>> Shabnam Erfani: Hi, I’m Shabnam Erfani. I’m Director of Technical and Business Operations at Microsoft Research. I’m excited to host Jake Kendall today from Gates Foundation. Jake is the Deputy Director of the Financial Services for the Poor Team. He runs the R and E. He’s here to tell us about the challenges that they’re facing in that space. This is a particularly interesting space for Microsoft. In terms of looking at the next four billion people who in the future will be on boarded to not only our financial systems, but also software industry devices and so on and so forth. Without further ado I’m going to hand it off to Jake and let him tell you about what he’s working on. [applause] >> Jake Kendall: Thanks a lot, thanks. Thanks everybody for having me out and coming out to hear what we’re working on. I think for me the goal of coming out here is really you know we’re and I’ll explain more about our teams work and our strategy. But we’re looking to; you know our core vision for our strategy is to create the next generation of financial services models that are relevant for the two billion or so people who currently don’t have any access to financial services. A lot of that is not a technology problem, right. A lot of that is a regulatory issue or a business model issue, or something like that. But a lot of it is a technology problem. You know we come out to see groups like Microsoft Research who have some of the top thinkers in technology. Hope to both get feedback on how we’re looking at the problems but also of course potential solutions. Really excited to be here and talk to you guys. I’m Jake Kendall, thanks for the introduction. I run the Research and Emerging Tech Team on the FSP, as we all ourselves, Financial Services for the Poor Group within the Foundation. Just a little bit of background on myself, I started in Physics at MIT as an undergrad. Did a couple of years in rural Zambia as a Peace Corps Volunteer, so that’s where my real development training started, I guess. Spent some time then trying to apply what I learned at MIT doing Cryptography with a couple of startups. But this was right around two thousand and one and was not a very good time to being doing startups. I ended up going back to get a PhD in Economics kind of bouncing back to the development role again. Spent a short time at the World Bank and then was picked up by the Gates folks and started there as a PO. I’ve been there six years and now I’m directing one of the four initiatives on our team. Just to give you some background on FSP. As I said our kind of core mission is that a group of two billion or so people who are currently outside the financial system. Then there’s probably another billion who are very poorly served by whatever their current financial products are to one or two more billion. There’s a large chunk of the world that just don’t have any financial services. We’re really working on how we solve that problem? Our team within the foundation is what I would say medium sized. We’re within the Global Development Group and we’re about a hundred and fifty million a year in grants and contracts. We also do things like equity investments, debt investments, all those kinds of other tools, all of with socially motivated purpose or a charitable purpose. Our teams about twenty-eight people now I think changing every other week. I think, so, but bounces around thirty. As I said we’re broken into four different groups, three of them sort of orient around geographies and the themes common in those geographies. Then my group is sort of a cross-cutting group. The first group is really focused on, I think the themes of the groups kind of show the breadth of the tools and the sort of many different angles through which we approach the problem. The first group is focused on government policy, regulation, and sort of donor policy. We run pretty significant advocacy campaigns within a lot of the countries where we work to advocate around particular regulatory changes, some of them fairly technical. But a lot of them having to do with regulations. That either blocks competition at the low end of the market by putting it all in the hands of banks and certain types of financial players who are just not very motivated to serve people at the low end of the income spectrum. Or regulations that just make it hard to serve people at low-low cost which is of course imperative for people who are very poor. Those are things like you know to what degree do you need to see identity, proof of identity, right? Do you need to see a proof of address you know? Better part of half the world doesn’t have proof of address, right. Those kinds of regulations rule people out unnecessarily, right, because the people at that level of income are not really adding risk to the system in any significant way. If you can find creative regulatory structures that actually allow them to get accounts and to get on to the financial system without adding risk of you know terrorism, finances, and the other things that the regulators are very risk adverse about. We also in that same initiative work a lot with government payments. Try to get the government to think of itself as a sustainer of the system and a stakeholder in the digital payments and financial system. If they’re doing lots of transfers to citizens for welfare programs, famine relief, you know maybe they’re supporting students in school, or whatever it is. Those payments are often done in a paper form or through a voucher, or whatever it is. You know massively corrupt usually and very inefficient. If you transition those things to digital forms, direct deposit into mobile money accounts or banks accounts. We’ll get into a little bit how some of that works. You know big efficiencies for the government but also sustaining for the system itself because that pumps volume and sustains the business model of serving those people at the lower end of the spectrum. Second initiative is focused on infrastructure. Sort of the core element of it is what we call the Level One Project. Which I’ll go into which is our design and reference model for a payment system that serves the poor. They’re focused in primarily Pakistan and Nigeria right now where you know we have good buy in from the government to try and build out these systems at the national level. We then have a third initiative focused more on East Africa and Bangladesh where the mobile money systems. I’ll go into it a little bit for those of you who don’t much about mobile money what those are. But mobile money systems in those markets are more advanced. You have adoption rates of maybe seventy, seventy-five percent for example of the adult population in Kenya of mobile money products. You you’re clearly getting down to pretty low levels of income within that group. You’re also getting pretty much ubiquitous adoption of a digital payment in financial services product which you know for us is quite a big win. But you know there’s a lot more to do in some of those systems. We have a team that works on both sort of ecosystem development and on sort of next generation business models, and that sort of thing. Then my group, which works on both client side research to understand the impacts and the needs of low income consumers and households, and that sort of thing, so almost some of it’s in the academic realm. Some of it’s more kind of data collection and surveys to collect the market intelligence. Then the upstream or emerging technologies, so you know looking at the emerging technologies globally in the financial sector but also in other sectors. How can we you know opportunistically capture the value that they can create in our sector and help those technologies transition into the market? That’s the broad overview of our team. I’ll go into a little bit kind of our broader theory of change. I’m getting to the more technology problems focus part. But I kind of want to give you all a broader sense of what our team does and why. Some of this is because I went out to your MSR Bangalore and dove right into the technologies stuff. They were sort of like well we don’t actually know what mobile money is or a lot of them didn’t and stuff. I wanted to give a bit more background here. If you start, this is our teams broader Theory of Change. You start out there on the right side of that column labeled Impact. You know that’s the goal that we’re going for, right. What’s the impact that we want to have in the lives of low income population and poor households? We start with the observation you know that a lot of the reason why people are poor is not that they’re sort of born poor and stuck in poverty forever. But often they transition out of poverty. In large numbers we actually looked at the data globally. You see you know anywhere from fifteen to thirty percent of people who are at below two dollars a day, which is kind of the nominal figure we use. Transition out of that over the course of a couple years but the problem is of course that a similar number transition back in. That’s because low income people live a very, a life of high uncertainty. They face a lot of shocks in particular health related shocks which can you know cause loss of income, loss of job. Of course the expenditure around trying to cure an illness or broken leg, or whatever it is. People end up you know in responding to these shocks, and shocks can also come in the form of you know a bad agricultural season for people who have rain fed agriculture and that kind of thing. Of course in responding to these shocks people are often pushed much further into poverty. Then you do see of course people moving out of poverty by getting a formal job or improving their entrepreneurial capacity. But that often also and this is the part, this is why we focus on sort of the cycle of poverty. Both of those margins are affected by the ability to generate resources and to use financial services, right. If you face a shock the ability to borrow, the ability to have insurance, or to reach out to family and friends, or get a transfer from the government. All