[From CFI blog, 23 December 2015]
Here are my best wishes for the CFI community, to be sung to the tune of the last bit of the twelve days of Christmas:
On the twelfth day of financial inclusion my true donor sent to me:
Twelve human centered designers
Eleven ground truths validated
Ten agent trainings
Nine peer networks learning
Eight logs a-framing
Seven innovation challenges
Six rapid prototypes
Five fast failings
Four big data thingies
Three evidence bases
Two RCTs evaluating
And an e-money issuer license.
[From the Institute for Business in a Global Context at the Fletcher School, 3 December 2015, with Kim Wilson]
Most of us in the digital financial services space have had to learn about digital payments and digital financial services on the job. While there are business school courses in banking and supermarket distribution networks, they don’t prepare you for a career in building the basic 21st century infrastructure of Digital Money. Digital money is our term for any transaction that securely enables the sending, receiving, storing, or tracing of money on digital networks.
As a result, the increasing number of companies building out teams in the emerging Digital Financial Services space face severe capacity bottlenecks, especially in developing countries. Their employees may come to develop deep operational expertise in the specific aspect of the industry they are involved in, but often lack a broader perspective of industry trends and issues, and remain unaware of alternative models that exist elsewhere.
In order to address this, the Digital Frontiers Institute (DFI), which was recently co-founded by David Porteous, Gavin Krugel and Ignacio Mas, is partnering with the Fletcher School to create the first university-certified course in Digital Money. This is an intensive, 12-week online course, and it will be offered for the first time this February, in English (see the course brochure).
The course is a foundational one, exposing students to the full lay of the land around digital payment ecosystems. It´s more about the forest than the trees; we focus on the big picture and integrate the component parts (banks, utilities, governments, third party providers) into an understandable whole.
We have designed a student journey that uses online to technology to move beyond a classic classroom experience. Students will not experience the course as a list of steps they need to take – videos they need to watch or assessments they need to complete. Rather, the whole course will be structured as an online discussion, which faculty stimulate by injecting teaching materials, cases, and open questions. Students can earn points towards their certificate requirement in a variety of different ways, so they can concentrate on those that best suit their interests and work commitments.
The course takes advantage of the broad range of Fletcher faculty interests and experiences in the form of a weekly “food for thought case,” which relates digital money to a variety of real-world issues, such as digital marketing, business digitization, migration, and ultra-fragile identities.
The global experience with online learning points to the benefits of blending online pedagogical resources with facilitated face-to-face classroom discussions. We will invite local organizations anywhere to work with us to create a blended offering for local targeted communities of professionals and students who enroll in the Digital Money course. By joining our Communities of Practice program, partner organizations can, if they so choose, create their own additional programing to be used in conjunction with the DFI-Fletcher Certificate in Digital Money curriculum. They can use this to develop closer relationships with industry participants, and perhaps also further their own capability assembling objectives and programs, building on top of the basic course curriculum.For the DFI, this course represents a first step in its objective of creating an ongoing professional development network around digital payments. For Fletcher, it represents a deepening involvement in the inclusive digital finance space, along with its Leadership Program for Financial Inclusion (FLPFI), Cost of Cash studies, and other research programs.
David Porteous, Gavin Krugel and I have teemed up to build capability in digital financial services and payments. We have now officially launched the Digital Frontiers Institute (DFI), as a non-profit.
Ultimately, our aim is to create:
Our first offering will be a 12-week online course in Digital Money, co-certified by our partner university, the Fletcher School. It will be offered for the first time next February, in English. This brochure describes the course in detail.
We are now accepting applications from individual students and organizations that want to sponsor a set of students. We have some scholarship funding which students can apply for as well.
We are also seeking partners who can help us ‘blend’ the course, by hosting weekly physical discussion forums for local students in their offices, and possibly adding content of their own to the basic structure of the DFI Certificate course. These partners might be financial institutions and payments-related companies wanting to create an in-house training facility; consultants and industry vendors wanting to develop an ecosystem of like-minded customers and partners; business schools wanting to create their own digital money offering; NGOs; etc.
