Open Banking, AI, Data and the Three-Legged Problem Of Credit (Foreword by Timothy Li, CEO, LendAPI)

Foreword

As someone deeply immersed in the intersection of finance, technology, and data, I have witnessed firsthand the transformative potential of innovation in reshaping the credit landscape. At Lend API, we’ve made it our mission to empower lenders and borrowers by leveraging the power of AI and Open Banking to bridge gaps that once seemed insurmountable.This article eloquently captures a pressing issue I’ve often encountered in my journey—the three-legged problem of credit. It sheds light on the evolving challenges faced by traditional borrowers, emerging entrepreneurs, and an aging workforce, all of whom are navigating an economic terrain redefined by demographic shifts, technological advancements, and new modes of employment.

From small business owners grappling with cyclical incomes to aspiring start-up founders struggling with credit invisibility, the need for a data-driven, AI-enabled approach has never been more urgent. I believe the solutions proposed here—rooted in the principles of transparency, real-time decisioning, and inclusivity—are precisely what our industry needs to embrace.The convergence of Open Banking and AI is not just a theoretical construct; it’s a path to unlocking opportunities for millions who have been marginalized by outdated systems. By rethinking how we evaluate creditworthiness and expanding access to innovative lending products, we can empower individuals and businesses alike to thrive in this new economy. It is my hope that this article will spark meaningful conversations among lenders, policymakers, and technology providers. Together, we have the tools and the mandate to build a more equitable, sustainable financial ecosystem.

Timothy Li

Founder, Lend API

Introduction

We are going to face a three-legged lending problem in the business of credit and it needs to be addressed by data and AI practitioners in partnership with lenders. In the business of Artificial Intelligence, data is supreme. It is the ore which turns into the gold of rich insights, inferences and predictions by application of models. In the business of lending, too, data is core to everything. However, as the definition of work changes, demographics start shifting and AI becomes more important, we may need to revisit the kind of data we use today and how this may need to change soon.  In this context, an essay in the MIT Technology Review(link here-https://www.technologyreview.com/2021/06/17/1026519/racial-bias-noisy-data-credit-scores-mortgage-loans-fairness-machine-learning/) throws up this universal problem. The fairness of algorithms makes no difference if data is not available. Generally, this issue of data has been focused on who can be serviced by bank loans. This is going to get more complicated over the next few decades and inclusion will be challenged severely. I have described the three legs below, with the caveat that these refer to the geographical context of Asia. However, one might argue it's a more universal issue. 

The Three Legs Of The Credit Problem

Leg One-People who have been eligible for bank credit will struggle to make the mark and yet will continue to be highly productive. This is a problem that bankers did not have to face earlier. Salaried individuals retired at 55, 58 or 60. Small business owners had an account with a bank, took loans regularly and passed on the business to the next generation who continued the same relationship. But an increasing number of salaried individuals are losing their jobs much earlier and some of them are becoming sole proprietors or director-owners of limited liability companies. While many may continue to retain the privileges of an earlier life, such as a credit card, they face difficulties in acquiring such privileges in their new lives. A home mortgage or a new credit card is likely to beyond reach now. The cyclical nature of entrepreneurial income may them the convenience of regular, predictable and documented income every month, for a long time.

Leg Two-Alongside them, is a new segment of entrepreneurs. Unlike the first category, many are young, some being still in university. A small percentage will receive venture investment over a number of rounds and migrate into the ranks of successful shareholders of well-capitalized firms. Some will not succeed and if they fail to go back to the ranks of the employed, they will start new companies. Lo and behold, the serial entrepreneur. It is not clear as to how much of a credit relationship they have with banks because this is rarely discussed in public. But the number are not likely to be small.

Leg Three-Individuals are living and working longer than before. The retirement age is also getting extended. Since most people outside government service do not earn a guaranteed pension, they either need to fall back upon the savings in their pension or provident funds or continue to work. Some, of course, do so because they want to remain productive or sound and continuing financial incentives. Yet, once you are beyond a certain age (55 in some countries), access to loans and credit cards becomes problematic. This is magnified when you are the owner of a business. 

