the amount of float held is higher
in markets that are not interoperable, as noted in Section 4.6.
4.2 Incentivising Liquidity Management

Providers can reward agents who consistently meet the standards of float management. While we cannot measure the
availability of cash at a specific point in time, we can see the adequacy of cash over time by examining the number
and value of withdrawal transactions. An instance of an agent who handles fewer or lower value transactions is a clear
indication of cash problems, especially if nearby agents are conducting higher volumes. Incentives can include interest
earned on float or one-time bonuses.
Another strategy is the education of agents by providers. Agents without adequate training or knowledge of how to
operate an agency business are often managing outlets. Providers, therefore, need to educate agents on the both the
importance of maintaining, and how to maintain, adequate float.Providers should highlight in their training and
monitoring visits that agents who study their need for float as the business grows can serve more customers and thus
earn higher commissions.
4.3 Seasonality of Demand

Agents experience high demand during holidays, school
opening, market days, weekends, social transfers etc.,
as noted in the MicroSave blog ‘The Ebbs and Flows of
Liquidity Management’. During periods of high demand,
agents experience more frequent issues with insufficient
float.
This results in denial of transactions and agents incurring

high rebalancing costs as they try to address the demands
for liquidity. At the agent-level, agent tracking of agency
business trends is crucial to enable an agent to know
whether they are in a deposit or a withdrawal zone.
Tracking of transaction trends in a month helps agents know

the time/days to have what type of liquidity to cater to their
customers’ demands. In turn, providers and master agents
should push their analytics further to predict agent liquidity
requirements based on historical transactions data.
Armed with this information – providers or master agents

must plan for high demand through encouraging higher agent float levels around peak periods, through pushing SMS
reminders, and through reward structures.
Visits by master agents and provider or third-party staff appointed to monitor agents, armed with this information, can
encourage better float management. Mechanisms to provide float include linking e-float accounts to bank accounts,
providing float on credit, and in the case of NovoPay in India, using advanced GIS systems to match agents with nearby
officers who can provide top-ups.
4.4 Agent Recruitment and On-boarding

Many providers have instituted minimum investment requirements, as well as minimum e-float balances, to ensure that
their agents have sufficient capital to transact.
However, the drive to get agent numbers, particularly if delegated to master agents, has often led to recruiters cutting
corners and on-boarding agents who are unable to maintain enough liquidity to sustain their agency operations.
Since induction training is often largely focused on the operational aspects of running the agency business while leaving
out the technicalities of float management, many providers struggle to ensure liquidity in their networks.
4.5 Liquidity Sources and Rebalancing Costs

While banks, followed by provider outlets, remain the primary source for liquidity in the ANA markets, some providers
have agreements with either other financial institutions or cash-rich businesses, or both, to provide liquidity in the
areas that are located far away from banks.
These cash-rich businesses may include supermarket chains or gas stations. However, incentivising rebalancing points
to serve agents requires them to charge for the service, which discourages the agents from using it. Some providers
have chosen to subsidise rebalancing costs for their agents, which has an impact on the bottom-line of their DFS
deployments.
4.6 Non-interoperable Agent Models

In Bangladesh, 94% of agents are non-dedicated, and 56% are non-exclusive. In turn, 22% of the non-exclusive agents
were dormant for at least one of the providers that they served. The ANA surveys found that non-exclusive agents had to
hold float for each provider. For some, this was a challenge.
Thus non-interoperable systems result in a proportionate increase in the level of liquidity held by agents (Figure 10).
Among the ANA countries, only Tanzania has a formalised interoperable system that allows agents to share e-float or
transfer e-float held between different provider wallets. However, the Uganda Bankers’ Association has been working
on a shared agents’ initiative.
Figure 10. Average total amount of e-float and cash agents hold on a daily basis (USD)
553 Exclusive
Tanzania
736 Non-Exclusive
Uganda 592
674
Kenya 1,064
1,123
860
Zambia
917
801
Senegal
1,170
5. Liquidity Management Models

Agent network deployments in ANA countries have adopted different models to facilitate liquidity management. These
vary depending on whether agents use their own money for operations, their rebalancing method and channel, whether
rebalancing is instant, and on the support that agents receive in managing liquidity.
5.1 Owning Float