of those are kind of you know kind of core coping mechanisms. Then the ability to capture an opportunity to get out of poverty often also requires some form of finance. That’s kind of the core motivation. If you go one step in to the left we think that in order to effectively bring Financial Services to the large number of people who currently don’t have them. You have to move people onto digital platforms. There’s been a longer history of course of microcredit which many of you’ve probably heard of. Other you know banking, low income banking and things like that. That have approached it from a much more analog process based approach. They’ve had some success; microcredit you know had a certain amount of success, reached a couple hundred million people. But stalled out and we think it’s largely because that model just isn’t sort of scalable and sustainable and has fundamental limits in terms of how low cost it can get and things like that. We believe it has to all be done over digital platforms in order for the cost curves and for the business model to work out in such a way that you can reach you know sort of the last mile of customer. Then these five larger inputs to that are kind of the five areas and since I covered what the teams doing. That kind of covers, but those are kind of the areas in where we think if we push on some of these levers we can start to see some of this happen. I wanted to give you next a bit of a background in mobile money which is what I would say is sort of the current state of play, right. This is mobile money is financial services primarily person to person payments who, that are delivered over mobile systems. These are primarily low end phones which is the type of phone that people in developing companies of course tend to have at low levels of income anyway. Currently they’re kind of the one success story to date. They’re, I’ll give you a sense of how broadly they are. We think there’s fundamental issues with them but they have demonstrated a fair amount of success. I wanted to give you kind of a background in what they look like. Then we can think about what’s the next step to kind of move beyond them. A typical mobile money system starts with the MNO which is the Mobile Network Operator up in the left hand corner. They’ll partner with a Partner Bank they will in order to create the digital value that’s going to sit on their mobile money system. They’ll usually put a deposit into the bank that will be backed essentially so the value on the system is backed one to one with deposits in the bank. The bank, largely that’s the role that the bank plays and they hold that trust account and make sure that the MNO has backing of one to one. Then as part of that as well they’ll often open up to agents who can then also go in and buy digital value themselves on the account of the mobile operator. Agents are the small retail shops where the person who’s going to come in and get their mobile phone signed up and then do transactions. That’s where they go in order to get cash out, put cash into the system. The mobile operator will sign up a network of agents. Agents can, I’ll give you some photos of what agents can look like. They’re kind of any kind of retail shop in a developing company. You see them in the smallest kind of informal market stall all the way to supermarkets and things like that. An agent just signs up. They have a special type of phone. It’s not necessarily a special device but their account is special. It allows them to do transactions with clients who they also sign up. They’ll sign up subscribers. The subscriber will get a phone; get the phone registered with the mobile operator, at the agent. Then they’ll, you know the first thing that they want to do is load value onto the phone, right. They give cash to the agent. The agent will accept the cash and basically transfer them value, electronic value onto their phone in real time. It’s a one to one transaction. There’s no credit for example. They don’t give the cash and then it appears later on their phone. It’s real time. It’s one of the nice things about the mobile money system is that there’s no, I don’t need to trust the agent basically I know that he gave me the value. I can walk away. The agent will eventually accumulate enough cash. He’ll go and put it into the bank and that’s how he buys more value. It’s almost like they’re e-value merchants, right. They’re just selling value. Sometimes they’re buying value back, right. If I actually want to pull money out of my phone because somebody sent me some money, I go in, I give cash, and I, or the opposite I get cash, and I transfer him value. That’s really the basics of how it works. The core subscribers can then send money to other subscribers who may live a long ways away. They’ll go to a different agent and pull money out there. It’s all sort of mediated on the backend. That’s roughly how mobile money works. I’m actually going to go to this slide first and then we’ll get back to the previous one. This is just to give you a little bit of color in terms of what do these things look like in different places in the world. This one up on top this is a rural area in Bangladesh where the B cash service is active. You can see you know that’s just a little tiny stall on the side of the road, right. But you know they’re setup. They have the signage up. They have pricing up and you know that guy has a phone that he can give cash out or put cash in. Or he can also help you know for people who don’t quite understand the service and that kind of thing usually. Similarly, Kenya I think that’s one of the slum areas, so peri-urban. You know pretty basic structure. Over here is actually also in Kenya is you know a pretty modern looking supermarket chain where you can go in and pay with a credit card or do whatever you want. You can also pay with mobile money, right. You can buy your groceries. You can put cash in or take cash out. You know all the functions that you can do at one of these more rural things. You can also do in these more advanced things. It’s kind of an elegant system in that sense in that it you know it’s very easy with a phone you can pretty much install an agent anywhere. If you think about comparing this to a bank branch, right. You know the cap X for building one of these is many orders of magnitude less than a bank branch. In fact the way that these things are franchised the owner of the agent is the one who builds the structure. For the MNO the cost, setup cost of rolling out an agent network of you know thousands and thousands of agents is words of magnitude lower than the bank branch network, right. >>: What kind of fee can [indiscernible] agent expect to earn on a [indiscernible]? >> Jake Kendall: Yeah, great question. I should know this but they changed not so long ago. I think that they get about fifteen cents, somewhere between ten and fifteen cents per transaction. >>: Per transaction. >> Jake Kendall: Yeah, might be closer to twenty at this point. But it’s somewhere in that range so they, it depends a little bit on the volume. If, you know they’ll, that’s for a typical transaction. Then it scales up and down based on how much they’re doing. They get for both cash in and cash out although the client basically pays only for cash out. You know this is a decent way for, if they can do ten, fifteen transactions a day they can add fifteen, sorry, a buck fifty, sorry, about three dollars something like that. A good agent, we sort of estimate that for an agent to be truly viable they need to be doing about fifty a day. Otherwise it’s not that interesting as a proposition. You know when they get up into the maybe ten bucks a day range that’s when it starts to be more viable. Some of them are actually; some of them are doing actually pretty well. They make a thousand or two thousand bucks a month. They can actually crank out quite a bit. Some of it depends of course on the volume. >>: How do the sellers manage their payouts from a cash flow? I mean it’s not always a balance of payments that anyone installs. >> Jake Kendall: Yeah, it’s a great question too. That’s actually one of the pain points I was going to go into. Man, these agent networks tend to be about fifty to seventy percent of the kind of cost structure of a mobile operation, mobile money operation. A lot of the rest of it is marketing. The backend systems actually, of course once you’ve built them they don’t cost very much to run. There’s not a lot else to it. These things while they are you know course dramatically cheaper than banks and ATMs, and things like that. They still do you know both between the commission structure and there’s a decent amount of kind of managing them, training them, recruiting them. All that kind of stuff that’s kind of ongoing. Sorry, repeat your question. >>: To me it was the payout. >> Jake Kendall: Oh, how do they manage, so, yeah and that’s a big aspect of it. For the actual you know the guy who runs this little place himself he’s got to figure that out. In most cases they kind of put that onto the local entrepreneur who’s running these things. You know generally that’s smart, right. Make the person who’s best able to solve the problem solve the problem. They’ll find ways to either pay friends and relatives to go to the bank and come back. Or actually go themselves at really slow periods during the day. >>: Like I have enough load to kind of cover in between. >> Jake Kendall: Right. >>: But they might do in a day to get back down to the bank to replenish. >> Jake Kendall: That’s right. >>: They have to kind of get to know their business there. Because I image some could be more on the payout side, some could be balance, some could be more receiving. >> Jake Kendall: That’s right, yeah. It depends a lot. Rural areas tend to be all payout. >>: Yeah. >> Jake Kendall: People receive transfers from relatives in the city as sort of the standard thing. >>: Yeah. >> Jake Kendall: Then they come in and so it’s all payout. Then just they’re running back and forth to the bank to get more cash. Cities can be more