We have read often enough about bias in decision-making and there is extensive debate about its impact on lending. Bias is a term that is traditionally applied to the marginalized in society. But what if marginalization is a process that is gradual, complex and affects a far wider range of individuals than we are given to think? Is a well-off businessperson, a young start-up owner or a middle-aged salaried employee at risk of being sidelined in the lending decisioning process? Let us take a look at data from ASEAN to get a sense of potential lending marginalization in the years ahead.

1. By 2030, one in four Singaporeans will be 65 years old or more(Source-https://www.population.gov.sg/files/media-centre/publications/Population_in_Brief_2024.pdf).
2. For Malaysia, 7.7% of the population today is 65 years and above(Source-https://www.malaymail.com/news/malaysia/2024/07/31/malaysia-set-to-become-ageing-nation-by-2030-perak-leads-senior-surge-dosm-report-says/145577).
3. The percent in Malaysia which will be 60 years and above by 2040 is projected to be 17% of the overall population(Source-https://www.thestar.com.my/news/nation/2024/09/06/elderly-population-to-hit-17-by-2040-says-stats-dept).
4. In Thailand, 20% of the population is above the age of 60 (Source-https://www.nationthailand.com/thailand/general/40037217).
5. By 2031, 28% of the Thai population will be 60 and above(Source-https://www.bangkokpost.com/business/general/2087527/uncertainty-for-the-elderly).
6. In every country in ASEAN, the percentage of population aged 14 and below will decrease by 2030 and then again by 2050. This means that the population cohort graduating into adulthood and economic productivity will decrease in percentage terms(source-https://thediplomat.com/2024/03/what-southeast-asias-aging-populations-could-mean-politically/).
7. A large percentage of start-ups in this region, perhaps approximating to the overall trends worldwide, close down regularly or are otherwise not successful. (Source- https://asiaconnectmagazine.com/tracking-startup-closures-in-southeast-asia-understanding-the-trends/).
8. Countries such as Thailand and Singapore are increasing their retirement ages as more people live longer and the youth segment reduces as a percentage of the entire population(source-https://www.bangkokpost.com/thailand/general/2890358/retirement-age-to-go-up-to-65-ministry)
9. In Singapore, the retirement age has increased to 63 and will go up to 64(Source-https://www.mom.gov.sg/employment-practices/retirement).

It’s clear that everyone who is “elderly” cannot be slotted into a retirement bracket.  An entrepreneur who is over 70 and active, may need as much access to financing as someone half his/her age running a similar business. Similarly, those who are above 65 and still employed in some capacity, may want to be able to extend their spending capabilities to purchase various goods and services. Above all, if 20% and above of the population in a country is classified as elderly and the percentage of youth in the future is shrinking, a large segment has to be removed from active consideration by lenders. That would impact the lending books severely. If the birth rate of that market is really low (as we see for Singapore and Thailand, and further afield in Hong Kong, Macau and Japan), then the serviceable base of borrowers will stagnate and eventually shrink. This can become a bigger problem between now and 2040 since younger citizens may need to borrow on behalf of older relatives in increasing amounts, if the seniors cannot get direct access to credit.

Let us look at the changing profile of employment. What about young and middle-aged entrepreneurs who either run start-ups or are part of founding teams in mid to small sized companies? They will age over a 20-year window but more to the point, they often remain not well-qualified for loans even when they are relatively young. This number is likely to increase as the position of salaried workers inside companies becomes more tenuous and some of them are regularly driven to start their own companies, join nascent start-ups or get involved in freelance and short-term contract work. The supply chain of labour in a technology-driven society is based on costs and mark-ups being passed up to the prime contractor with terms and conditions favouring the party with leverage over the ultimate client. AI models can be expected to start to learn routine, repetitive and administrative tasks and also eat into more senior staff and non-technical/non-sales jobs. It is possible that posts and designations may be collapsed into areas of deliveries attended to by a smaller team assisted by machines. The heavy lifting in the supply chain for work has always been done by small-sized intermediaries, whether they use more intelligent machines or more people in their mix. This percentage may grow dramatically as corporates seek to use technology as a reason to carve out more yield. The individual worker and the small-scale company may start to become even more indistinguishable as compared to today. Yet, these smaller companies or freelance individuals cannot find easy ways to seek credit when they need it. The cost of capital available can be very high with terms and conditions difficult to achieve. The term SME, too, can be misleading with significantly larger firms with well-established cash flows falling within that category. The question then arises-what is the future for small business and entrepreneur borrowers from the perspective of lending institutions?