[From NextBillion blog, 9 September 2015, with Ross Buckley]
The essence of banking is taking calculated risks, and banks’ profitability comes from taking such risks. That is not to say that bankers are inherently risk loving; they often display a strong conservative bias, which is a natural form of self-protection against excessive risk-taking. Calculating risks appropriately requires getting as much information as possible on the underlying sources of risk. Bankers therefore seek to establish ongoing relationships with their customers as a path to capture further information.
On the other hand, the essence of payments is offering transactional services with the minimum amount of risk. Profitability comes from customer service and convenience, not taking risks on behalf of customers. So payment systems are designed to offer customers maximum functionality, speed and convenience, at adequate levels of security and certainty for all parties. Modern payment systems seek to minimize risk by conducting transactions on a funded basis and by operating as close as possible to real time. Technology, rather than relationships, lies at the heart of transaction speed and certainty.
In a new paper, we analyze the ways in which digital payments are emerging as a specific field of expertise, and how and why it differs from banking. The principal differences between the two fields are that banks prosper greatly from managing risk and little from network effects, whereas payments providers typically seek to avoid risk and prosper greatly from network effects. This leads to fundamentally different outlooks between the practitioners in each field.
For payments providers, being able to handle transactions on a funded basis and in real time is enormously liberating because it enables transactions with less well-known parties. It makes it possible to opt for mass-marketing channels, without having to worry as much about screening customers. It also makes it possible to engage indirect service channels; for instance, offering cash in/cash out through a network of thousands of retail outlets. This is not to say digital payments do not carry risks, but the aspiration is always to limit them.
In payments, the quality or depth of individual customer relationships matters less than their number and breadth. This is because payments can only be understood in the context of a network, and the size and breadth of the user base are defining characteristics of the network itself. Payment systems are subject to strong network effects (the more users on it, the more valuable the service is to any given user) and operate in a multitude of two-sided markets (there needs to be buyers and merchants, bill payers and billing companies, wage earners and employers).
That´s not necessarily the case with banking: There may be scale effects because serving more customers is cheaper than serving few, but one customer doesn´t directly benefit from there being a large number of other bank customers. Banking is fundamentally about the functioning of institutions (how they manage risks and build enduring customer relationships), whereas payments is more about the functioning of ecosystems (who is in it and how big it is).
As payments become more deeply researched and its practitioners more specifically educated in it, these differences in economic drivers of banking versus payments ought to create a differential regulatory treatment. Traditional banking regulation seeks to limit the risks banks assume, because when banks fail the money they lose belongs to ordinary people, who vote, and the broader economic consequences of bank failure can be severe. For both these reasons, politicians feel the need to bail out failing banks. Payments are traditionally regulated as part of banking regulation, and often by the same regulatory institutions, but the imperatives that drive banking regulation should not drive payments regulation. A failure of a payments provider should not necessitate a bailout with public funds. While it may prove highly inconvenient to many people, it is difficult to imagine the failure of a payments provider causing financial market contagion, as did the collapse of Lehmann Brothers, for instance.
Payments are their own industry and they deserve their own regulatory regime, finely attuned to the relatively minor risks that payments generate. Changes in banking in the past 20 years have been substantial, but the greatest changes and opportunities in the next 20 years are likely to arise in payments.
[From CGAP blog, 21 July 2015]
When it comes to understanding the needs and behaviors of low-income people, the financial inclusion literature is full of contradictions. Experts celebrate poor people for their complex, active financial lives, but then seek to educate them financially. Researchers document how resilient and purposeful their informal practices are, but then investigate ways to protect them against their own financial habits. Giving the poor a wide range of financial choices is an admirable goal, but do we really need to “nudge” them to change behaviors, as if the choice had already been made for them?
Education is often identified as a barrier preventing customers from using digital financial products. In reality, however, teaching someone to use money in a new way - digitally - by starting with education may be the toughest path. Change only comes with practice, and people will see little reason to change without a compelling reason. It may be easier to inspire the use of digital financial services if we flip the script around.