While DFS providers often impose minimum float requirements as part of agent on-boarding, and operational guidelines
to ensure that agents hold sufficient working capital for the agency business, agents are not always expected to own
their liquidity. As mentioned, in some operations, agents make use of float or cash from master agents or super agents.
However, an advantage of having the agent take ownership of liquidity is it gives them full control of their operations
and the autonomy on when to rebalance.
When an agent owns the liquidity, the recruitment process is key to ensuring that on-boarded agents have the capacity
to raise the necessary working capital. For example, Safaricom requires an existing business (providing six months’
bank statement) to operate M-PESA. An agent must start with three outlets and each is required to invest KSH 100,000
(USD 1,000). Thereafter, outlets are required to hold balances ranging from USD 1,000 to USD 5,000 according to the
volume and value of transactions they are conducting. Another provider in Kenya also requires agents to hold varying
e-float balances ranging from USD 1,000–USD 5,000 daily, depending on their location (rural or urban). Providers have
faced challenges with enforcing this method, as agents may borrow the funds initially to satisfy the setup requirement
and subsequently may not sustain the required liquidity balances.
5.2 Delivery as Opposed to Pickup

In ANA markets, some providers offer on-demand or routine liquidity deliveries, while others require agents to travel
for rebalancing (Figure 9). Deliveries are trickier for cash than for e-float due to security and cost concerns.
Figure 11. Percentage of agents who report getting liquidity delivered to their outlet
Bangladesh 98%
Pakistan 95%
Uganda 52%
Indonesia 45%
Tanzania 40%
Senegal 33%
Zambia 20%
Kenya 2%
In highly non-dedicated markets like Bangladesh, Pakistan, and Indonesia, where DFS is an add-on to existing
businesses, providers have generally sought to facilitate agent rebalancing. In Pakistan, one provider has made use of
the existing support teams from the airtime distribution network to facilitate rebalancing for the agents.
In Bangladesh, distributors conduct regular liquidity deliveries and are available on-demand. In Uganda, providers and
master agents use ‘liquidity runners’ to deliver float to some of the larger and more accessible agents. However, master
agents interviewed in Uganda expressed their fears of loss in using runner services to deliver cash.
Providers also assist select agents in Senegal and Zambia. Generally, however, agents are expected to pick up liquidity
for their operations from rebalancing points, identified by the provider. The agent incurs the cost of transport to the
rebalancing point and the charges if any, to obtain liquidity.
5.3 Instant as Opposed to Delayed Rebalancing

Agent cash-outs are generally immediate. However, obtaining e-float can be instant or delayed, depending on the
integration of platforms between a provider and partners that provide rebalancing support. For example, one bank in
Kenya requires agents to operate using an agent account, through which they can conduct transfers from a regular bank
account into the agent wallet, thus enabling instant transfers.
When an MNO partners with a bank or other institution to facilitate rebalancing, agents may experience delays in
obtaining e-float or cash due to lack of integration of platforms. In nascent DFS markets like Nigeria, bank agents can
also experience delays with reconciliations effected at the end-of-day. At times, the processes may even be manual. In
Malawi for instance, a bank partnership with an MNO requires agents to send a picture of the receipt to the provider
after making a deposit to receive e-float on their wallets. Such cumbersome rebalancing procedures increase the costs of
liquidity management for agents and providers alike.
5.4 Self-rebalancing as Opposed to Credit

Increasing non-dedication rates across most ANA markets point to the advantages of running DFS operations as an
add-on to a parallel business (Figure 12). From the standpoint of providers, non-dedication reduces the pressure on the
provider to ensure hefty commissions and enables smoother cash-management at the agent location.
Figure 12. Percentage of non-dedicated agents (running DFS operations in parallel to another business)
Wave I
Bangladesh 96%
Wave II
94%
Pakistan 77%
96%
Uganda 55%
37%
Kenya 54%
64%
29%
Tanzania
57%
The exception is Uganda, where dedication has increased with the

introduction of non-exclusivity. With just two major mobile money
networks and well-developed master agent networks that actively
manage liquidity in urban areas, it is common to see booths or small
shops dedicated exclusively to mobile money in the country.
Non-dedicated agents can rely on their existing business to facilitate

