The final GAFIS report is out. “Big banks can’t serve low-income clients because small balance accounts just do not offer a sustainable business proposition” has been a truism since forever, and for good reason. This report takes a good stab at how that doesn’t always have to be the case, and how some of the biggest banks in sizeable developing countries are finding creative, segmented solutions to expanding the envelope of financial inclusion.
Between 2010 and 2013, the GAFIS project engaged five banks – Bancolombia (Colombia), BANSEFI (Mexico), Equity Bank (Kenya), ICICI Bank (India), and Standard Bank (South Africa). Together, they serve 77 million clients, and have combined assets of $250 billion, many of whom are the low-income segment. Obviously not all low-balance accounts belong to low-income clients, but the vast preponderance of low-income clients will have low-balance accounts; hence the focus on making them sustainable.
The final impact was described as follows (p.g. 38):
Collectively, the GAFIS banks have opened more than 4.2 million new accounts in GAFIS-linked products in less than three years. More importantly, they now serve approximately 420,000 “new, poor savers” as a result of the project (see Box I). This number is measured according to the GAFIS project definition, which requires evidence of both savings activity (870,238 accounts) and the poverty status of the account holders (543,119 new accounts meet both criteria); and which also weights the level of attribution to the project according to how directly GAFIS was involved (reducing the 543,119 to 419,654). Even with delays in launching new products at several banks, these growing numbers provide early indication that large banks can achieve scale outreach with their new propositions.
Very importantly, the business case for these 420k new, poor saver accounts was improved significantly through a combination of strategies:
- Costing methodology adjusted to product and channel specifications: It doesn’t make sense to load up branch costs into an acquisition expense if the product is mostly mobile phone based, for example.
- Using agent channels, where agents are essentially mom-and-pop stores set up to function as mini-banks: Transactions cost about a fifth as much with agents, where the agent is paid a commission. Account origination costs are in the $1-$6 range at agents, while those at branches are $16-$25.
- Increasing cross-sell: This might mean serving credit needs of clients, or fees from intermediating inflows from government or other sources of credit payments.
Here’s a summary of what each bank did to improve the business case (p.g. 22):
Still, all this does not make the business case for the accounts a positive one. It does significantly improve it though, from losing $2.79 per account per month, to losing $1.02 per account per month (p.g. 32):
Lot’s more details in the report – do check it out!
About three months ago, J-PAL had a great conference celebrating ten years of incredible work that arguably has furthered evidence-based development work more than any other institution out there. All the videos from the event are available here, but one particular one stood out for me. This session, titled J-PAL: The Next Decade happens right at the end, where J-PAL Directors Abhijit Banerjee, Esther Duflo, Rachel Glennerster and Benjamin Olken discuss what they would like to see happen over the next ten years.
Two things Prof. Banerjee said particularly resonated with me because they speak directly to the technique and underlying motivation for my PhD work, which I paraphrase below.
[9:40 - 10:30] First, he notes that there will much more extensive use of big data – data that is already collected “administratively”, and not with particular programmatic deliberation, such as cell phone call or m-pesa transfer records. One can assume that these will become more easily available This in turn will allow us to test very fine hypothesis with the power afforded by that much data – very small effects can be tested.
[13:05 - 13:55] Second, he expects to move more and more into theory-building based on the data from all these experiments. It starts with taking the results and then fitting it back to the narrative. As experiments accumulate, facts that don’t fit the story help build better stories as they are refined. New experiments are also designed to check the fit to those stories.
Remember that J-PAL is built mostly around RCTs, which take one hypothesis of causal connection within the context of a much larger story, and test that, and only that. To go from there to talking about much more freeform “big data”, and about theory-building as a natural evolution from theory-testing are two major steps away from their bed and butter. Given the complexity of the financial lives of the poor and the increasing availability of such data though, this is certainly a necessary and welcome evolution.
Mobile money is all the rage these days in inclusive finance circles, and for good reason. Mobile penetration has increased dramatically across the developing world over the previous decade, providing convenient rails for financial services to piggy-back on. It’s cheaper than Western Union and MoneyGram for the client, and cheaper than having to erect brick-and-mortar branches for banks. It’s faster than virtually any other method to send money to a relative in a pinch at the other end of the country, and safer than carrying cash around. And it has spawned an incredible array of services.
One of those is M-Shwari. It’s a savings-and-loan product offered by Safaricom in Kenya, in partnership with the Commercial Bank of Africa (CBA). M-Pesa clients can save into their M-Shwari account and earn interest, and take out short-term loans for a fee. It’s been hailed as a revolutionary product and discussed by the likes of CGAP, GSMA, NextBillion etc. highlighting fascinating aspects of mobile money adoption. And Safaricom promotes it as seamlessly blending convenience, safety and affordability.