[From Helix Institute´s Digital Finance in the Field blog, with Mike McCaffrey, 14 Jul 2015]
Agents are critical to the customer experience of digital money services because they represent the first and most tangible service touch points for most end users. Agent networks are also probably the most operationally burdensome and costly element of the digital financial service value chain, typically costing anywhere between 40 and 80 percent of revenues generated from the business. Providers therefore need to approach agent network development and operation with a high degree of strategic clarity to drive a sufficiently tight operational focus.
The importance of agent networks is only rising. The MMU´s State of the Industry reports shows that since 2011 the amount of active agents providing digital finance services has grown by almost 800%, while the average number of agents per provider has increased by over 260%. Further, this has not just been happening in East Africa, but in regions around the world, like South Asia where bKash in Bangladesh and MobiCash in Pakistan have scaled much faster than any network in East Africa ever could have dreamed.
The successful design of an agent network must ensure that it is structured appropriately to deliver a specific value proposition to a chosen target market, while making business sense for the provider. In a new paper, (Designing Successful Distribution Strategies for Digital Money), we document the variety of ways in which digital finance service providers (including banks, telecoms and third parties) have gone about assembling and then managing networks of third-party retail agents. We start our analysis with some strategic decisions involved with choosing stylized agent management models. However, based on seven case studies included in the report, and additional ones published on the Helix Institute website, we found that providers often evolved and especially hybridized their agent network strategies over time. We identified several core reasons behind the increasing diversity of models employed by the more mature providers.
Some providers who initially relied heavily on flat, centralized channel structures feel they need to embrace more scalable models to grow faster and avoid overly burdensome operations. For instance, Airtel Uganda evolved from a centralized build model to a master agent model to better manage growth, and UCB in Bangladesh opportunistically partnered with a third party specialist (MobiCash) that was building its own agent network.
Once they have successfully built a strong proprietary agent network, some providers have tended to feel safe in bringing in partners to complement their own agent network. For instance, both UCB in Bangladesh and Equity Bank in Kenya have been opening up to partnerships with retail chains.
But the trend is not always towards more outsourcing and partnering. Some providers who initially relied heavily on retail chain partners to roll-out their agent network may feel they need to regain some control over geographical coverage. For instance, BBVA Bancomer in Mexico and Eko in India added a centralized channel build to areas in their network with low coverage, while M-Sente in Uganda implemented master agents, to better extend their coverage into specific rural areas.
A related situation is where providers who were initially happy to work with non-exclusive channel partners or share retail agents with other providers in order to grow faster feel they need more control over the customer experience and bring back some differentiating elements. For instance, BBVA Bancomer in Mexico needed a direct, exclusive channel that could focus on customer acquisition (rather than merely cash in/cash out) and recruited agents directly that could do this to complement their retailor partnerships. Islami Bank (IBBL) in Bangladesh found that it had to provider better liquidity management services for agents, and brought in master agents for support. In the case of Easypaisa in Pakistan, an increasing level of competition in the market meant that more control was needed over at least part of the network, therefore increasing the strength of the direct relationships they have with the agents.
A change in the agent network model may also be required when a new service is added that puts pressure on the existing agent network, either because the new service requires a higher touch sale and service model or because it is targeted at a demographic that is not adequately served by the existing agents. This has been observed with banks that agree to distribute Government-to-Person (G2P) benefits and suddenly need to build a denser network in rural areas.
Sometimes the changes in agent network structure and operations happen organically over time, as the agent channel itself differentiates and it becomes clear to providers that certain agents are better at registering customers, or tend to have more float and do substantially more transactions. It becomes evident that not all agents are equal, and it does not make sense to treat them as such. In this case, providers usually implement systems to start segmenting their agent network and offering different levels of support based on performance or other salient criteria. These trends all seem to be natural progressions that channels make as they become larger and more sophisticated overtime, and are certainly a sign of a continually maturing industry.
[From Brookings Institution´s TechTank blog, 3 June 2015]
One of the foundational notions of digital financial services has been the distinction between payment rails and services running on the rails. This is a logical distinction to make, one easily understood by engineers who tend to think in terms of hierarchies (or stacks) of functionalities, capabilities, and protocols that need to be brought together. But this distinction makes less sense when it is taken to represent a logical temporal sequencing of those layers.
