Despite the many successes microfinance has achieved in delivering financial services to the world’s poor at scale, it has often been criticized for the relatively high interest rates it charges on loans. Microfinance practitioners know these high interest rates are not a function of risk but are necessary given the high operating costs associated with making small loans to low-income entrepreneurs in developing countries.
Download “Microfinance in the Digital Age”
In 2015 I wrote a blog post entitled “Microfinance in the Digital Age” in which I observed how technology was profoundly transforming the microfinance business model and wondered if this might enable a future in which more micro-borrowers could be reached more cheaply and efficiently. Could technology reduce the operating costs of microfinance institutions (“MFIs”) and allow them to charge lower interest rates to their customers? In essence, could digitization achieve the “Holy Grail” of microfinance? Eight years on, MFIs are still charging relatively high interest rates, and journalists are still criticizing them for it (most recently Bloomberg in 2022). So why hasn’t the promise of technology been realized?
In June I attended a workshop organized by CGAP(the World Bank’s “Consultative Group to Assist the Poor”) entitled the “Digital Future of Microfinance: A Realistic Path towards Big Dreams.” It was an eye-opening discussion. Participants included MFI networks and members of CGAP’s team of digitization experts presenting lessons learned from automating the core business of MFIs over the last several years, many of which struggled to generate customer or business value from these investments. The insights shared were interesting and shed light on the root causes of why digitization efforts can fail to live up to expectations.
The following is a distillation of the learnings and experiences shared by participants at the workshop. I have withheld attribution to maintain confidentiality.
At a very high level, digitization projects can run into trouble for reasons related to governance. The market and the technology for automating financial services are moving fast, so shareholders and the board must understand what is needed and have confidence that the C-suite has the knowledge and experience to implement technological changes. Digitization often creates tension between the CEO and the chief risk officer (“CRO”), who might not be comfortable, for example, allowing an algorithm to make credit decisions. By the same token, the chief technology officer (“CTO”) must also be up to speed on the latest trends in digitization, including everything from middleware to how to select the right solution providers. If the CEO does not have the insights and knowledge to know when the CRO or CTO are not up to speed with current trends, or just plainly out of their depth, then progress on digitization could be held hostage by one or both.
When it comes to digitization projects, time is also not on your side. If they take too long to implement, they can become distracting and expensive. At a certain point, practical considerations set in, and a project can get killed. To keep momentum going, you have to move fast, fail fast, iterate, and prove concepts in a short amount of time. Digitization cost can be a significant vulnerability. One participant noted that the more expensive the digitization project, the higher the chances of failure. “If you are making a big bet and putting $2,000,000 on the table to do something, the pressure from your Board and investors to get it right is sort of paralyzing,” he observed. This creates a negative feedback loop. Rather than feeling agile and iterative, you feel the need to overanalyze to get things right, causing implementation to take longer than expected. The irony is, the software needed to automate a loan decision (for example) can be quite inexpensive. Not all digitization requires large, sophisticated, time-intensive (and therefore expensive) amounts of infrastructure. The solution can be simple and inexpensive, so knowing what solutions are available and appropriate for the value you wish to create is crucial.
Which leads to the next important lesson learned from digitization: the importance of articulating and quantifying the value that you wish to create, both for customers and the business. Many MFIs can articulate the value of automating a process in terms of time saved, but then be unable to measure that value to demonstrate it has been achieved. “When there is clarity around defining and measuring value, it is amazing how much smarter everybody gets,” said one digitization expert. Through clear value definition and measurement, you can achieve a dynamic practice that automatically generates momentum toward discovering the product features that create value. If you cannot articulate value and translate that into measurable “bottom line” indicators, then it will be hard to convince your C-suite that the digitization effort is worthwhile.
Not seeing the forest from the trees was another common refrain with respect to product design and implementation. For example, MFIs that automated their loan approvals were understandably hyper-focused on risk because they had to make sure the algorithm was lending to creditworthy customers. But what about the customer experience? It turns out, automating a loan does not necessarily make it more attractive. Some MFIs even reported customers opting for traditional loans over the automated ones even if the latter offered them a larger loan size. For digitization to add value, the automated loan product has to be significantly better than the traditional one, not just for the customer but also the loan officer. The latter must be willing to manage a portfolio of loans that they did not originate, particularly if some go into default. More importantly, they must be willing to use the time saved to manage an even larger loan portfolio. This was a key lesson learned and one that was observed by several of the MFI networks in the room. Automation can save a loan officer time, leading to increased job satisfaction, but it won’t necessarily translate into larger loan portfolios unless monetary incentives to boost productivity are introduced. This was generally seen as a “mindset” challenge. Management was so focused on risk (not losing money) that they failed to make the automated product more attractive to customers or put in place the financial incentives needed to encourage loan officers to manage larger portfolios.
Most astonishing to me, was that none of the automation efforts led to lower interest rates. Even if risk-based pricing (i.e., charging lower rates to lower-risk customers) is difficult to do in practice, at least MFIs could have tried cost-based pricing (charging lower interest rates on the lower-cost automated loan). If you are struggling to make an automated loan more attractive, one sure way to do so would be to offer a lower interest rate on it. I asked the participants this question, and the main response was that because automation is so expensive, there are no cost savings to pass on. But this answer flies in the face of what was mentioned earlier, i.e., that technology for loan automation does not have to be sophisticated and expensive.