I am totally sold on the “convenience” and “safety” elements of M-Shwari, but to buy the claim of “affordability” requires a somewhat unconventional understanding of that word.
But First, the Conventional Take …
So, how much does M-Shwari cost?
Q: Do you get charged interest on your M-Shwari loan?
A: The M-Shwari loan DOES NOT attract any interest. The 7.5% charged is a loan facilitation fee payable only once for each loan taken.
Q: If you have not paid your loan within 30 days, what will happen?
A: Your loan repayment period will be extended for an additional 30 days and you will be charged an additional 7.5% facilitation fee on your outstanding loan balance.
Q: If you pay your loan before the due date, will you still be charged the loan facilitation fee of 7.5% on the loan amount?
A: Yes, the 7.5% is a facilitation fee charged on the cost of processing the loan. Early repayment will increase your future loan limit qualification. Remember your loan limit is dependent on your previous loan repayment behaviour and usage of other Safaricom services such as Voice, DATA and M-PESA.
That’s a 7.5% flat charge on the nominal loan amount, with a maximum term of a month. The nominal and effective Annual Percentage Rates (APRs), “annualized” interest rates if you like, are:
- Nominal APR: 7.5% * 12 = 90%
- Effective APR: (1 + 7.5%)¹² – 1 = 138%
For additional details/nuances on APR voodoo please refer to MFTransparency or Wikipedia. XIRR comes to 146%, assuming the 7.5% is collected when the loan is repaid, and 164% if collected when disbursed.
Calling it a “facilitation fee” doesn’t change anything – for a month-long loan, the effective annual cost to the client is 138% or so.
In fact, that it’s a flat fee makes things worse. There is a pretty good reason why the 7.5% can’t be called an interest rate, since interest rates are time-defined. If the cost was actually 7.5% per month, someone borrowing for a fortnight only would be liable for a 3.75% charge. Not so with M-Shwari – you could repay in a day, but you’d still pay 7.5%. The nominal and effective APRs for a twice-a-month engagement comes out to be 180% and 462% respectively; anything shorter, and the APR is astronomically higher.
We’re hovering dangerously close to the “u” word, don’t you think?¹
Conventional Carrot-and-Stick Included
Two other “features” jumped out from the FAQs:
Q: What is the loan duration?
A: The loan is payable within 30 days. However, you can repay the loan before the due date and borrow again. If you pay the loan in less than 30 days your loan limit qualification will increase.
Q: If you have saved Kshs 5000 in your M-Shwari and have a loan of Kshs 2000 and do not repay within the loan duration (30 days), what happens to the money in your deposit account?
A: – When you borrow the Ksh 2,000, the money in your savings account will be frozen to the loan amount and the loan fee (loan amount Ksh 2,000 loan plus a facilitation fee of Ksh 150).
- You will only be able to access any balance above the frozen amount. The frozen amount will be accessible once you pay the loan. However you can continue to deposit money. Note: During the period the frozen savings will continue to earn interest which will be paid into your M-Shwari at the end of calendar quarter.
M-Shwari thus encourages faster loan repayment, offering an increase in the loan limit as the carrot. CBA sweetens the language even further by noting, “The earlier you repay your loan the better your chances of getting your credit limit increased!”
M-Shwari also seems to be fully collateralized, with both principal and “interest” held in escrow. Good incentive for borrowers to repay, also because they run the danger of being reported as delinquent 90 days after the loan is due, and potentially ineligible for another loan in 7 years. (See CBA FAQ, pages 6-7.)
This is all well and good from a profitability and prudential lending point of view, but it could be argued that simultaneously making the current loan costly and increasing future indebtedness on one hand, and conducting a complete bait-and-switch between savings and credit balances so that there is no net increase in availability of funds to the saver on the other, are not inclusion-friendly “features”.²
The “Affordable” Case
There is no math that can known down the glaring APR of 138%. Context, however, can help explain why M-Shwari has 2.4 million active users within 1 year of operations. Here are some of the reasons I think it’s been so :
- M-Shwari is primarily a savings product, with an option to take an emergency loan, as promoted by Safaricom and CBA. The maximum amount one can save is KES 100,000, but the maximum loan amount is KES 20,000.
- People are willing to pay fees on short-term loans akin to a service fee. It’s why payday loans can charge $15 for every $100 borrowed in the US. People are also willing to pay a fixed, simple fee compared to a more complicated, constantly adjusting rate, which is why microcredit interest is often charged at a flat rate. There may also be an element of pay-as-you-go, where bank clients are willing to pay a transaction fee each time they use a service, as opposed to pay a fixed monthly ledger fee.
- APR is not particularly useful for short-term loan products. No one expects a borrower to roll over 12 times in one year, bleeding KES 90 to access KES 100 through the year – we know even the unbanked are more sophisticated financial users than that.