It is not too much of a caricature to portray the argument —and, alas, much common practice— like this: I’ll first build a state-of-the art digital payments platform, and then I’ll secure a great agent network to acquire customers and offer them cash services. Once I have mastered all that, then I’ll focus on bringing new services to delight more of my customers. The result is that research on customer preferences gets postponed, and product design projects are outsourced to external consultants who run innovation projects in a way that is disconnected from the rest of the business.
This mindset is understandable given limited organizational, financial and human resource capabilities. But the problem with such narrow sequencing is that all these elements reinforce each other. Without adequate services (a.k.a. customer proposition), the rails will not bed down (a.k.a. no business case for the provider or the agents). In businesses such as digital payments that exhibit strong network effects, it’s a race to reach a critical mass of users. You need to drive the entire stack to get there, as quickly as possible. Unless, you develop a killer app early on, as M-PESA seems to have done with the send money homeuse case in the Kenyan environment.
It is tough for any organization to advance on all these fronts simultaneously. Only superhero organizations can get this complex job done. I have argued in a previous post that the piece that needs to be parceled off is not the service creation but rather cash management: that can be handled by independently licensed organizations working at arms length from the digital rails-and-products providers.
What are payment rails?
Payment rails are a collection of capabilities that allow value to be passed around digitally. This could include sending money home, paying for a good or a bill, pushing money into my or someone else’s savings account, funding a withdrawal at an agent, or repaying a loan. The first set of capabilities relates to identity: being able to establish you are the rightful owner of the funds in your account, and to designate the intended recipient in a money transfer. The second set of capabilities relates to the accounting or ledger system: keeping track of balances held and owed, and authorizing transactions when there are sufficient funds per the account rules. The third set of capabilities relates to messaging: collecting the necessary transaction details from the payment initiator, conveying that information securely to the authorizing entity, and providing confirmations.
Only the third piece has been transformed by the rise of mobile phones: we now have an increasingly inclusive and ubiquitous real-time messaging fabric. Impressive as that is, this messaging capability is still linked to legacy approaches on identity and accounting. Which is why mobile money is still more an evolution than a revolution in the quest for financial inclusion.
The keepers of the accounts —traditionally, the banks— are, of course, the guardians of the system’s choke points. There is now recognition in financial inclusion circles that to expand access to finance it is not enough to proliferate the world with mobile phones and agents: you need to increase the number and type of account keepers, under the guise of mobile money operators, e-money issuers or payment banks. But that doesn’t change the fundamental dynamics, which is that there still are choke point guardians who need to be convinced that there is a business case in order to invest in marketing to poor people, that there are opportunities to innovate to meet their needs, and that perhaps all players can be better off if only they interoperated. A true transformation would be to open up these ledgers, so anyone can check the validity of any transaction and write them into the ledger.
That’s what crypto-currencies are after: decentralizing the accounting and transaction authorization piece, much in the same way as mobile phones have decentralized the transaction origination piece. Banks seek to protect the integrity of their accounting and authorizations systems —and hence their role as arbiters of financial transactions— by hiding them behind huge IT walls; crypto-currencies such as Bitcoin and Ripple do the opposite: they use sophisticated protocols to create a shared consensus for all to see and use.
The other set of capabilities in the digital rails, identity, is also still in the dark ages. Let me convince you of that through a personal experience. My wallet was stolen recently, and it contained my credit card. I can understand the bank wanting to know my name, but why is the bank announcing my name to the thief by printing it on the credit card, thereby making it easier for him to impersonate me? The reason is, of course, that the bank wants merchants to be able to cross check the name on the card with a piece of customer ID. But as you can imagine, my national ID got stolen along with my credit card, and because of that the thief knows not only my name but also my address. That was an issue because I also kept a key to my house in the wallet. None of this makes sense: why are these “trusted” institutions subverting my sense of personal security, not to mention privacy?