balance, certain parts of cities. >>: Yeah. >>: You had said that the starts one day would help with that mobile operator. Is that always the case or do other entities also start… >> Jake Kendall: Good point, yeah, so that’s a typical. But it’s not always the case. Actually B Cash here in Bangladesh is not a Mobile Network Operator. They have a very close partnership with Mobile Network Operators. But they are actually one of the few third party systems. In some countries, Pakistan for example, regulations stipulate that there has to be a bank who’s actually holding the funds. Whereas in the model that I showed you it was more a bank that was holding a trust account on behalf of MNO. In those markets there what’s happened is essentially the Telco’s have bought banks and essentially have them as sort of sub-lines of business. Technically it’s a bank but on some level it’s also a mobile operator. >>: In any one country is it just winner takes all? Like there’s one dominant system or are there multiple systems? Are there brokers or exchange between them? >> Jake Kendall: Yeah, let me, I’ll get to that in a second actually. We’re going to give you; I’ll sort of give you the sort of global landscape of what it looks like in terms of the market structure. Just real quickly I’ll run this then. This is probably fairly obvious at this point. You know, here’s some of the common products or common things you can do with the mobile wallet. You can do cash in and cash out of course. Airtime top-up so I can buy more airtime for my phone. Course most as you guys probably know most of the phones in developing countries are prepaid plans. I buy minutes and then I use them. P two P or person to person money transfer that’s another, that’s sort of the main use case, bill payments growing, so you pay your utility bills, your school fees, your cable bill, whatever it is. Merchant payments in store, everyone’s always kind of believed that there’s this big promise for in store merchant payments. But it’s been really hard and has not taken off so that’s a very small sliver of current transaction volume. Bulk payments still one-to-many. Now more and more we’re starting to see stuff here like down in the bottom three where banks are leveraging these platforms. Essentially, so they don’t have build these branches everywhere and so they can reach people through their phone, through these agent networks. But allow them to use their more advanced banking products through that platform. Digital credit, where they’re using kind of the transaction data flow around the product to do automated lending and then there’s a few groups that have experimented with Freemium life insurance. Where essentially they use it as a promotion and then try and get people to buy up. But, you had a question. >>: Is person to person transfer kind of the main incent mechanism that gets people who are previously doing cash you know on to this? >> Jake Kendall: Yeah, so I’ll jump around a little bit just to answer the questions in real time. This is the global product mix. It’s data collected by the GSMA. That’s the kind of Global Mobile Operator or Industry Association. We actually support a group within the GSMA who tracks mobile money and tries to act as kind of a thought leader or think tank for the field around using mobile money for serving low income populations. You can see here on the volume side most of the transactions are airtime top up. Of course mobile operators love that, right. Cuts out the agent commission that they pay on all the guys who are selling the minutes because they don’t have to pay anybody a commission on that, and is convenient for clients because you can buy airtime at any time from your phone. You know good, decent volume of it is that gray is P to P. Then there’s a bunch of smaller stuff that’s pretty low, some bill pay. Then the value of course P to P transfers are much larger. I mean no one buys fifty; rarely do you buy fifty or hundred bucks’ worth of airtime. But of course when you do transfers to your family and a lot of it is. You know a lot of it traditionally has been I’ve been working somewhere and I want to send money back to my family. But actually now you see all kinds of things, you see a lot of small business owners, informal business owners who are you know using it to buy things from distant suppliers. Or manage their money in some way, even pay their employees. A lot of people support their family in other context like kids going to school and all that kind of stuff. More and more you see kind of people doing all kinds of other use cases for it. But one of the driving use cases was essentially the sort of family income transfer thing. Yeah, it’s almost always long distance. But more and more you also see with, especially with the small business merchants and stuff they’re starting to do it face to face. >>: If you go back ten, fifteen years ago, I mean what would people be doing instead of this to do those [indiscernible]? >> Jake Kendall: Yeah, so in a lot of cases they were actually sending physical cash. You know there were studies in Kenya which is probably the most advanced market now, as I mentioned. Pre MPESA and it you know people would put a, literally they’d go find a bus driver who’s going in the direction that they wanted to send money and say hey here’s an envelope please give it to so and so. Then they have to call that other person and say meet the bus driver because you know if you don’t then he’s probably going to take it. Or you know they would happen to have a friend who’s going in that direction. Or they would just wait until the person came home Christmas or whatever it is. There was a lot informal mechanisms that were very inefficient, fairly risky, hard to depend on, that kind of thing. We’ve done, you know we’ve done sort of pretty rigorous academic studies looking at when mobile money rolled out. Comparing people who just by the chance of where they lived and where mobile money rolled out first had access to mobile money versus those who didn’t. Suffered some sort of medical shock or some sort of other significant shock to their income, or whatever where they needed some money to deal with it. You can see a big difference in the coping patterns. In fact, the welfare outcomes of the people who actually had the mobile money because they were able to get money from further away from a broader group of people. Larger amount and faster than the people who had to rely on sort of the regular cash transfers and things like that, you know it really does affect even just that. We actually see I’m going to get to you know there are issues with mobile money. It’s not open and you know as you saw in that sort of diagram. If you think about the standard credit four corner model where you know banks can join and merchants can join, and that kind of thing. It’s not an open system like that. It doesn’t allow kind of the flexibility to create new products and new offerings. It’s pretty limited. But never the less, you know it does improve people’s MODs. We think of the digital stuff as actually you know has that value of being faster and all that kind of stuff. But it’s also really just a platform that you can build so many other things on that’s the real value of it. >>: How ubiquitous is mobile money in these markets? For example is it predominantly with the lower incomes? Or is it predominantly with the higher incomes or pre… >> Jake Kendall: Yeah, well so in terms of where it falls in the income spectrum. It tends to; I mean our data shows that it tends to start with sort of upper middle income. Often people you know who are traveling a lot. Or just you know kind of new, more kind of willing to try out something new, probably have less to lose on some level. It kind of starts in that middle income band and may spread in both directions. But eventually does go kind of down market. We’ve sort of tracked it in Kenya since the early days. Definitely seen like you know it wasn’t very high uptake in the sort of lower quintiles of income until more recently. But as it saturated people did start to pick it up. Some of that’s driven just by mobile phone adoption. In Kenya it’s kind of getting close to bumping up against overall mobile phone adoption in general. >>: Is cost structure modulate as more adoption goes in? For example are there sources for the lower income that can entice more to join in or is it pretty standard throughout? >> Jake Kendall: Yeah, they, so a lot of it is already geared to be a pretty universal need. The ability to move money over distance I mean you know it’s pretty fundamental, right. The low income populations you know why they didn’t adopt right away. I think a lot of it was just not being in kind of the bigger cities and not having probably been exposed to it very much. But also I think you know there’s a pretty strong network effect. People that had a really clear need like they’re a small business owner who’s going back and forth between two towns a lot. Just wants to move money like you know there’s clear willingness to pay and that kind of thing. Whereas I think over time you know the network effects probably started to drive it more with the low income population. But there hasn’t been a lot of really interesting low income population only oriented products. I think, yeah? >>: [indiscernible] confused like there was also international reticents. >> Jake Kendall: Yeah. >>: How does that work? >> Jake Kendall: Well so in, well international reticents get a lot of press I would say in the, especially in the development world, international development world. They’re actually, they’re not really used by, they’re actually mostly for people who are better off. I mean you really can’t afford to send someone to another country, certainly not Europe or the United States unless you’re actually better off. They’re actually pretty small percentage of the population in Africa and South Asia where we work. Certain countries accepted, right, so certain parts of certain countries in particular. There’s you know a lot of cross border