- But most importantly, it’s convenient – it’s on that phone that is attached to your hips, it’s on the M-PESA rails that is ubiquitous in Kenya, and it’s near instant, secure and private. I know I would pay a 7.5% premium for a service like that.
M-Shwari therefore seems to be “affordable” despite having an APR of 138%. With no cash-handling or client interaction costs at branches, a relatively cheap source of capital in deposits from the same clients³, and non-performing loans at around a low 3.8%, this is one partnership that must make for quite nice margins for CBA and Safaricom.
¹ The “u” word is usury, in case you’re still wondering.
² It’s not clear if one can borrow more than one’s savings amount – the product pages and FAQs are deliberately vague.
³ Tiered interest rate is in the 2-5% range per annum.
Author: Ignacio Mas, Senior Fellow, Center for Emerging Markets Enterprises at the Fletcher School, Tufts University
With M-PESA and the whirl of innovations that it has triggered, there is no doubt that Kenyan payments are becoming more electronic. But, at the same time, are they any less paper-based? It’s hard to argue that is the case, if one looks at the Central Bank’s data. (Currency and GDP are from its statistical bulletins, and check volumes are from its Annual Reports; and remember that M-PESA was launched in April 2007).
The value of currency in circulation has remained essentially flat, especially if you discount the 2007 high blip (I doubt that M-PESA’s cash-busting bang was largest during its first nine months of operation). Likewise, the volume of checks is on an exceedingly gentle decline. The average check value has dropped quite significantly, but surely that’s due to competition from electronic funds transfers at the high value end rather than from M-PESA with its small-ticket transactions.
To me this lack of visible impact on paper highlights the two key pending transitions that M-PESA –and mobile money more generally—needs to undergo.
First, customers need to see value in storing their balances electronically. As long as most customers have the practice of withdrawing any electronic money they receive immediately and in full, M-PESA will remain essentially a cash-to-cash service, and as such it sustains rather than reduces the role of cash in the economy. (My recent mantras: M-PESA is better cash, not better than cash and you can’t go cash-lite on empty accounts.)
Second, businesses need to see mobile money as an easier way not only of paying and getting paid, but also of managing the information around those payments. It needs to link with order management, invoicing, accounting, reconciliations; possibly even inventory and fleet management. Mobile money needs to have the kind of flexible application programming interfaces that allows corporates to handle transaction flows seamlessly within their own systems rather than as a separate universe of transactions. It must solve basic trust issues that arise when there is no prior relationship between buyer and seller. (My recent mantras: solve business paint points around mobile payments and think of cash as a highly-evolved visual-acceptance payment instrument.)
Without these two transitions, to more electronic storage of value and flexible interfaces into business IT systems, mobile money will continue to be an extremely useful extension of the Kenyan payments system, but it will hardly be at the core of it. The core remains very paper-centric.
Ignacio Mas is currently a Senior Fellow at the Fletcher School’s Center for Emerging Market Enterprises at Tufts University, a Senior Research Fellow at the Saïd Business School at the University of Oxford, an Associate with Bankable Frontier Associates, and an independent consultant. You can find further details of his work at: http://www.ignaciomas.com.
Yes, this blog was in a bit of a hiatus. While I put together some more original stuff (and hopefully welcome additional contributors), let me share two maps with you that are pretty cool not just because they are quality products in their own right, but also because of how they were put together.
Source for everything in this post: 5 Maps That Could Help Solve Some of the World’s Most Daunting Problems
My favorite of the five shows the outreach of financial services in Nigeria:
… the Gates Foundation’s Financial Services for the Poor program launched an ambitious geospatial project in Nigeria. Armed with BlackBerrys and traveling in a grid pattern, 30 surveyors undertook a three-month campaign in Nigeria to pinpoint all of the services available—from bank branches to microfinance institutions to post offices and more. They logged GPS coordinates and took photos. … The resulting website gives the best picture yet of how financial services work in a developing economy. Jake Kendall, the Gates Foundation program officer who heads the project, hopes financial institutions will use the map to figure out where to open new locations. It can also guide public policy: Kendall’s team used this map to show the Nigerian central bank governor the need for mobile-based financial services, since it demonstrates that more than half of residents live outside a 3-mile radius of brick-and-mortar options.
A good map inspires curiosity. I find myself asking, how do you get to people who do not have mobile coverage? (Those are the non-purple parts.) Mobile money agents are often touted as a great list mile solution – yet, they often overlap with existing infrastructure, and in a fair few cases, are missing where other delivery channels, such as post offices, are present. Are mobile money agents really that redundant? Are there alternative delivery channels for financial services one should explore?
The second map is of the biggest slum in Kenya, Kiberia (the Wired piece lets you zoom into the map):
African maps are notoriously problematic. Much of the data is old; roads, particularly footpaths, languish unnamed. Africans often navigate by informal landmarks like bars or gas stations, places not represented on standard maps. The slums have it even worse: On Google Maps they figure as blank expanses, in keeping with their reputation as shadowy, marginal places.