rebalancing, as cash from parallel sales can be used for customer
withdrawals. Adding on a mobile money operation entails few marginal
costs other than that of rebalancing. It, therefore, provides an attractive
way to diversify income, increase the foot-traffic, and cross-sell products
and services to the clientele.
Some agents receive liquidity management support through self-sourced

or provider-funded credit facilities. While some banks, such as KCB and
Equity Bank in Kenya provide credit facilities for their agents, more
mobile operators have been implementing overdraft facilities (in some
cases with digital credit providers like Jumo) to bridge the gap between
agent rebalancing trips. Usually, agent performance is the key qualifying
factor for this credit.
This is unfortunate, since the agents most affected by liquidity shortages

and those in remote areas with high rebalancing costs would be least
likely to receive this support. The key concern among providers is to
ensure that the agents use the funding given to them for agent operations
and do not redirect the funds to other uses.
6. Five Innovative Solutions to Liquidity Management

From the challenges identified through the ANA surveys, what are the potential innovations which can reduce the
liquidity challenges that agents face?
6.1 Innovative Agent Platforms

Centralised monitoring systems can help identify agents who consistently fail to hold adequate liquidity. Alerts can then
be sent to agents whose float levels have dipped below a recommended level to encourage rebalancing.
These platforms could facilitate a variety of rebalancing mechanisms. These include rebalancing at ATMs, as well as
through inter-agent transfers, where agents can ask for and receive e-float from fellow agents. Agents may also choose
to deposit or withdraw money from their personal account into the float account remotely without involving the bank.
Master agents have already been doing this informally through WhatsApp groups set up by them to manage their agent
networks. If providers are able to monitor these activities, they could monitor compliance and define standard operating
procedures for their agents.
6.2. Uber-isation of Agents

Building on the ideas around inter-agent transfers, a recent MicroSave/Helix workshop discussed the potential to
reduce the dependence on agents by empowering almost every customer to act as cash-in/cash-out points. This, of
course, is already being done on an informal basis across the globe – particularly in remote areas that agents, who
formally receive support from providers, are unable to serve effectively. Often, local business people or community
leaders provide services to convert cash into e-value or vice-versa in an informal manner for a small commission. Using
such an approach would mean an increased network of CICO points as well as reduced agency management costs for
providers. Customers would benefit from the convenience of proximate services.
In ‘Reimagining The Last Mile – Agent Networks in India’, MicroSave

highlighted that “fintech companies can come up with smartphone
Eko Financial Services Pvt. Ltd
applications that enable any user to act as liquidity merchants has launched an app called Fundu,
― mimicking what Uber has done for transportation. Similar which is being geared up for a
initiatives can help address cash-needs in the ‘last few hundred pilot-test in Kenya. “This app
yards,’ while agents provide the ‘last-mile’ backup, underpinning a
will allow you to act as an ATM…
more decentralised cash market.”
Whenever a Fundu app user near
The elegance of this solution is that the users do not need to meet or
you needs cash, you will get a
even know one another. The shortcoming is that there is no scope for
the exchange of physical cash.
notification. If you have cash and
are willing to provide it, you can
Our discussions with industry stakeholders show concerns for
the security of agents who currently operate on such ‘Uber-ised’ accept the request.” The individual
solutions. Their concerns are that if agents highlight that they have will transfer the money to the
cash at their outlet, it is an invitation for robbery and/or fraud, both user’s bank account using his or
of which have been growing at an alarming rate. her virtual address. – LiveMint.
5. This section is an extract from a MicroSave blog ‘Liquidity – solving agents’ perennial problem’ developed from insightful inputs from a group of industry practitioners.
6.3 Use of Data Analytics to Predict Demand

Data analytics could be used to monitor transactions and facilitate liquidity management based on historical experiences
and trends. This idea is built on a recommendation made by The Helix Institute in 2014 and by Harvard Business School
in 2017. Using the DFS platform data to identify trends in agents’ demand for e-float or cash will assist in planning
for peaks and troughs. The platform itself could automatically share this information with agents and master agents
to assist them in maintaining adequate levels of liquidity. The analyses would need some modifications to account for
unusual and exceptional events that create spikes in demand for liquidity, such as general elections, large sporting
events, and intermittent bulk payments like remittances to refugees or government subsidy transfers.
Nonetheless, regular SMSs to agents that predict the likely demand for liquidity on a monthly, weekly, and daily basis
would help them to plan better. It would also inform the kind of support needed by providers and master agents, such
as facilitating e-float overdrafts for agents (see below) or organising cash pick-up or drop-off at agent outlets. Providers
can also use this data to monitor agent activity, which will help identify unusual or fraudulent practices, such as remote
deposit, split transactions, and float-hoarding.
6.4 Credit to Allow Agents to Access Working Capital