Bottom line: it seems that the promise of technology to transform microfinance has not been realized because digitization is challenging for the reasons mentioned above. The workshop summarized the lessons learned from automating operations into 5 “core principles” for successful digitization:
1. Resource and empower the product development team. The product team has to have executive power and have data analysts working for them to understand customer behavior in order to understand the problem for which you are solving. Digitization efforts should be led by the product team, not the IT team.
2. Define and measure the value you aspire to create. It is not enough to create efficiency gains if you cannot measure those gains and translate them into higher productivity and customer satisfaction.
3. Prioritize the product features that create value. Test fast, fail fast, redesign and iterate so that you can focus on the technological features that create and achieve the value that you defined at the very beginning (the problem you are solving for).
4. Test the prototypes with simple technology. The technological solution does not have to be sophisticated and expensive. Having the right staff, with the right skills and knowledge will help with vendor selection. The solution provider also has to be nimble and agile, so selecting the right vendor is key. In the words of one participant, “pick a technology that you can manage and that won’t drown you.”
5. Create a good user experience for everyone. There will be less resistance to change if the final product is high quality and adds value, not just for the customers, but also for internal staff.
Ultimately, creating a product management culture can be paramount for success. If the product teams have data analysts identifying customer behaviors, they can use this “business intelligence” and technology to identify and solve for problems like poor loan renewal rates, or low uptake of new products. This type of product management culture engenders trust, where management believes the product team knows what they are doing, not just in terms of product design, but also a risk management and compliance. As one participant summed up, digitization is, “hard, but if you take a structured approach, and break down the process into problems you can solve for, you can find and implement appropriate solutions.”
Easier said than done, for sure, but at least it seems there may be a path to reach microfinance’s Holy Grail. It just might take a bit longer than everyone hoped and expected.
Thanks for this update, Anthony. You’ve captured an enormous amount of experience very succinctly, and your main points ring true. It’s especially disappointing to know that interest rates have not dropped. One point you didn’t address was whether NPLs differed for digital loans, which would be a cost factor.
Great write-up and very informative! While digitization seems to be a straightforward goal, it invokes organizational change that can have unforeseen ripple effects and require MFI realignment across numerous areas of activity. MFIs are optimized towards a certain way of business and simply digitizing some isolated business activities can raise new hurdles (i.e. your example of time-saving efficiencies vs. incentive attainment). This also begs the question as to whether MFIs are even equipped to lead in digital microfinance or whether newer fintech companies can provide a better template (it seems they face a different set of challenges).
A wonderful job Anthony with this piece. Just adding a few views and also elevating others.
Most probably the board than management, needs to understand why, how and what to digitize? If the board has no clue, then perhaps start by enhancing the board or continue with your BAS. Management can come along through workshops and constant coaching because they have time, but the board does not have that time. The shareholders should be willing to devote resources to the process.
10 years ago, fintechs seemed to find a solution to and scaling and cutting costs. They predicted that MFIs would go out of business within 2-5 years, but MFIs still exist. Why? Although fintechs have brought great solutions to customer’s problems, their algorithms alone are falling short in predicting risk due to limited data, hence high NPLs and elevated interest rates. The MFIs on other side are sitting on a lot of idle data accumulated over the years. Perhaps the solution is convergence, where the fintechs work with the MFIs to solve problems together. A couple of FIs have teamed with fintechs with revenue share arrangements and managed to scale significantly.
Also, there are some aspects which are hard to digitize. There is still relevance in human relationship unless the market becomes all GenZ and Millennials overnight. Perhaps a certain mix of tech and touch with more touch in the initial stages when establishing relationships and then deploy tech for repeat cycles after enough data has been accumulated.
And remember to build as you go, have a sun box, for innovation and test, then implement or discontinue if it does not work.
Thank you for sharing. This is a great summary and a lot of what it has sounds about right. At Kaleidofin (www.kaleidofin.com), our tech stack helps a lot in digital transformation of Microfinance institutions. And at least in India, I see some things changing. Inertia though is high and more than cost, I feel it is the time to implementation and subsequent non familiarity of the team that becomes a challenge. Most entities do not have a strong team and there is a significant reliance on external vendors.
Also in most cases, a product manager who thinks about the customer doesn’t exist making tech projects fail.
A long time ago, I had a chance to witness a pilot conducted by a known payments entity in partnership with a leading MFI in India to digitise transactions. In that age, mobile connectivity was not great and the solutions needed to work even in the absence of good connectivity.
So the solution offered was that each borrower will be issued a chip based smart card that can store offline data. The field officer will carry a POS device that was a variant of the credit card type machine but longer, heavier, bigger screen, more possibilities of input.
The process change was enormous though. In the past, the field officer had to sign in a pre printed card 20 times if it was a 20 people group. Everything else, including the date was printed. Count the cash and issue a single receipt to each group.
In the new process, each card had to be inserted manually into the machine 20 times over (once for each customer). Several buttons pressed to authenticate, amount added for each transaction and save data which in those days wasn’t instantaneous.
A thermal receipt was printed for each customer. It used more expensive paper 20 times over and a paper that would fade in a month max. The customers would then come to the branch to ask for some printed record.
In many cases the field officer was pushed to fill up the old loan cards and sign it by the customers as he did in the past.
The challenge was that no one thought of the customer and the field officer. Smart cards and pos devices were available and that became the basis. In many cases even to date, I see several projects of digital transformation that would do much better if a good products person would be involved in designing such projects.