traffic. But the basic way that they work is you know from MPESA for example they will open up a partnership with either certain banks, international banks that offer international [indiscernible] services. Or I think with MoneyGram for example they offer to, they have like a partnership. They do some sort of backend integration. Then someone sends you a remittance and it gets put into your mobile money account basically. It’s pretty similar to how, I mean from a client perspective it works pretty similar to how you just do it normally. Except the sender would go to you know a money transfer outlet in a developed country or wherever they are, yeah. >>: I imagine in most of these countries the majority of population is not added the traditional banking systems. >> Jake Kendall: That’s right. >>: What is the adoption rate for mobile money? Then the percentage that are not on board to mobile money what do those guys use? Is it still a cash economy? >> Jake Kendall: Yeah and actually even for the folks who are adopting mobile money they’re usually adopting it around a couple of specific use cases. They want to move money you know twice a month from here to there from their, you know husband to wife. Or whosever transferring, whoever you know someone’s working in a particular area. The overall adoption rates vary a lot by country. A lot of that has to do with regulatory, how recently regulations have opened up. Most countries in Africa and South Asia have some form of enabling regulation. Some of them are still kind of stuck and the regulation is basically blocking it. But you know in East Africa where it’s most developed the adoption rates are anywhere from about seventy, seventy-five with Kenya down to I think thirty, mid thirties in Uganda. Then in other countries it gets lower than that. You know in India for example it’s maybe one, two percent, so it’s you know almost nothing basically. It really depends on the level of market adoption. It’s really a network effect sort of a product. You know if I’m the only one who has it it’s totally useless, right. It’s like a phone. I mean the more people that have the more useful they are. You do sort of see often kind of these countries will sit in the doldrums for awhile where they just you know the operators are kind of investing in trying to get over the hump. Then eventually they’ll kind of get a takeoff. East Africa has had that. Few other countries in Africa have had that. This data by the way is also from the GSMA. You can see sort of the global landscape of where things are. Some countries have one, two, and now more and more this map would have looked a lot more sparse three or four years ago. Now most of them have more than one offering which is good because you do see the price drop a fair amount. You can see here regionally anyway sort of the total subscriber base. Twenty-one services now have more than a million active users. We see a million as a pretty good threshold for where things are really taking off and they’re getting good traction and that kind of thing. That’s been encouraging. That number has moved probably doubled since you know two or three years ago. Then five services have more than five million. That again, you know five million is a pretty active mobile money deployment at that point. But what you’ll also see is I think if you look at these circles. The dark center circle is the active user base. The clear circle of the area proportional is the registered user base. You can see that there’s large numbers of registered but inactive users. That’s a big problem that the industry faces. People will sign up and then either they don’t find the product compelling enough or they forget their pin. Or they don’t remember how to use it or what have you. That drives fair amount of cost in the system. It is again one of the issues we want to solve. >>: [indiscernible] can speak to that. >> Jake Kendall: What’s that? >>: [indiscernible] can speak to that. >> Jake Kendall: In the sense that they have a lot of inactive usage. Yeah, fair enough, maybe for different reasons. But, so I mentioned a couple times that you know we see mobile money as this kind of great demonstration case on some level and potentially a platform. But it has a lot of issues. Part of it is that you know we see this wheel is suppose to represent like you know what we think is possible in the longer run. Using a better digital platform and a better set of technologies and business models and you know the ability to kind of capture and support all of these different payment use cases, right. People getting their salary, people paying suppliers if they’re a small scale farmer or entrepreneur, tax collection, you know pension and other social transfers from the government. All the different payment use cases and then of course all the financial use cases on top of them that have payments as kind of integrated into their, into the system, right. If you’re doing credit you want the data that’s coming off of these systems. But you also want the ability to put into someone’s wallet immediately the loan and not have to have them visited by a loan officer and that kind of thing. That’s a huge difference in the cost structure if you can do it over a digital platform. You know E-commerce eventually, physical retail, all this kind of stuff. We think that in the long run you know if you want to cut out the high cost things in the system. High cost things are cash and physical person to person interactions, right, which tend to go together. What you want is so that the value starts to circulate in digital form and stays in digital form, right. Right now I send you money. You take it out and go buy stuff. You know we’re not really getting very far. It’s better than putting money on a bus. But you know you’re not really kind of creating a financial system where the sort of per transaction cost is you know minuscule and [indiscernible] small amortized over you know the whole system. You still have to have all those physical access points and people and stuff involved. You know if we think about can we make people loan cash in the sense that, or long digital value, so we pay them in digital value, pay for their crops in digital value. Get their money from the government digital value, all that kind of stuff. That gives them an incentive to spend it in digital value. But then you also have to give them the capability to do commerce, day to day merchants, all that kind of stuff. The world is very far, you know the mobile money world is very far from being able to kind of actualize on that vision. Now a little bit getting into what our teams trying to do. That’s really the background. We were involved quite a bit in the, globally in the scale up of mobile money systems. We gave you know exploratory grants to some of the early movers and that kind of thing. Have you know like I said supported the industry because we’ve seen it as a valuable thing. But we also think you know and we think its part of our role is that the industry itself you know there’s a next step to take really, right. There’s a, there’s much further that they can go in terms of creating an open scalable, more robust platform. I’m going to show you sort of two areas, right. One is the platform work that we’re doing, that our teams doing. As I said we’ve designed kind of a reference model for what we think the payment system needs to look like at that lower levels of income, or to support those types of transactions. That was a long a pretty in depth process of studying kind of the current systems and what their shortcomings are. Then I’ll give kind of a next generation set of technology pain points, or pain points that we think technology could readily solve. Just jumping on the platform stuff you know I was telling you the mobile money systems tend to be pretty bespoke, built for purpose, and not very robust, right. They’re built around this person to person payments, use case, very narrow sort of use cases. They are very unreliable, MPESA use to have you know I think probably three hours, three or four hours a week of down time. They’re better now. But you know that kind of latency or that kind of down time would just be you know unheard of in a credit card system. I don’t know if you’d use a credit card or a bank if you thought they were going to go offline a couple hours a week at you know completely unexpected times. You know that kind of lack of robustness. As well as, the inability to be open to allow third party players to integrate through APIs and that kind of thing. The mobile money systems are pretty basic and hard to work with. Then the sort of more in you know credit card based systems or banking systems that can support some these use cases have a lot of issues in terms of their cost structure primarily that make them hard to do. We did, like as I said, kind of a design exercise. You know talked to people all over the world and ended up building a prototype that are now going out to build a real reference model that will deploy in a couple countries. But we started with kind of the first principles, right. It needs to be digital for fairly obvious reasons, has to be robust, open, low cost, and secure, right. We wanted to be able to assure that those five principles were instantiated in any platform. We then kind of looked at what is the user need, right? What is the user experience need to be able to support? An easy account opening, you know lots of paper processes, and having to go to three different bank branches and all that kind of stuff you know really not going to work, right. You should be able to open your account yourself. If you have to go into a physical location it’s only to meet some sort of regulatory requirement to show ID or you have a fingerprint, or whatever. You need easy access to cash in and cash out. Now we think in the long run we want to get rid of that, right, by giving people better options than to cash in and cash. But no one’s going to start off using a system if they think they can’t get their money back. Obviously, it’s got to be safe