Agents often cite lack of resources or working capital as the key impediment to financing their liquidity requirements.
These impediments are sometimes (but not always) temporary, as a result of seasonal fluctuations. While few mobile
network operators are willing to take the risk of lending to their agents, extending e-float on credit provides a significant
opportunity to improve liquidity and enhance agent loyalty. If lenders use methods like data analytics, they would be
able to predict liquidity-needs and assess the past performance of agents. This should allow lenders to significantly
reduce the risk inherent in offering credit to agents.
Furthermore, a system that provides agents with e-float overdrafts to allow them to rebalance using their mobile phones
could unlock significant value. It would also reduce the number of transactions declined for want of liquidity. Safaricom,
for instance, offers their premium M-PESA agents short-term weekend/public holiday financing to meet their liquidity
requirements. This not only boosts the availability of float but also increases the number of agents working over the
weekends when banks and other super agents are closed. A few banks, such as Commercial Bank of Africa and Kenya
Commercial Bank, are already taking steps towards this. However, given the sophisticated data analytics and credit
platforms required in the process, fintech companies may be best-suited to provide these lines of credit.
6.5 Set-up Digital Ecosystems

A core focus should be on deepening digital ecosystems. Digital ecosystems consisting of open APIs and fully interoperable
platforms would facilitate and encourage the use of digital payments. This can reduce demands to cash-out and the need
for agents to rebalance. Similarly, when FMCG suppliers insist on payment for supplies in e-value rather than cash, it
can help rural agents use the e-float they accumulate.
We can achieve high-functioning digital ecosystems only if all the players collaborate to increase opportunities for
additional digital transactions. Effective liquidity management is key to any trusted and successful agent network. Yet
the much-vaunted challenges are all manageable, particularly if providers make use of the capabilities of fintechs and
data analytics.
Appendix A: Agent Network Accelerator (ANA) Studies