Enter Spatial Collective and Map Kibera. These two organizations, a company and a nonprofit, are mapping a Kenyan mega-slum called Kibera—the name is derived from a word meaning “jungle”—according to how its 200,000 inhabitants actually navigate it. The maps started with crowdsourced landmarks important to locals: water taps, schools, pharmacies. Residents with Internet access were invited to add to an open source map; others contributed data by SMS or attended community workshops, where they wrote on giant empty maps.
Spatial Collective overlays these community-generated maps with official data. One project tacked on sewer-line data (depicted here) from Nairobi City Water to find the most valuable spots to build new public toilets. Another mapped community-reported crime data to help the World Bank understand where to place safety interventions like lamps. These kinds of projects make Kibera more legible to its inhabitants and to outsiders.
In addition to the ability to provide efficient placement of new services, the fact that this map was crowdsourced is particularly noteworthy. By collecting information deemed useful to the residents of the slums themselves, whatever work done based on this will presumably be more relevant to them. Incidentally, comments in the article suggest that this is built on OpenStreetMaps’ work in mapping Kibria (http://mapkibera.org/) – that in itself is worth checking out.
… or, A Story of How a Coding
C*Screw-Up Made Bangladesh One of the Least Tolerant Countries in the World. (Spoiler: It isn’t!)
What We Thought We Knew
Yesterday, the Washington Post put out a story, A fascinating map of the world’s most and least racially tolerant countries. In that map, India and Bangladesh stuck out like a baboon’s butt as bastions of intolerance in the world. It’s reproduced below; red implies that more people said they “would not want neighbors of a different race”.
In fact, the percentages are more than a little damning:
“India, Jordan, Bangladesh and Hong Kong by far the least tolerant. In only three of 81 surveyed countries, more than 40 percent of respondents said they would not want a neighbor of a different race. This included 43.5 percent of Indians, 51.4 percent of Jordanians and an astonishingly high 71.8 percent of Hong Kongers and 71.7 percent of Bangladeshis.”
Three thoughts occurred to me in this order:
- Wow that’s an odd basket of countries to be lumped together as the most intolerant!
- Ouch! Yeah I’m Bangladeshi.. and while I’ll be the first to admit we have our own favorite national stereotypes and periods of ethnographically inspired excitement, least tolerant? Really?
- I wonder if someone fat fingered on this big time.
Thanks to the fact that both Max Fisher of WaPo and World Values Survey folks freely shared their sources, we can take a dive into the data that generated the map to explore thought #3 to our heart’s content.
The short answer is, yes, someone did fat finger this big time. “Yes” and “No” got swapped in the second round of the survey, which means that 28.3% of Bangladeshis said they wouldn’t want neighbors of a different race – not 71.7%.
26K Facebook likers and 2.5K Tweeters, take note.
Now, the long version for the data wonks amongst you. By the way this piece is restricted to Bangladesh – time, and ability to read primary questionnaire being main constraints.
What the WVS Data Really Says
(Spoiler: Data says it’s confused…)
There were 5 waves of data collection:
Bangladesh has data for the third and fourth wave.
First, lets reproduce the 71.7% number. We can use the interactive query tool WVS has set up on their website at: http://www.wvsevsdb.com/wvs/WVSIntegratedEVSWVSvariables.jsp?Idioma=I
“Mentioned” basically means “Yes” (we’ll explore this in detail in a bit).
What about the 1996 survey? Here’s the same page with 1996 data (table shown only):
If the oddity isn’t jumping out to you yet, let’s use something a little more visually friendly to compare the 1996 and 2002 numbers. WVS also makes available a Online Data Analysis toolkit at http://www.wvsevsdb.com/wvs/WVSAnalizeQuestion.jsp. With a literal tinkering, we can get to this screen:
“Mentioned” (the measure for less tolerance) went from 17.3% to 71.7%, while “Not Mentioned” went from 82.7% to 28.3%.
a) something WTHBBQPWN&%$#* happened in those 6 years that made 54.4% Bangladeshis do a 180 degree on their tolerance levels, OR
b) the coding got messed up and “Mentioned” is either 82.7% and 71.7% in 1996 and 2002 respectively, or 17.3% and 28.3% respectively.
For a survey size of N = 1,500ish, b) is always your safer bet when no obvious change agent is involved, endogenous or exogenous.
(If you’re Bangladeshi, you’re also probably laughing your behind off at a) since the 54.4% number involves tens of millions of people in a generally syncretistic society where appreciation for that heritage has arguably only increased with the younger generations, but that’s “anecdotal” for the purposes of this piece, so we’ll leave that thought there.)