you know it’s got to be trustworthy. You have to be able to move money to other individuals. I think that’s a difference in the developing world that’s probably different from here. That’s not as big a use case in the developed world, right. I rarely need to pay other people. But in the developing world where lots more transactions are informal and that kind of thing you want to be able to move money to other people. You need to be able to do direct deposits. You know people can get their wages and other things, direct deposit of wages, and these are government benefits. You need to be able to purchase things. You need transparent and low fees, right. You can’t have people thinking that they’re kind of getting charged all the time. That leads us to some of the design principles. We think you want self-issued account and sort of tiered levels of KYC. KYC is Know Your Customer, so how much do diligence you have to do should be tiered. Push payment model, most credit cards are all on a pull payment model. It’s kind of subtle as to why push payment; push payment is initiated by the user’s device. It’s not really, I think part of the reason why the Legacy systems are built around the pull payment model is that it wasn’t feasible to do real time communication when credit card systems were first developed. What happened is I give you my information. You as the merchant then go pull that money out of my bank account, right, with my “permission”. Then the digital credit card systems were just built on top of that model. That adds a lot of transaction backend stuff that’s in particular I as a merchant now can pull from anybody, right. I mean at least I could try. That’s where most of the fraud comes from in the system. There’s a whole bunch of extra risk and fraud mitigation stuff that has to be in place. Whereas in a push payment model I’m the only one who can activate my account, it’s through my phone and I have pin. If something goes out it clears immediately and is immediately transferred and that’s another principle. There two as well as the irrevocability lead to a lot lower cost structure. Most, a lot of the cost structure in credit cards is that you have the right to sort of revoke things. Merchants can charge you for no reason and so someone can fake being a merchant and that kind of thing. Of course open loop we want you know other third party players to be able to integrate in and use the system. System wide fraud and risk, so not that everyone has an individually siloed fraud and risk approach. Then of course high volume and low value payments are feasible within the cost structure of the system. >>: Does open loop include interoperability with other, two different services or multiple services? Or are you thinking more just open loop in terms of currency, merchants? >> Jake Kendall: Yeah, so open loop. Yeah, I mean in general should incorporate the ability for lots of different third party players who want to be able to move money in ways that the system support safely should be able to do that, right, through… >>: But it means sending money to a person that’s on a different service? >> Jake Kendall: Yeah, yeah, right, sorry, interoperability across platform as well, yep. >>: Okay. >> Jake Kendall: We actually see this, the level and project we’re thinking about it really at the payment systems level, so you know at the national level essentially. What do the solution sets enable some of these? You need an account opening service, consumer agent and account management service, smart. You know these are just sort of different account management modules essentially, interoperability for across platform interoperability and then fraud and risk. Then the broader national strategy has to be more than just the technology platform of course, right. You need the technology but you also need the regulatory framework. You need the right business models and scheme rules. You need certain governance so that the different scheme players feel ownership, operations, and the right level of communications, and that kind of thing. This was sort of the background. This is I’ll jump over this slide pretty quickly just because we’re getting on in time. But this is kind of the high level schema. We, I think the thing to look at here is this is sort of the old, the Legacy system, right, commercial banks, card networks, the central bank, etcetera. These systems are not very well geared for things that are explicitly mobile and very low cost and high volume, so, low-end consumers in particular but also small merchants, and things like that. The other thing to note is in this new system there’s kind of a set of core services that are shared. Then there is the stuff that each digital service provider is going to build on their own and where this is where the completion is in the market. I’ll show you some pain points; these are sort of the next generation issues. What we did to develop those was really to go through the different players inside this box. The DFS providers and the different end user stakeholders and think through kind of what are their. Given the system you know if it were setup what would they still have as pain points in order to sort of execute? >>: It’s the notion that you have a national digital financial a really good tie in backed with central bank that in terms of your policy, currency controls, currency leaving the country like Nigeria. Is that, that’s really embodied into that central pillar, right? >> Jake Kendall: Yeah, so central banks, right, I mean they, their kind of core mission is usually systemic stability and bank regulation. But they also often run various aspects of the payment system. They’ll setup and manage the ACH and all that kind of stuff. >>: Yeah. >> Jake Kendall: Then they setup the rules for, yeah, international transfers and all that kind of stuff. Some of them have all kinds of stuff like capital controls and things, but. >>: Yeah, that’s, I read about that. >> Jake Kendall: Yeah, yeah, this just too kind of give you a sense of all the different components that go into it is the next level down in detail schema for the reference model. You can see maybe in particular the stuff we see ourselves as commissioning to build directly versus stuff that’s done through partners at different levels of. Some of it is you know direct support versus supporting through partners. Then some of it is just stuff that we think other people will eventually build on their own. But kind of the core stuff of you know the interoperability integration and security, and things like that are kind of the things where. Then also potentially some of the merchant processing stuff, at least the core elements of it is things where we think we can add value. Because you know that’s not something the market will necessarily do on its own. Where we think we can work with the government for example to setup kind of a core public good system. Not necessarily an NGO, right, could be a for profit company. But we see it as a public good in the sense that it’s shared infrastructure for the broader economy. >>: You’re actually funding third parties to build to your specifications some of us would hope? >> Jake Kendall: Yeah that’s right. Yeah, so we will build a reference model. We’re talking to some of the bigger system integrators and other players. In the process of setting up like I said Pakistan and Nigeria we, some of these components already exist in each country. It’s a little bit of like filling in the gaps on some level. But we plan to build a full reference model. At the current plan instead of an open source, [indiscernible], and hope that and try to support an open source community where others can build to the same reference model. Some of those other builds may not be open source. We’re not sort of fixated on open source but more that we want to, if we’re going to pay for it, it should go into the public domain, basically. But that’s the platform stuff. That I think to me is an answer to the question what is the right architecture and schema, and business rules for the core of the system. In the time that we have left this is, these are kind of what we see as sort of next generation issues. In areas where we think more you know sort of direct point solutions from kind of emerging technologies are potentially relevant. These are areas that even if you have the architecture right and that you know L one P system in place are still hard and can still you know benefit by people inventing new ways to do them, basically. I thought I’d just go through them. But I’d love to hear you know I think we can kind of discuss individual ones and if you all have thoughts and things like that. The top one is fraud and cyberattack prevention, right. You know this is sort of obvious I think in a financial system we’re constantly being attacked. It’s usually kind of proportional to the amount of money involved. Currently there hasn’t quite been, maybe hasn’t quite been enough. I think also partly because the systems are built for purpose and closed loop, so all run within the datacenter of one mobile operator. There hasn’t been that we know of a whole lot of massive cyberattacks. But there’s certainly fraud of different kinds. They’re probably, and there’s certainly been insider fraud, so internal cyberattacks if you want to call them that. People who are insiders within the company and have managed to kind of like you know jigger it to put ten cents of every transaction into their own bank account or whatever it is. Of course you know most of that stuff just never comes. You know they don’t allow that to be publicized in the newspaper if they can avoid it. It’s hard to know how much of this stuff there is. But it’s certainly an issue in the long run because the bigger the systems get the greater the pressure on them will be. There’s a lot fraud of the form kind of petty fraud, person to person fraud. You know calling up and pretending to be the, you know whatever the services operator, trying to get peoples pins and all this kind of stuff, fisching stuff with text