The Solution

The answer lies in being able to evaluate some segments that could be classically considered as marginal, to be in potentially good standing for adequate, timely and reasonable-term credit.  How can lenders decide this and contribute to greater inclusion, given their compulsions and obligations? For this, we need to talk about data. A question which may arise here is about the availability and quality of data for small businesses, especially small-scale retail. This is no longer as difficult as it might have previously been. This is because of the following reasons-a. consumer payments have moved to cashless in a big way across the board; b. a growing percentage of business payments are shifting from cheques to electronic bank transfers as well as real-time or next-working day payments. This means that there is digital data which can be extracted, optimized and transferred for processing. The availability of this kind of data can change long-held assumptions about risk profiles and credit worthiness. How would this kind of data be seamlessly available to lenders and their decisioning engines?

One hears usually about venture capital in the context of start-ups. But Credit cannot be underestimated. It offers choice. It should be easier to tap. Ease of borrowing helps small firms and entrepreneurs to cover costs of manufacture, ship orders within schedule and grow business capacity.  Some of these could be as follows- real-time loans with limited shelf life of offer; flexible tenors tied to varying risk levels; multi-tier offers to family run businesses allowing them to spread risk between members; hybrids between secured and unsecured finance; leveraging new IP and nascent cash flows. Above all, the power of AI can be used to predict business performance, especially cash-flows. A question which may arise here is about the availability and quality of data for small businesses, especially small-scale retail. This is no longer as difficult as it might have been. This is because of the following reasons-a. consumer payments have moved to cashless in a big way across the board; b. a growing percentage of business payments are shifting from cheques to electronic bank transfers as well as real-time or next-working day payments. This means that there is digital data which can be extracted, optimized and transferred for processing. The availability of this kind of data can change long-held assumptions about risk profiles and credit worthiness. How would this kind of data be seamlessly available to lenders and their decisioning engines? That may not be straightforward because there are regulations around risk as well as due diligence to be undertaken. So, the framework of data collection has to change.

An ideal way could be to have an open-data permissioning regime whereby a lender can seek consent from a borrower for data from his/her existing banking relationships. This is the way Open Banking works. But we propose going several steps forward.

After all, Open Banking itself will remain an attractive artifice but nothing more than that, unless it provides meaningful change or solves significant extant problems. This consent framework will need changes in legislation and regulations, some of which are being done already. However, it will help if ease of credit access is made a key outcome. While financial institutions may have relevant software available to provide key analytics, the real value will be obtained if the data processing outcomes are exposed in real-time to decisioning engines. This may need changes in the credit decisioning infrastructure itself. It is likely that mid to large sized financial institutions will call for the previously-mentioned intermediaries to play a more active role in providing real-time outputs and to manage their decisioning engines as these acquire greater complexity. If all the above can work, we are looking at the foundation of new categories of lending products.

Ease of borrowing helps small firms, nascent start-ups and entrepreneurs to cover costs of manufacture and procurement, meet operational costs, ship orders within schedule and grow business capacity.  By leveraging data, lending institutions may provide new products to balance the growth of their business with risk control. Some of these could be as follows- real-time loans with limited shelf life of offer; flexible tenors tied to varying risk levels; multi-tier offers to family run businesses allowing them to spread risk between members; hybrids between secured and unsecured finance; leveraging new IP and nascent cash flows. The flow of credit to the end-borrower and the repayment record can be utilized to provide scores in real-time to all concerned parties, alongside both incentives and penalties. Once again, Open Banking can provide the API-led and consent-driven layer of interaction between all parties.

Real-time decisioning and Open Banking will together engender changes in the technology stacks of lenders as well as fintech companies addressing this sector. These changes will help better quality of decisions in shorter timeframes to deliver much-needed support to new, high-potential segments. These segments need not be lost opportunities anymore.