If your data analysis foo is up to the task, I’d encourage you to check out the raw data itself to confirm that this is also the case there. The dataset is generously available at: http://www.wvsevsdb.com/wvs/WVSData.jsp
I used the STATA file for the “WVS FIVE WAVE AGGREGATED FILE 1981-2005″ dataset; you can also choose SPSS or SAS formats if they suit your toolkit better. For Stata, the relevant command is:
tab S002 A124_02 if S003 == 50
Giving the result:
| Neighbours: People of | a different race Wave | Not menti Mentioned | Total --------------------+----------------------+---------- 1994-1999 | 1,261 264 | 1,525 1999-2004 | 425 1,075 | 1,500 --------------------+----------------------+---------- Total | 1,686 1,339 | 3,025
(S002 is the Wave, A124_02 is the question under study, and S003 contains the country, with Bangladesh being code 50.)
What Should The Data Say?
(Spoiler: Bangladeshis are a tolerant bunch – it’s ok to come visit.)
Ok, now that we are reasonably certain the data is confused, which way will it point once we de-confuse it? For this, we need to turn to the actual questionnaires. These too are available at http://www.wvsevsdb.com/wvs/WVSDocumentation.jsp?Idioma=I.
First, the 1996 survey. The relevant section is reproduced below (Bangladesh_WVS_1996_1.pdf, pg 10):
The first column, উল্লেখ করেছেন, means “Mentioned” – to be selected if the respondent notes a particular group of individuals (V51 – V60) are unwelcome neighbors.
The second column, উল্লেখ করেন নি, means “Not Mentioned” – this is for the more chillaxed bunch.
Now, let’s look at the 2002 survey. The relevant section is reproduced below (Bangladesh_WVS_2002_1.pdf, pg 19):
The first column, প্রতিবেশী হিসেবে পছন্দ করবো, means “Would like [X] as a neighbor” – to be selected if the respondent notes a particular group of individuals (V68 – V77) are welcome neighbors.
The second column, প্রতিবেশী হিসেবে পছন্দ করবো না, means “Would not like [X] as a neighbor” – this is for the less chillaxed bunch.
’1′ and ’2′ stand for totally the opposite things in the two surveys.
Unless we are willing to allow that the data input folks consciously converted a ’2′ in 2002 to a ’1′ in 1996 to connect প্রতিবেশী হিসেবে পছন্দ করবো না to উল্লেখ করেছেন, I think it’s reasonable to assume that ’1′ was also coded for “Mentioned”/উল্লেখ করেছেন in the 2002 dataset, leading to a flip in the results for that question for that wave. Spot checks also suggest that this is what would be consistent with surveys from other countries.
With everything righted the right way, here then is what the final numbers look like:
Yes, 28.3%, not 71.7%.
- I only looked at the question that was used for the tolerance/intolerance WaPo piece. This inconsistency shouldn’t be extrapolated to any part of the rest of the Bangladesh survey unless one has double checked for that.
- I ran this for the Bangladesh dataset only. No idea if any of this applies to any other country dataset – same caution as above applies.
A little while back, we realized we were nipping at the heels of “big data” territory, arriving at near 1 terabytes of data. We being the team at Bankable Frontier Associates I work with, through a partnership with the Center for Emerging Markets Enterprises at Fletcher. Not quite a size that would titillate folks who drink Hadoop and sleep in Elastic Clouds, but it was a sobering moment that caused for some reflection on the limits of what we could do with all this information, even as we strained a pretty souped up machine to it’s limits.
Most of my concerns stem from the fact that big data has the disconcerting property of confessing to something – anything – under sufficient coercion. It’s a variation of the age-old problem of statistical correlation, aptly captured in the XKCD to the right –>
As the venerable Nassim Taleb points out, “We’re more fooled by noise than ever before, and it’s because … with big data, researchers have brought cherry-picking to an industrial level. … I am not saying here that there is no information in big data. There is plenty of information. The problem — the central issue — is that the needle comes in an increasingly larger haystack.“
When you’re dealing with data on tens of millions of accounts and billions of transactions from financial institutions serving clients in eight countries, that’s a rather massive haystack to get lost in. In situations like this, it is ever so important to have a set of null hypotheses that can be proved/disproved conclusively, thereby keeping us honest, instead of chasing spurious connections.
Which brings us to correlation vs causation. Yes, we have granular transaction data over a course of years for each account holder, meaning we know everything they are doing with that account. We can also have up to twenty characteristics of the client and the account type – age, gender, income, occupation, age of account, interest rate paid, etc.
But unless such studies are paired with detailed financial diaries, we know nothing of the individuals motivations for why they do what they do, or of the rest of their financial portfolio and financial tools at their disposal. This means we usually cannot say things like, “the average account holder saves Ksh X for her child’s school uniform”.
And that’s ok.