The problem is that the current financial regulatory framework is premised on a direct binding of every transaction to my full legal identity. As David Porteous and I argue in a recent paper, what we need is a more nuanced digital identity system that allows me to present different personas to different identity-requesting entities and choose precisely which attributes of myself get revealed in each case, while still allowing the authorities to trace the identity unequivocally back to me in case I break the law.The much-celebrated success of mobile money has so far really only transformed one third (messaging) of one half (payment rails) of the financial inclusion agenda. We ain’t seen nothin’ yet.
[From Center for Financial Inclusion blog, with David Porteous, 29 April 2015]
In a recent new paper, we look at identity from two opposite but complementary perspectives. The first is a narrow biological perspective, under which identity is associated with one´s uniqueness as an autonomous living organism with a unique genetic makeup. The legal basis for identity tends to be based on this perspective, and leads to questions that focus fundamentally on the confidence with which identities can be asserted and confirmed.
Beyond the definitional question of what it is about one´s person that creates his or her individuality, there is the empirical question of how it can be verified by someone else, such as a financial service provider, through observation. Generally your identity is established indirectly, by demonstrating your command over some proxies (e.g. a signature, a card, a PIN) that have been linked to your identity. The core decision for providers is therefore to determine when they consider that they know someone with good enough probability.
The second perspective is information-based, and views individuals as an irreducibly complex web of personal information and attributes. Digital markets tend to take this view of identity, with customers characterized more in terms of defined attributes, preferences, and transaction histories that can drive customer segmentation than on intrinsic uniqueness. This perspective leads to questions that focus fundamentally on what information about themselves it is legitimate to expect people to reveal to build up their identity, and what information they have the right to keep private.
Why do we so resist websites´ attempts to extract personal information from us? Why do we distrust organizations that appear to squirrel information away and use it to build profiles of us? Oxford University philosopher Luciano Floridi (see chapter 5 in this book) argues that people want to be in control of their personal information because that makes them feel more in control over their own identity. If I were completely transparent and held no secrets, anyone would be as enabled as I am myself to define who I am. By withholding personal information, we feel that we have some control over how we project ourselves. Our management of personal information is central to how we shape our identity in various spheres.
We manage our personal information, and through that our identity, chiefly by compartmentalizing it based on the different roles or personas that we assume in different circumstances. You don´t present yourself in the same way to your employer (you are employee number X and report to Y), your family (you are a stern-but-kind parent), your friends (you want to be seen as fun and easy-going), and indeed at the passport office (you are a neutral, non-suspicious face). Again, you are able to have different personas to the extent that you are able to control which information is disclosed and acted upon in each circumstance, based on what´s most relevant.
These facets of your being can be construed as distinct identities, linked to the same person. Each of these identities is attached to a different —and maybe even contrasting— set of personal attributes. Who we really are is no more and no less than the combination of these distinct identities, but none is necessarily more real than the others. We want to be able to shift easily among them as we go through our daily life.
The two perspectives on identity are profoundly different. You are an unvarying sequence of genes, but also an evolving social being. You are an indivisible entity, and you are a loose accretion of diverse personal traits and roles.
And yet these two perspectives need to be reconciled because they bring together the two key trust aspects or gaps that are at the heart of most identity problems: security (i.e. the confidence with which identity can be established) and privacy (i.e. the sensitivity and sense of personal control with which the information associated with one´s identity is revealed and distributed). These two aspects are often seen as being at odds: to be sure of who you are, I need to know more about you. But when security and privacy are not handled appropriately, trust gaps appear between social entities, between customers and providers, and between citizens and the State. You need to see my full name, exact date of birth, and ID number to let me into a bar or to pick up a parcel at the post office? Really?
The most promising approach to reconciling these diverse notions of identity is building digital systems that (i) permit an unbundling of personal information, and (ii) put users in control of how these unbundled bits of personal information are linked and exposed. Here is how it might work (see this book by David Birch for a fuller explanation):
• My uniqueness can be represented abstractly by a unique number assigned to me by a trusted authority (likely a government entity). Imagine that this number is linked to my biometrics, so that only I can claim to be the person represented by the number.