messages and all that kind of thing. Certainly more to come I think as it starts to get into the merchant business and that kind of thing. Proximity payments is another one and I think that’s one where the, why are we interested in proximity payments? I sometimes get that question, right. You know it’s not that we think poor people benefit by paying with a digital phone rather than with cash when they buy their groceries. I mean that you know it’s not really clear. But as I said before what we think for the long run sustainability and health of the system you need to achieve the ability to keep cash in digital form. For that to happen it’s really critical to get the proximity the ability to buy things in the store. Because that’s most of someone’s sort of you know day to day transactions are you know very, very small value proximate payments to someone who they’re buying. You know three tomatoes from in the market or something else, right, or soap, or whatever it is. But this is a real challenge, right, because cash just works super good as a payment medium for proximity payments. I mean just you know I hand you cash, I can count it out easily, I’m familiar with it. You know it’s super fast, easy, you know all that kind of stuff. >>: No fees. >> Jake Kendall: There’s no fees. >>: But today if you want to buy three tomatoes couldn’t you just do that by entering the merchant’s phone number into your service? >> Jake Kendall: Yeah, I mean you can but the guy selling three tomatoes on the side of the road typically isn’t a merchant. But even if he was the… >>: Except he’s a consumer or she’s a consumer… >> Jake Kendall: Right, so you could pay them… >>: Chances are she has a phone herself. >> Jake Kendall: Yeah, so there’s a couple aspects of it, right. One aspect of it is the fee and that’s not a technology issue obviously. But another aspect of it is just the user experience, right. I’ve got to punch in this guy’s phone number and you know like not get it wrong. Then you know go through the menus and all that kind of stuff. The current systems are really bad at that. We’re looking at stuff like NFC and barcodes, and other ways to you know put all the information for the transaction quickly from one phone to the next. You know some things that we think about and weigh as to whether we can make the assumption or not. That for instance one of the parties has a smart phone if they’re a merchant you know those systems, those approaches become a lot easier. Because obviously with a smart phone you can, you know you can read a lot more things. Because they have sensors and you can project a lot more data and things like that. >>: You’re thinking phone to phone? >> Jake Kendall: Yeah. >>: And phone to terminal, phone to point of sale terminal? >> Jake Kendall: Right, yeah you’d want to integrate in. Have the ability to integrate with Legacy terminal based POS systems and things like that. But, yeah, NFC, it could not be a phone, right, on some level, right. If you just had like an NFC chip or something like that. Yeah? >>: Are the phones for these markets built with mobile payments in mind? Or is it like they try and wear it on top of text messaging or something else? >> Jake Kendall: Yeah, so I actually skipped over that slide but the primary medium, so they’re not built with mobile payments in mind for the most part. The sort of the communications protocol that most of them tend to use is USSD which is just a really basic session based protocol. That a low end phone can initiate with just a code. You’ll have USSD short code. It’s like three numbers and it’s sometimes like star three numbers and hash tag. That pops up like a menu. But some of them also use just sort of the direct integration with the SIM toolkit and the ability to put a menu item in the actual SIM card menu. That’s better in sense of from the user experience it’s a bit more integrated and smooth. But often requires a SIM swap and so some operators have just opted not to bother trying to get all their clients to swap a SIM. Then some have tried to do it with kind of structured text message and things like that. But so the obvious issue with structured text message it’s not all secure. There’s no way to really prove you know the, which phone it came from and all that kind of stuff. Yeah? >>: Jake I’m going to have to take off but thank you. I run payments for Microsoft. >> Jake Kendall: Oh, okay. >>: I’ve spent some time with Roger and Costa before. >> Jake Kendall: Okay, yeah, yeah, yeah. >>: On some of this stuff. But it’s always good you… >> Jake Kendall: If you’ve got a card leave it and we’ll keep in touch. >>: Pardon. >> Jake Kendall: If you’ve got a card leave or send me an email and we’ll keep in touch. >>: I’ll send you a card. >> Jake Kendall: Okay. >>: Good. >> Jake Kendall: Just too quickly go through the rest, identity, woops. Identity is you know how do you identify people? We’re looking at a lot of biometric stuff. Thanks a lot. >>: Yeah, thank you. >> Jake Kendall: Yeah, there’s you know there’s a lot of challenges with, if you could easily figure out who someone is both for the transactional integrity. But also the onboarding, the KYC part where you’re actually signing them up. Most regulators want to know who people are so they can mitigate against terrorism, financing, and money laundering, and all that kind of stuff. That simple act of trying to figure out who someone is can actually add a fair amount of cost at least relative to the amount of revenue generated by a really low income person. We’re looking at like are there really quick ways to, can we capture a fingerprint through a mobile phone camera, basically. Or things like that so that we could really cheaply in software deploy identity solutions across some of the systems. Yeah? >>: [indiscernible] the essential the phone number, right. There is no… >> Jake Kendall: That’s right. >>: It’s what you can all anchor up, right. >> Jake Kendall: Yeah, so, well that’s your, that’s what you have, right, but it’s not really who you are. >>: [indiscernible] >> Jake Kendall: Anyone can have that phone number. >>: That linkage between them is sort of missing [indiscernible]. >> Jake Kendall: That’s right, yeah. It’s especially hard in the initial sign up process, right. Because if I sign up. >>: Yeah. >> Jake Kendall: I mean anyone can say I’m you know whoever, right. Like India has set up this really interesting system called Aadhaar or the Universal Identity. They have basically biometrically fingerprinted and iris scanned nine hundred million people at this point and, over the past couple years. They have a really kind of simple core system that basically you send Aadhaar number or identity number and the biometric. It’ll give you an answer is this the person you know it’s that biometric mass that match that number. They can also send back name, address, one or two other fields, really simple, right. But just that capability in and of itself is really core to all kinds of identity questions. That kind of system you know especially to the extent that new phones are starting to come out with biometrics and things like that. The main issue is that like sometimes deploying biometrics broadly over lots of different retail points and things like that can be challenging. Being able to, I think actually if we could do it in software only on a mobile that would be a pretty interesting… >>: Do you have any leads in how you could link those to? >> Jake Kendall: Yeah, so I’ve seen, there are some folks that like have a biometric, like a palm based biometric that you can do through mobile camera. Some extent it depends on what kind of verification you want to do. If it’s you know capture biometric and do a look up. You know the false positive array is just too, you can’t look up nine hundred million people and find the right you know exact individual, right. Or you’re going to get like fifteen yes’. But if you’re saying if you want to do it for like a transactional or if you’re going to capture that and then just put in the person’s name or something like that that’s fine. The main issue with Palm is that no ones, like there are no Legacy databases of Palms, right, in most countries that I know of. We’d love to have something that actually does fingerprints since that’s what a lot of these countries have actually over the last few years often supported by the UN or somebody else around a voting. You know an election or something like that. They’ve actually got big biometric databases that we try to boost of. But if you have to have some specialized biometric that’s not going to work. Like there’s ear prints and you know there’s all that kind of different stuff that you can do with a, potentially with a mobile phone camera. But actually fingerprints are harder, so that could be a, I think that’s an area that if someone came up with a solution it would be pretty important. Client activation is another one, has to do with again signing people up, getting them onboarded. But also kind of communicating with them after the point of actual physical contact with messaging and things like that. To keep them engaged and give them instruction and give them value, and things like that. Analytics is also a, you know clearly an area where you know mobile phones are creating a huge amount of data. If they’re also doing your payments and money transfer and things, that’s also you know a whole other level of data where there’s a lot of signal in terms of who you are and what you’re doing, and that kind of thing. Risk scoring for credit of course is a big area and currently in Fintech in the developed world where the people are using all kinds of alternative data sources to score. You’re seeing that trickle down. People are trying to use the mobile calling patterns and transactional patterns, and things like that. There’s also just in general you know the current you know mobile operators. They are sophisticated in very specific ways and then very unsophisticated in a lot of other ways, right. They’re very good at their kind of core system integrity business