The Agent Network Accelerator (ANA) project is managed by MicroSave, with funding from the Bill & Melinda Gates
Foundation, the United Nations Capital Development Fund (UNCDF), Financial Sector Deepening – Uganda (FSDU),
and Karandaaz Pakistan. It is the largest research project on agent networks in the world that aims to increase the global
understanding of how to build and manage sustainable cash-in/cash-out (CICO) networks in poor communities and
identify factors that drive their success or failure. The research is designed to distil the most salient aspects of strategic
operations in agent network management for the DFS industry, including agent network structure, agent operations,
agent viability, liquidity management, quality of provider-support, and agent compliance.
MicroSave’s Helix Institute of Digital Finance launched the project in 2013. Since then, The Helix has conducted over
38,700 agent interviews in 11 countries, providing assessments to over 40 leading agent networks around the world.
We carried out quantitative assessments in countries where the population of active agents exceeded 10,000 according
to recent and reliable data. Where networks were nascent, the team carried out qualitative strategic assessments,
interviewing providers, agents, and other DFS stakeholders (See Table A).
Table A. Study type and sample size, by country and year of data collection
Year
Country
2013 2014 2015 2016 2017
Quantitative Quantitative
Bangladesh
(2,841)* (2,309)*
Benin Qualitative
India1
(4,437)* (3,199)*
Quantitative
Indonesia Qualitative
(1,383)*
Kenya
(3,220)* (4,126)*
Nigeria Qualitative Qualitative
Pakistan
(3,151)* (2,563)*
Quantitative
Senegal
(1,639)*
Tanzania
(2,052) (2,066)
Uganda
(2,028) (2,288)
Quantitative
Zambia
(1,350)*
*Includes booster sample for key providers. Outside Tanzania and Uganda, core random samples were ‘boosted’ with additional interviews for
specific providers in order to obtain statistically relevant sample size.
1. Second wave India data was being finalised at the time of paper writing. Because the Indian market underwent a dramatic transition following the demonetization of INR 500 and INR 1,000
denomination banknotes, papers do not present data from 2015 as it has lost relevance.
While MicroSave’s Helix Institute of Digital Finance directly conducted the qualitative agent network strategic
assessments, The Helix managed the quantitative studies with data collection outsourced to local data collection and
management firms.
Between 2013 and mid-2015, data collection, quality control, data cleaning and analysis were outsourced to the local
survey firms. The Helix provided the survey teams with the core ANA questionnaire which was administered using
Computer Assisted Personal Interviewing (CAPI)2. From September 2015, the survey was streamlined to reduce the
number of questions and in-house most of data quality control, data cleaning procedures, as well as all data analysis.
Across all countries, we designed the ANA surveys to be nationally representative at the country, rural/urban, and
provider levels. The study methodology varied slightly from country to country depending on the agent population data
available and which The Helix and the local survey firms were able to obtain. In Kenya, Tanzania, and Uganda, we used
agent censuses conducted by BrandFusion as sample frames for the studies. In other countries, The Helix compiled
publicly available data on agent locations and solicited agent lists from the countries’ leading providers.
The sampling strategy in all countries was two-stage stratified cluster random sampling, with administrative units being
stratified by region and rural/urban classification3, then drawn at random. Agents are subsequently sampled from the
randomly chosen administrative units in proportion to the agent population. In markets where agents serve multiple
providers, agents were interviewed about their operations for a provider, randomly selected from the list of all providers
for whom the agent has conducted at least one transaction in the preceding 30 days.
Each study was analysed to produce publicly available country reports4, which contain essential information about the
performance of agents and providers who manage them. Leading DFS providers also received confidential reports with
business intelligence comparing their network to competitors. In addition to country and provider reports, MicroSave’s
Helix Institute of Digital Finance has synthesised ANA data to enhance industry understanding of best-practices and
benchmarks for building and managing agent networks across the globe in blogs as well as the following publications5:
• Designing Successful Distribution Strategies for Digital Money helps providers understand their goals for building
an agent network. It subsequently helps them think through the model of building an agent network that best fits
their needs.
• Successful Agent Networks builds on the understanding that networks are the channel providers used to deliver
distinct value propositions to different customer target groups. It lays out a comprehensive analytical framework for
analysing agent network success along several key dimensions.
• Agents Count: The True Size of Agent Networks in Leading Digital Finance Countries lays out a framework for
understanding agent network size, drawing the distinction between agent tills and agent outlets. It also discusses
agent activity rates and calculates customer to agent outlet ratios, providing updated benchmarks for the industry.
2. ANA questionnaires were adjusted to capture market specificities, while preserving the core of the survey.
3. National census rural and urban classifications were used in Pakistan and Indonesia. In Africa, larger and densely populated regional, provincial and district centres are classified as “urban”
whereas sub-districts or locations outside major districts are classified as “rural”. Similarly, in 4Bangladesh, Thana and Village Headquarters are classified as “rural” with eight divisional
headquarters and districts classified as “urban”.
4. Tanzania Country Report based on 2015 data remained unpublished due to the Tanzanian government’s restrictions on conducting
nationally representative surveys.
5. MicroSave’s Helix Institute of Digital Finance has also authored the following landmark pieces on DFS product and business model evolution:
• Finclusion to Fintech: Fintech Product Development for Low-Income Markets This paper is designed to help fintech innovators understand the unique money management strategies used
by low-income people in the developing world. It summarises insights from 15 years of financial inclusion research and suggests how cutting-edge technological innovation in the fintech
industry could better serve developing world markets.
• Redesigning Big Data for Digital Finance This paper proposes important strategies that digital finance providers (mobile network operators [MNOs], banks and third parties) should adopt to
manage the influx of fintech (technology firms) players into the developing world. It argues that to compete or collaborate with fintech players, providers need to augment their customer data.
• OTC: A Digital Stepping Stone or a Dead-end Path? discusses the pros and cons of Over the Counter (OTC) transactions and argues that they should be seen as a stepping stone to mobile
money account adoption and use.
This compilation of papers draws on the rich ANA data, with the exception of India, to benchmark agent training and
support, liquidity management strategies, as well as risk levels across agent networks. The compilation also takes
into account the framework presented in the Successful Agent Networks paper. Each paper uses a distinct analytical
approach:
• Benchmarking Agent Support classifies 27 leading providers into three groups, according to the providers’
agent network management approach: direct, indirect, or hybrid. It further analyses trends between Wave I data
collection (conducted 2013–2014)6 and Wave II data collection (conducted 2015–2017). Slight variations in data
collection approaches across markets as well as differences in levels of market maturity constitute the methodological
limitations of this analytic approach. Nonetheless, we believe that the data offers interesting, even if indicative,
evidence on the levels of training and support each agent network management models can achieve as well as the
effectiveness of agent training and support.
• Fitting Pieces of the Liquidity Management Puzzle relies primarily on the latest wave of data collection for
each country and country-level analysis, supplementing it with trend-related data as well as provider-level nuance.
• Measuring Risks in Agent Networks draws on both supply-side (ANA) and demand-side (Financial Inclusion
Insights, FII) data to propose indicators for different types of risks. Both datasets are analysed at the country-level
to offer country-wide benchmarks for providers to use.