Causality in the social sciences is a hard problem. It’s not possible to hold “everything else constant” like we can with the hard sciences. Human free will allows for a mind-boggling array of choices, people may not always take the same decision despite being faced with the same choices, and somethings an effect may have multiple contributing causes.
Quantitative researchers do the best they can to account for all possible explanatory variables and then attribute degrees of causality to certain variables. Because we don’t have all possible explanatory variables when dealing with big data, we restrict ourselves to demonstrating strong correlations and usually end up indicating potential causal connections and let others take it from there – such as field researchers who can conduct focus groups to dig in deep.
This may not sound intellectually gratifying, but it is once you get into the thick of things. Let’s consider the example of two savings types: A, which is a short-term, low-balance almost transactional behavior, and B, which is accretionary savings over the course of a year leading to a decent balance. A is strongly correlated with ATM card usage, while B is strongly correlated with branch usage. 20-40% of all savings accounts seem to display A-type behavior, while about 1% display B-type behavior across many of the financial institutions we have looked at and I can talk about. What questions come to mind? How about:
- Do ATMs make it hard to save larger amounts over the long term because it’s just so easy to take money out? Do branches make it harder for folks to withdraw funds willy nilly and therefore save more over the long term?
- Or.. do clients self-select to use ATMs in cases where they need easy access to money and intermediate small amounts through that channel, leaving large amounts of transactions aimed at towards building that large lump sum for some purpose to happen at branches, not least because they don’t feel safe hauling a satchel of catch to an ATM in the middle of nowhere?
The implications of potential answer(s) can be profound. The first would imply that while we have celebrated ATMs as a successful de-congestion measure for banks, reducing staff load, client wait-times and operational expenses associated with physical branches, they have also caused people to save less, which can be antithetical to the cause of financial inclusion. On the other hand, the second would imply that branches still have certain benefits that are not being captured by other channels, and more effort needs to be made to address this convenience/security factor.
Of course, as with any complex system, the actual answer probably contains kernels of truth from both possibilities, and then some. Unless ridiculously fortuitous natural experiments present themselves with just the right incentives, say through subtle product rule changes intended to “nudge” a certain type of behavior, it’s well nigh impossible to seek answers to these kinds of questions irrespective of how big “big data” is.
(Btw, having 1% of accounts display a particular type of behavior across different banks in different countries is highly interesting in itself, since there is nothing definitional that would force this to happen. But that’s another story.)
I, for one, sleep peacefully at night knowing that often, all I can expect to get from “big data” are glorified correlations; anything else is gravy.
Although most of my work is with microfinance institutions, I have the good fortune to catch up with great institutions doing quite innovative work every now and then. D.Net is one such institution, who I interned with way back in 2006. Most of their work is in the realm of ICT4D (ICT for Development) and over the years they have pushed out a couple of quite successful technology-reliant solutions.
One of the latest one they are working on is this concept called the Info Lady. It’s an entrepreneurial woman who is provided skills training and a notebook and other digital equipment, who goes around in a bike in her own community and those near it and provides livelihood information, usually for a fee. Sounds a little … far-fetched, right?
Turns out, it actually works and is a sustainable business model. The original idea was inspired by D.Net’s success with Pallitathya Kendras, or Village Information Kiosks where people would come to these centers and obtain all manners of livelihood information for a fee. The Info Lady model simply extends this idea by taking that information to the doorsteps of clients, literally.
There are 90 possible services that could be provided, but based on demand most Info Ladies provide 20 to 25 services. Offerings include everything from health services (blood pressure, pregnancy tests etc.) to information on income generation activities to assistance for workers seeking employment abroad to taking pictures or videos at weddings. In fact, the last one has turned out to be quite popular and is also a good source of income for the Info Ladies.
Info Ladies require a fair bit of training on domain knowledge and technical expertise, which requires an extensive training program over a couple of weeks. The program is being rolled out in a handful of districts. There are 57 Info Ladies in the field at the moment, with 105 undergoing training. The target is to have 300 Info Ladies offering their services by the end of the quarter.
There is an upfront cost to getting the Info Ladies set up. The total up-front cost can vary between $1,500 to $2,000 depending on the equipment options. D.Net is subsidizing a part of the costs, but the Info Lady is expected to come up with most of it on her own, or take responsibility for most of it. One option is a 3-yr loan from the public National Bank that comes with a 3-month grace period and a 9% interest rate (commercial interest rates are around 19%). The average earnings reported is about $150 per month, which makes a monthly loan servicing amount of $50-$60 conceivable. By the way, the highest consistent earnings reported is a little over $600; while this can’t be taken as the average scenario, it does show the earning potential of the practice.
D.Net is also trying to get the Bangladeshi Diaspora involved by getting them to sponsor Info Ladies – another hallmark of D.Net projects.