and that kind of thing and keeping the mobile signal up and all that kind of stuff. But they’re not very good at kind of other kinds of analytics and developing products and things that sort of leverage their data. Another one is CICO agents. Like I said this is an area where you know CICO agents cash in, cash out networks. They cost probably like I said fifty to sixty percent of the cost structure for these things is just keeping up those things up and running. There’s a huge problems with cash stockouts or digital equidity stockouts. Agents don’t, you know they don’t stock enough. There’s problems with how do you place agents in the right place? How do you manage them? A lot of those things I think could be solved with data analytics. I mean, there’s a guy at Harvard I talked to recently who’s just you know run a model if you could get the agents to restock their cash. According to his model you know when his model predicts they should restock and that sort of thing they’d save you know thirty, forty percent of their cost structure, or at least their cash in and out cost structure. Now whether or not you could actually get them to behave that way is another question, right. You have to figure out how to motivate them. But I mean that, you know that kind of thing I think the prediction stuff using the data that’s flowing out of the agent networks is currently not done at all. It’s actually incredibly unsophisticated the way agents are managed. There’s almost no sort of automation of messaging and prediction, and things like that. I think that’s potentially a really big area. Maybe there’s also sort of hardware stuff at the point of sale, right. Agents you know they get passed a lot of counterfeit currency in certain countries and being able to detect it, some simple cheap way to detect that would be really interesting. You know, I don’t know maybe there’s security devices really low cost, etcetera. >>: Did you say in many cases an agent is not typically operating on their own? They’re part of some group of agents that’s… >> Jake Kendall: Yeah. >>: Funded by a third party entity or an individual. >> Jake Kendall: Yeah, you see different models often the mobile operator will contract with a series of super agents who then manage a string of fifty to five hundred other agents. Those guys often do some sort of basic frontline support. You do see, but you see different structures. Some of them, some of it’s almost all in house and there’s no real subcontracting structure. It just kind of depends, yeah. Then another big area is financial managing and budgeting tools for end users. You know as with anybody low income people have a hard time keeping track. They have actually very complicated financial lives which is surprising to some people. Everyone things it’s very simple, right, you know zero balance and that kind of thing but like they’re actually very complicated. There’s, you know if you look at, we’ve done very careful studies tracking their financial patterns over many years for example. You’ll see that for example maybe ten times the value of their income will flow through their house in and out. They’ll borrow and lend and you know receive gift from somebody and give a gift to somebody. You know maybe ten X what they actually spend and consume goes in and out of their household. They really have a lot, a very rich life of informal loans, gifts. You know money transfers to friends in times of need you know all that kind of stuff. I think in addition and this is something that has come out in recent years in the research. The fact of being poor and kind of being pressed every day to make decisions and living a life where you know there’s very little certainty. They’re often they’re sort of finding piece work from day to day, no steady job, no steady paycheck. That adds a huge amount of cognitive load on people. It’s just something that sort of behavioral scientists are getting better understanding of. But it diminishes their capacity to make sort of long run planning and decision making and things like that. We think there’s a huge amount of potential for this kind of stuff if you could do it over the low end interface that a mobile phone represents, right. We’re doing experiments with like text messaging and trying to come up with kind of the right messaging framing formats to motivate people to save for example. Or motivate them to pay back debt on a regular schedule or that kind of thing. I think you know helping people with their budgeting. Helping them kind of you know with their “nudges”, behavioral nudges and things like that. Giving them maybe optimization tips, so like you often see people are borrowing at high interest and holding liquidity at zero interest, right. You know just sort of pointing those kinds of things out. I think there’s a lot to do here. It’s complicated because a lot of it is more behavioral science than it is technology. But if you can build the two together I think there’s a lot that can be done. >>: If you look the under bank you know population in developed countries. I mean there’s a huge you know when there’s such an asymmetry of financial power there tends to be. The people who are under banked tend to be overcharged you know. >> Jake Kendall: Right. >>: Things like check cashing services and payday loans and stuff like that end up tip toeing toward predatory stuff. How do you keep all of that kind of stuff out of this? Is MPESA like they’re for profit, right? >> Jake Kendall: Oh, yeah, yeah. I mean that’s, yeah. That’s a huge question for us. Especially like in the digit credit space, right. Like you see you know some of these guys are offering either through a mobile app or through kind of a backend service that’s white leveled through the mobile operator. You know what’s very attractive on some level, right, because I can on my phone go on, register, signup, and apply for credit, and get an answer within you know thirty seconds, right. I might get you know fifty, a hundred, two hundred dollars loan. My loan term is typically two weeks. For the short-term emergency that’s enough. But then my interest might be three, four percent over two weeks which if you sort of do the math over the year is, it’s about a hundred percent or something like that or a little more I think. You know I’m not pundit at doing math very good. >>: It’s just like payday loans here it’s worse than that, right. It’s like you know thirty dollars to borrow a hundred and fifty for you know ten days. >> Jake Kendall: Yeah and you see that kind of stuff too, right. >>: But digital credit in these markets that’s a new phenomenon, right? >> Jake Kendall: Right. >>: That’s, it’s just getting started. You said earlier it’s really you have to have cash to get the value on to your phone. You have to have the value on your phone to get cash out, right? >> Jake Kendall: That’s right. >>: You’re not floating or crediting people generally? >> Jake Kendall: No, I mean the system itself doesn’t have inherent credit built into it, right. But people are always trying to create credit products for people. >>: In the… >> Jake Kendall: Yeah and our worry is like the potential of it seems great, right. But then how do you make sure that eventually that interest rate comes down to a level that’s not you know frankly gouging people? How do you make sure that they actually get value out of it and aren’t just pressed into it by the urgency of their circumstances? I think competition is probably an answer to that in some sense, right. In the kind of fees that the mobile operators charge we’ve seen them come down when more entrants successfully enter the market and start to grow the size of their client base. You know I think that’s one of the main motivators, one of the big motivators for our level one project is to build a system where the mobile operator can’t just set up a one party system and keep everybody out. But to actually build a more open architecture for the payment system as a whole so that you get competition at different levels of the value chain that will actually drive the lower costs. But, yeah it’s certainly; I mean you can gouge people over digital just as much as you current gouge people in sort of manual and face to face transactions. That’s certainly true. That’s definitely something we think about a lot. >>: As a comment to your last point, are there places where like you [indiscernible] where you can actually get them to save and have [indiscernible] an account, and… >> Jake Kendall: For sure, yeah. I mean there are you know successful saving products and things out there. One of the areas where we’ve seen success is what we call commitment accounts. Accounts that have some sort of friction associated with taking money out. It can be as simple as like you commit to a group of other people who are savers that you’re working, you know that are part of your social network. You all kind of chastise each other, right, if you take money out. It can be that basic. But it can also be sort of more formal account features where you pay a fee to take money out if you take it out before you’ve reached your goal. You might set a goal of I don’t know two hundred dollars because you want to buy a sewing machine and start business or something, right. If you take it out before that you get a penalty. Or you want to save up for planting season so if you take it out before planting season comes you get a penalty, something like that. We’ve seen, you know we’ve run sort of randomized experiments with people who have those features and don’t. You actually see a fair amount of success in increased ability to save. Then even you know the knock on effects of that which are better planting, better investments in planting and farming, and that sort of thing. >>: [indiscernible] on the backend link there’s a bank account for them. Or like where is this money being put? >> Jake Kendall: It depends, in that scenario I was just mentioning was a bank. But most of the commitment products have been through banks mostly because the mobile operators, mobile money platforms