Fitting The Pieces of The Liquidity Management Puzzle: Nancy Kiarie, Ian Odongo, and Vera Bersudskaya

Uploaded by

Copyright:

Available Formats

Fitting The Pieces of The Liquidity Management Puzzle: Nancy Kiarie, Ian Odongo, and Vera Bersudskaya

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Fitting The Pieces of The Liquidity Management Puzzle: Nancy Kiarie, Ian Odongo, and Vera Bersudskaya

Uploaded by

Copyright:

Available Formats

Fitting the Pieces of the

Liquidity Management Puzzle

1. Defining Liquidity Management

E-float refers to the balance of electronic money (e-value) present

E-float management is the process of ensuring that the amount of

2. Describing Liquidity Management in ANA Countries

Unpredictable fluctuation in client demand 7

Long wait at rebalancing point 6

Lack of resources to buy enough float 4

Need to shut store to rebalance 2

Figure 2. Rebalancing costs as a percentage of agents’ monthly revenues

Figure 5. Non-exclusivity rates across ANA countries

Pakistan ‘17 78%

Tanzania ‘15 70%

Senegal ‘15 66%

Uganda ‘15 64%

Bangladesh ‘16 56%

Kenya ‘14 13%

3. Impact of Failed Liquidity Management

3.1 Denial of Transactions and Reputational Risk

Figure 6. Percentage of daily transactions denied due to lack of float

As alluded to above, non-exclusive agents have a harder time managing

Access to provider data would enable further analysis of liquidity

Figure 7. Percentage of daily transactions denied due to lack of e-float, by exclusivity

3.2 Reduced Agent Profitability

3.3 Loss of Customer Trust

3.4 Agent Workarounds

3.5 Compromised Digital Innovation

3.6 Reduced Agent Investment

2. With the exception of Indonesia.

Figure 8. Average amount of e-float held by ANA agents (USD)

4. Why is Liquidity Management so Difficult?

This section describes factors that make it challenging for

4.1 Whose Responsibility is It?

Figure 9. Liquidity delivery to ANA agents

Uganda ‘15 27%

4.2 Incentivising Liquidity Management

4.3 Seasonality of Demand

This results in denial of transactions and agents incurring

Tracking of transaction trends in a month helps agents know

Armed with this information – providers or master agents

4.4 Agent Recruitment and On-boarding

4.5 Liquidity Sources and Rebalancing Costs

4.6 Non-interoperable Agent Models

5. Liquidity Management Models

5.1 Owning Float

5.2 Delivery as Opposed to Pickup

5.3 Instant as Opposed to Delayed Rebalancing

5.4 Self-rebalancing as Opposed to Credit

The exception is Uganda, where dedication has increased with the

Non-dedicated agents can rely on their existing business to facilitate

Some agents receive liquidity management support through self-sourced

This is unfortunate, since the agents most affected by liquidity shortages

6. Five Innovative Solutions to Liquidity Management

6.1 Innovative Agent Platforms

6.2. Uber-isation of Agents

In ‘Reimagining The Last Mile – Agent Networks in India’, MicroSave

6.3 Use of Data Analytics to Predict Demand

6.4 Credit to Allow Agents to Access Working Capital