• I can then link this unique number to different personas (represented by pseudonyms) that I want to assume in different circumstances, for different reasons – say my electric utility, library card, or my Amazon log-in. These entities don´t need to know specifically who I am and what my unique number is, just that I have one so that they can consider me a real person and that a trusted party has this information in case I break the law.
• I can then link specific personal attributes that are relevant to each of these different pseudonyms. For instance, I’d reveal my physical address to the electric utility so that they can service my house and the fact that I am over 18 at a bar, but I may not want the bar to know my address or Amazon to know my age. My personal attributes would be digitally confirmed by a host of different entities that are in a position to verify it.
The first step with dealing with digital identity is, therefore, breaking out from unitary notions of one trusted party knowing everything about me, or one ID card serving all purposes. Users can be in control of their identity, supported by trusted third parties who help them assert digitally whatever personal information they wish to establish. Financial institutions would seem to be well placed to become such trusted third parties, since they acquire substantial amounts of personal information through mandatory Know Your Customer (KYC) requirements, loan and other product applications, and regular customer usage. They could put this customer knowledge at the service of each of their customers, by validating specific attributes that customers wish to have confirmed to others.
[From Fletcher School´s CEME InclusiveFinance blog, 7 April 2015]
We have been handed down a fresh round of evidence on the impact of microcredit from six randomized control trials (RCTs) undertaken by eminent economists (see summary paper here). The results are what most reasonable people would expect: that microcredit is useful for many but by no means transformational. Perhaps the more novel finding is that the impact on average is zero, not negative.
So does this finally settle the debate then, some forty years after Muhammad Yunus’s first (non-random) experimentation with microcredit? Alas, I don’t think so, though it will certainly affect how arguments are pitched henceforth. There are four reasons why evidence from RCTs alone will not close the debate.
First, RCTs are pure empiricism, devoid of any theory. Give a bit of something to some, withhold it from others, and compare. So when the results come out, all one can do —indeed, what the dozens of online commenters on these studies have done— is to interpret them by retrofitting them into one’s own prior theories and mental models. How did the impact happen? Why the differences between men and women, young and old, richer and poorer? Would this work elsewhere in the same way, or if the interest rate were lower? But all we have learned is that six concrete programs are not impactful, on average. The rest is inference and theorizing. We are back to extrapolating on the basis of isolated (now n+6) data points.
Second, RCTs measure impacts on a number of variables: consumption, investment, schooling, female empowerment, job creation, etc. Results are often a mixed bag, so we back up into the problem of how to total up these impact categories to get a net-net impact. There’s now the inevitable suggestion that we should be measuring how people’s happiness is affected, as a sort of grand bottom line. How can we hope to measure ultimate impact if we lack even a basic definition of what we mean by impact?
Third, there’s the question of what we are supposed to make of averaged impacts. We know that debt is not right for everyone, so why would or should debt be held to an average standard of goodness? Surely debt is good for some people, in some circumstances, at certain times. What’s the point of diluting —and possibly cancelling— the benefits to these people statistically? The issue is one of targeting interventions, not passing across-the-board judgments on them.
Finally, for all the sophisticated statistical tools deployed, it all hinges on the quality of the data collected. Given that RCTs are fundamentally an empirical exercise, it is odd that in the dozens of online commentaries on the six studies there hasn’t been much discussion of the empirics itself. The research papers are often not written for people like me to understand (have you tried reading them?), so I am relegated to being a user of the abstract, introduction, and conclusions sections. But what I can read —what we can all process— is the survey questionnaire. I would urge all those who feel they need to have an opinion on these studies to start by reading the questionnaires. Take this one, for example. In your mind, is this fit for the purpose of evaluating the impact of microcredit? Does this meet the lean research standards proposed by Kim Wilson? Try to have the questions answered tonight by your partner or spouse. Then try to imagine what kind of data you’d get if you were going around asking these questions of perfect strangers, showing up at their doorstep unannounced.
The suggestion that such empirical studies can settle the impact question of microcredit is just as naïve as the prior suggestion that the best way to get poor people out of poverty is to get them into recurrent cycles of debt. I agree entirely with the findings from these RCTs. Was anyone really hopeful that the six RCTs would turn out otherwise?