Here’s an article that came out in The Daily Star (link to an archive because the paper’s website is currently migrating to a new one..). And here’s a powerpoint that gives a bit more detail on what an Info Lady does:
There’s poverty everywhere …
We’ve spent a fair bit of time talking about various methods to study poverty and development in this blog, ranging from plain vanilla surveys to financial diaries, RCTs and portfolio analytics, and it has all been within the context of developing countries. It is a reality that poverty exists in the developed world too, including the most vibrant economy in the world, the US. This post will talk about the US Financial Diaries program.
Just to put things in context, here’s a map of the world showing poverty percentages compared to the national poverty line:
… including the US
This tells us between 10% and 20% of the population live below the poverty line in the US. The Oracle provides us with some more detail:
Poverty is a state of privation or lack of the usual or socially acceptable amount of money or material possessions. According to the U.S. Census Bureau data released Tuesday September 13, 2011, the nation’s poverty rate rose to 15.1% (46.2 million) in 2010, up from 14.3% (approximately 43.6 million) in 2009 and to its highest level since 1993. In 2008, 13.2% (39.8 million) Americans lived in relative poverty. In 2000, the poverty rate for individuals was 12.2% and for families was 9.3%. In November 2012 the U.S. Census Bureau said more than 16% of the population was impoverished, and almost 20% of American children live in poverty.
And details can be found in all its glory right from the horses mouth; some of the highlights are copy/pasta-ed below:
- In 2011, the family poverty rate and the number of families in poverty were 11.8 percent and 9.5 million, respectively, both not statistically different from the 2010 estimates.
- As defined by the Office of Management and Budget and updated for inflation using the Consumer Price Index, the weighted average poverty threshold for a family of four in 2011 was $23,021.
- The poverty rate for males decreased between 2010 and 2011, from 14.0 percent to 13.6 percent, while the poverty rate for females was 16.3 percent, not statistically different from the 2010 estimate.
- In spring 2012, 9.7 million young adults age 25-34 (23.6 percent) were additional adults in someone else’s household. The number and percentage were both unchanged from 2011.
- In 2011, 13.7 percent of people 18 to 64 (26.5 million) were in poverty compared with 8.7 percent of people 65 and older (3.6 million) and 21.9 percent of children under 18 (16.1 million).
- The South was the only region to show changes in both the poverty rate and the number in poverty. The poverty rate fell from 16.8 percent to 16.0 percent, while the number in poverty fell from 19.1 million to 18.4 million. In 2011, the poverty rates and the number in poverty for the Northeast, Midwest and the West were not statistically different from 2010. The poverty rate in the South was not statistically different from the rate in the West. In addition, the Northeast poverty rate was not statistically different from the rate in the Midwest.
More than one in eight individuals living in poverty is bad, right, given the relative abundance of wealth in the US? Consider the fact that 16 million (22%) of all children live in poverty – a number exacerbated along ethnic lines (38% of Black and 35% of Hispanic communities) and one can see how this is an important issue to address in the US. [Source for numbers.]
Enter the US Financial Diaries
There’s been much renewed interest in the lives of the poor since financial armageddon a few years ago. Michael Barr’s No Slack: The Financial Lives of Low-Income Americans is a recent publication that based on 1,000 in-depth surveys that elaborate various ways in which the financial system “fails the most vulnerable Americans”. The U.S. Financial Diaries project is driven by similar motivations – it wants to understand how low-income individuals and households manage their financial lives.
You’re probably familiar with the financial diaries methodology championed by PoTP – conducting repeated surveys exploring every detail of a HH’s financial life for an extended period of time. This provides unprecedented detail, specially into behaviors that are difficult to pick out in one-off surveys, such as extent of engagement with the informal sector, and savings intermediation habits. As one can imagine, the presence or absence of various financial instruments creates a completely different ecosystem than say what we see in Kenya or India or the Philippines.
Non-probability sampling techniques are used, so the results will not be representative of the entire US. “Sites have been chosen to ensure geographic, ethnic, and racial diversity of households and urban/rural communities,” according to the project website, with sites including “Cincinnati and surrounding areas, Northern Kentucky, Eastern Mississippi, San Jose and surrounding areas, Queens, and Brooklyn”.
The project is in the data gathering stage right now, and we can expect analysis to start coming out in 2013 itself. It’s got the heavyweights in this field behind it – it is being administered by New York University’s Financial Access Initiative (FAI), Bankable Frontier Associates (BFA) and The Center for Financial Services Innovation (CFSI), with funding from Ford Foundation
and Citi Foundation and the Omidyar Network.
I for one can’t wait to see what comes out of it – the chance to compare and contrast with similar studies around with world will be quite interesting, as will be any follow up work that utilizes this information to design interventions to address the poverty situation in the US.
Saving in a Lending-to-save Product
We know that folks who have to deal with incomes that are low, irregular and uncertain have to resort adapting available financial instruments to meet their idiosyncratic needs. This is another post on one of my favorite datasets – P9 – that illustrates a simple but powerful adaptation. (You can read previous post here.)