are not flexible enough to instantiate that kind of account feature. They’re very inflexible. There just sort of really basic wallets. But I think they’re moving in that direction. Most of them are upgrading platforms and things like that. But our, I mean what we’d like to see is you know if that’s a niche product that some people want. You know you should be able to offer it without setting up the entire mobile money system, right. The creative entrepreneur who comes up with just the right sort of marketing and system of you know Avon groups. Or whatever the right thing is that people actually then start to buy into it should be able to access that system. Deliver that product over the system through an API or whatever it is without having again to build the whole thing from scratch. That’s what we’re hoping will happen. It is starting to happen in certain cases where some of the operators have started to allow a little bit more access to their systems. I don’t have, I’m out of slides. I’d love to hear any thoughts you guys have. You know if you just kind of have an idea down the road for things, technologies we should be looking at, or ways in which we could collaborate. I mean all; we have you know partners on the ground who are kind of struggling with a lot of these things. A lot of them are not very technologically sophisticated. You know the ability to kind of experiment, help them experiment with new approaches to this would be really valuable for us. I think you know for the reason to come out and talk to you guys is kind of, from my end to kind of get you aware of the problems. Then I guess if you come up with things, thoughts for me, or just want to talk about it, get more involved, etcetera. You know definitely reach out. I’d love to hear from you. Yeah? >>: You talked about making some of this reference implementation open source. What’s the timeframe for that? How would you kind of engage community to build some momentum behind that? >> Jake Kendall: Yeah, so that’s something we’re still really trying to figure out. Because as you know open source you know often sort of flops, right, and doesn’t really create that sort of energetic community around it. I don’t know what our strategy is going to be to be honest. We’re still like I said really wrestling with it. But the timeframe I think the, I think we have a prototype which is not to say a referenceable implementation. But really just a prototype that, like kind of like checks the box and will do basic things. But doesn’t have, kind of like the performance characteristics and ability to parameterize and things all that kind of stuff that you would want from a full-fledged system. That may go into open sort of public domain sooner rather than later next few months maybe. But again that’s, I don’t know how interesting that is except to kind of play around with. I think the bigger project will probably not be until at least the end of the year or maybe probably next year. My guess, but I’m not sure. It’s not actually, that one’s run by the other team, the one that’s doing infrastructure. I’m not probably as in touch with their timeline is. But my sense is it’ll be aways down the road. >>: Sorry, not to over resimplify but I will for a second. It seems like analytics, although you called it out on number five. It seems like analytics would play a role in number one, number three, five, six, and probably seven. >> Jake Kendall: Right, yeah. >>: I think about where you know Microsoft has been demonstrating some strength recently as in the Cloud, Machine Learning, and Analytics… >> Jake Kendall: Yeah. >>: [indiscernible] since I don’t know how much of the infrastructure that or the reference designs that you’re doing have Cloud based aspect to it. >> Jake Kendall: Yeah. >>: That could tie into the Machine Learning that’s coming along. But that would seem like an area that could unlock some value in a bunch of these different areas. >> Jake Kendall: Yeah, that’s a great point. It’s true these are not really kind of a MECE list of things, right, the fraud and identity being you know, right. But, and analytics definitely cuts across a lot of it. I think, yeah the idea of having a Cloud especially maybe in the fraud area or any of the shared services where data’s being shared… >>: Yeah. >> Jake Kendall: In particular probably a really good idea. One challenge is in a lot of countries. In some countries there are explicit laws that state certain types of data can’t leave the national boundary. Cloud based solutions are really tough in those countries for that reason. In other countries it would just be politically difficult. >>: Yeah. >> Jake Kendall: You know from an optics perspective. But might still be feasible might just take a little bit more sort of you know getting everyone on board. But I do think, yeah, that makes a ton of sense. Actually the whole thing could frankly be I mean most of it could be in the Cloud. >>: Yeah, well like in China for instance you’d really have to setup joint ventures. >> Jake Kendall: Right. >>: To basically operate within the national you know licenses and stuff like that. >> Jake Kendall: Yeah. >>: That COM, so you’d have to probably do something similar in other markets. >> Jake Kendall: Yeah that’s right. >>: Then the other thing is then, right, I think I probably know the answer. But, yeah the fact that we’re struggling on our Windows Phone market sharing and even with the Nokia acquisition. But we do have a lot of you know large screen devices. I just don’t know how much… >> Jake Kendall: Yeah. >>: Of the unbanked, how much of the overlap is there between… >> Jake Kendall: Yeah. >>: People that use mobile paid services and have larger screen you know computers at home or something like that. Whether that could… >> Jake Kendall: Yeah, I mean I, you know the vast majority of “our clientele” are really at best on a low end phone. >>: Right, yeah, a flip phone or with a candy bar… >> Jake Kendall: Yeah, exactly, yeah, if they’re on anything else as an operating systems it’s probably Android. >>: Yeah. >> Jake Kendall: I mean you guys probably know the actually the market share on the phone side you know in some developing countries. >>: Painfully we know it. We know it painfully well, yeah. >> Jake Kendall: Yeah, I feel like sometimes when I travel or I probably move the number by a fraction. >>: [indiscernible]. >> Jake Kendall: But, yeah, I mean the, yeah, I mean there’s not a lot of. I mean it’s even worse than here in the sense that like people haven’t even bothered to deploy a lot of the computers that they would have. Because they’re doing almost everything through, right, through Smart Phones. >>: [indiscernible], yeah. >> Jake Kendall: Yeah. >>: It’s just related on the analytics. I mean right now I mean all of that data is siloed per carrier, right. There’s no kind of if you want to do analytics for fraud or anything else, right. MPESA has its silo of data and everyone else theirs really. We can speculate about what kind of patterns they’re seeing and how things might improve. But there’s no actually no public access, right. >> Jake Kendall: Yeah, there’s certainly no kind of formal channel public access. We support different academic researchers. Some of whom have managed through to sort of hand shaking and cajoling get access. Sometimes to you know datasets that are maybe six months old or something like that. >>: Yeah. >> Jake Kendall: They do see pretty interesting patterns. We’ve done research on that dataset you know call detail records and event detail records that come off of mobile phones, right. As you said they’re all mostly trapped in the mobile operator silo. But they’re really fascinating datasets. I mean we’ve don’t everything from plug them into epidemic transmission models to chart where, like where is malaria going to go next and that kind of thing. Not even financial services related. But then also looking at like I said optimizing agent network, cash in, cash out patterns, and things like that. You know we’ve even tried to use them to figure out whose rich or poor, or who’s male, female. I mean you could just try to figure all kinds of things about the population if you have some sort of labeled dataset on the other side of who these people are. >>: Like whose data were you using for those classes? >> Jake Kendall: Each research project tends to be some other you know sort of specialized… >>: Are those one by one kind of negotiation for this particular project? >> Jake Kendall: Yeah, yeah, it’s very painful, takes like a year per dataset. >>: Yeah. >> Jake Kendall: But it is very interesting stuff. We’ve got a facility, a grantee that we’ve setup. One of their missions is to come up with a more systematic sort of standing order type approach. Potentially think through like what’s the right, because it’s very sensitive data, right. You can’t just take out the phone numbers and you can still figure out who everyone is, right, just by where they’re moving and all this kind of stuff. You, one of our grantee’s is sort of tasked with trying to figure out what’s a kind of like an aggregation format that would leave enough resolution in the data to be actually interesting. If you wanted to chart, for example, maybe mobility patterns for which there’s a lot of uses for knowing that kind of thing. But low enough resolution that you couldn’t [indiscernible] individuals. There’s you know obviously fairly technical ways to do that. But their goal is not, I mean that part actually is not so hard to work out in any given context. But then setting up the political process around it and potentially like a shared datacenter resource or something like that. That would then make it, make the data available appropriately to researchers or others you know in a kind of standing order basis. >> Shabnam Erfani: I have to interrupt at this point. We have another meeting coming up at three o’clock with folks who are calling in from Austin, so thank you Jake for the talk. >> Jake Kendall: Thank you guys. [applause] Like I said send an email or a text…