You’ll recall that P9 is a lending-to-save product, where a certain proportion of the earmarked amount is held back as savings, which is then replaced with cash flow from the client once the loan portion has been paid off. This implies that you have to pay off the loan amount first, before you can really save. If nothing else, the discipline of paying off the loan in small increments is transferred to saving in small amounts towards a large lump sum.
Except, what if you only wanted to save, and didn’t need or want the loan?
It seems that a certain portion of the clients at the Hrishipara site (P9 is offered in two sites – Hrishipara and Kalyanpur) have adapted the product to this end by paying off the loan within the first day of disbursement presumably using the same amount they had taken out, and then spend the next few weeks or months saving up. Clients thus seem to have taken the conscious decision to do away with the lending half of the “lending-to-save” model but have voluntarily taken on the discipline expected of them as they save up towards the amount held in escrow on their behalf.
Tracking Down the “Only Savers”
The first clue that something was not going exactly according to plan was this plot:
This plot tells us what percentage of the tranche is paid off as the first payment. To fully grasp what this is showing, let’s first set some expectations. Say you decide to pay off an outstanding amount of Tk 1,000 in 10 equal installments of Tk. 100. How much of the tranche are you paying off per payment? Why, 10% of course (Tk. 100/Tk 1,000). What if you decided to pay it off in 20 installments of Tk. 50? Each payment would then constitute 5% (Tk. 50 / Tk. 1,000).
Of course, this can also be calculated by taking the reciprocal of the number of payments as a percentages – 1/10 = 10%, 1/20 = 5%, and so on. We wouldn’t expect the first payment to be anything different per se from the “average” payment, so our expectation of the size of that first payment would also be 10%, 5% or x% depending on whether we expect 10, 20 or n payments, where x = 1 / n as a percentage.
Thus, the graph above tells us that in 56% of the tranches, the first payment is 10% of less than the entire disbursement amount – something we would expect. But check that 27% in the blue circle – these folks have paid off around half the disbursement amount through the first payment. And the clients in the green circle – the 5% – have paid off almost all, or all, of the disbursement amount right at the first payment!
What is going on with the folks in the red circle!?
The examples are pretty self-explanatory. The table below is for the blue “Save Only” folks – you can see the almost-equal amounts for the loan and the repayment made, with the delta essentially being a fee of Tk. 10-100.
And the table below is for the green “Ramp Up” folks – you can see that the repayments are equal to the disbursement amount:Yes, clients are paying off the entire tranche amount. This is generally done because you have to cycle through smaller tranches before you are earmarked a larger tranche, and these guys have simply decided to do that cycling in one go. Most clients will cycle through one or two such tranches, but one particularly adept client went through 7 tranches in 8 days, cycling from Tk. 3,000 to Tk. 13,000.
I have to say, it’s not often that a pattern jumps out like this – if only portfolio analytics was generally this readily discoverable!
Adaptation Behavior Over Time
How consistent is this “savings only” behavior? Do they do the same thing tranche after tranche, or do they go back to taking advantage of the loan option? If you consider the blue circle folks as “Saving Only” and the green circle folks as “Ramp Up” clients, with the remaining as “Neither”, you can envision a 3 x 3 transition matrix between each tranche where a client in any of the three “states” can choose to be at any of the other three “states”.
The complete state transition figures are given below as a percentages of the number of accounts that have gone through that tranche. We stop at the 20th tranche because less than 50 accounts have gone through more, resulting in a lot of noise.
That’s a lot of numbers.. so let’s just focus on these three rows: “Neither -> Neither”, “Neither -> Save Only” and “Save Only -> Save Only”. The first goes from 74% to 44%, the second fluctuates between 2% and 14%, and the third goes from 8% to 26%. Thus, fewer and fewer clients continue the lending-to-save model, and more and more save only.
A closer snapshot of this dynamic is given below by focusing on the two states of “Neither” and “Save Only” and looking at the 2nd, 10th and 20th tranche:
What Does This All Mean?
Well, at the end of the day it’s fairly simple – P9 at Hrishipara has certain rules that its clients found a way to serve their need better when they were interested in saving only. Quantifying the phenomenon gives us a sense of how widespread it is, and allows product designers to account for deviations from expected behaviors. (I haven’t looked at the P9 Kalyanpur data yet but my sense is that the product there is more flexible and accommodates this behavior already.)
One subtlety that you’ll probably appreciate is this usage behavior indicates the preference clients have of having the option to draw down a loan amount even if they do not exercise that option all the time – in fact, around the 20th tranche, about a tenth of the tranches exercise the option to draw down after saving only in the previous tranche.
The write-up on which this post is based can be found at the P9 Databank. It benefited greatly from Stuart Rutherford’s feedback.