AI found a barrier to female funding in Kenya (the answer was boring, and why that’s exciting)

What's been holding back access to entrepreneurial financing for women-led small and medium-sized businesses in Kenya?

Ask development experts and they'll give you a list. Gender bias in lending. Lack of collateral. Financial literacy gaps. Cultural barriers. All true. All documented. All real barriers that deserve attention and resources.

Nobody mentions ID cards.

That is, unless you're someone like Jane Muthoni. In July 2024, she was preparing to join Kenyatta University. She'd done everything right. Passed her exams. Got accepted. Applied for a loan from the Higher Education Loans Board to cover tuition. The HELB loan was designed for exactly this situation. Thousands of students relied on it every year.

There was just one problem. Jane didn't have her national ID card yet. She'd applied for it months earlier, but she was still waiting.

Without the ID, she couldn't complete her loan application. HELB requires a copy of the applicant's national ID card.[1] Kenya's anti-money laundering regulations mandate Know Your Customer verification for any financial transaction, and the national ID is the primary credential.[2] No exceptions. No workarounds.

"I've been waiting for my ID for months, and now I'm worried I won't be able to apply for the HELB loan in time," Jane told reporters.[3]

She wasn't alone. Thousands of students prepping for university that month faced the same problem.[3] So did entrepreneurs trying to get business loans. People trying to open bank accounts. Anyone trying to access a formal financial service.

The National Registration Bureau had stopped issuing ID cards. Not officially. Not explicitly. They'd just stopped. With that, access to the entire financial system froze for hundreds of thousands of people.

Nobody in the development economics world seemed to notice. Why would they? It wasn't self-evident that Kenyan ID card printing had much to do with women's entrepreneurship financing. It may seem obvious now that I’m pointing it out, but without the benefit of hindsight, ID cards and women’s financing seemed like unconnected variables that were worlds apart.

The Puzzle Nobody Was Solving

The Kenyan government had been running a fund specifically designed to get money to women entrepreneurs. Since its establishment in 2007, the Women Enterprise Fund had disbursed loans to over a million women.[4] International development institutions had poured funding into programs with the same goal.[5]

Despite all this effort, only seven percent of women-owned small and medium enterprises in Kenya could access formal finance.[5]

Seven percent.

For decades, development economists have been working on questions like this with dedication and sophistication. When programs failed, the reasons were often elusive. Even when experts had theories about what went wrong, proving causation in complex systems remained difficult.

The Kenya story is a perfect example.

Enter the Nerds

This year, a team of data scientists tried something different. They mapped (through industrial-scale data harvesting, AI, machine learning, etc.) all the value chain pieces that had to come together for a woman entrepreneur in Kenya to get a business loan. Not just the application process. Every piece of the ecosystem. Every dependency. The policies, regulations, financial infrastructure, risk models, government offices, technology systems. The many invisible constituencies and processes that had to align.

This is what makes an algorithmic approach unique. Traditional development work picks a hypothesis (gender bias, collateral requirements, financial literacy), then designs studies and interventions around it. The data science approach didn't start with a hypothesis. It started with measurement. Map everything. Measure throughput at every node. Find where things actually stop.

They were looking for bottlenecks. Chokepoints. The slowest parts that held back the system as a whole. They were as surprised as anyone else when ID cards topped the list.

How Printers Broke the System

Stay with me here, because this gets interesting.

In November 2023, the Kenyan government started rolling out a new digital ID system. The Maisha Card. It’s chip-enabled, has enhanced security and is very modern.[6] They were phasing out the old prior-generation cards and routing new applicants to the new system.

Then civil society groups sued, expressing legitimate concerns. The government hadn't done required data protection assessments. So, on December 5, 2023, the High Court agreed and issued conservatory orders.[7]

Here's the thing: the court order didn't technically ban all ID card issuance. It specifically halted the Maisha rollout. In theory, the government could have kept printing the old cards.

In practice, they didn't.

Kenya had already pivoted to the new system. New applicants were being routed to Maisha. When the court suspended Maisha, ID production effectively stopped.

For eleven weeks.[8]

Think about what this means. To access any formal financial service in Kenya, you need a national ID card. No ID means no bank account. No loan. No access to the financial system. There's no workaround. Despite everything that had been done by the Kenyan government and development agencies to create access to entrepreneurial financing for female-led small and medium-sized businesses, none of it mattered without an ID card (which had been an afterthought).

In July 2024, it happened again. Another lawsuit. Another injunction. Another suspension of the Maisha system.[9] This second freeze lasted about two and a half weeks before the court lifted it.[10]

That's when Jane Muthoni found herself stuck. In fact, an estimated 600,000 Kenyans like Jane got caught up in the ID freeze. [11]

Jane eventually got her ID, but unfortunately the problems didn't stop when the government resumed printing ID cards. It’s hard to stop and start a process like that without a domino effect of consequences. First, the government struggled to make enough cards to keep up with those who had been waiting, plus the regular cadence of people who typically need new cards or replacements going forward. Making things worse, by August 2024, more than 476,000 printed cards sat uncollected in government offices.[12] A mountain of cards had been printed, but the government struggled to deliver them to their owners (especially in rural and hard to reach areas).

Here's what makes this uncomfortable: each actor followed their incentives perfectly. The government wanted to modernize Kenya's ID system (reasonable). Civil society groups identified legitimate data protection concerns (important safeguards). Courts paused the rollout to ensure proper procedures (correct legal response). The government, having pivoted to the new system, didn't scale back up production of old cards (administrative inertia).

Nobody made a bad decision. The system failed anyway.

Why AI Changes Everything, In a Good Way

Consider what just happened. A team of data scientists gathered data and modeled the highly complex value chain for female entrepreneurial financing in Kenya. The algorithms not only uncovered the ID card crisis but linked it to female-led financing as a critical bottleneck. The team didn't figure this out; the AI did. That represents an extraordinary level of contextual awareness for a machine. It found and connected multiple needles in a very big haystack.

Capabilities like this give governments and development organizations a remarkable advantage. They’re trying to make progress on problems so entangled that no one ever knows what might blindside, constrain or quietly derail their work. Beyond female lending in Kenya, the same process is proving both effective and surprisingly delightful across a range of social and environmental challenges. Each time, the data brought familiar issues to the surface alongside unexpected and often counterintuitive bottlenecks that had been hiding in plain sight. Patterns appeared that no one had thought to look for until the AI revealed them.

For example, it connected the use of microbes in turning plants into hormone ingredients as a key barrier to women's reproductive health, spanning contraceptives, hormone therapy and treatment for polycystic ovary syndrome. It linked a current shortage of fingerprint scanners with the prevention of domestic child abuse. In each case, the constraint was far from obvious until after it had been pointed out, because there are simply so many variables and interdependencies to consider in any complex social issue that it's hard to know which to focus on.

That’s the big picture, and why it's exciting. We can now grasp both the sweeping complexity of social systems and the fine-grained details where those systems break down, without having to compromise one for the other. Development economics has always drawn brilliant people to hard problems. Traditional approaches have produced real successes, but also baffling failures born from the sheer difficulty of seeing where the true constraints lie.

It's detective work on an unprecedented scale. The computer doesn’t care about theories or assumptions. It just follows the data and finds the places where a small, precise action can unlock progress across an entire system.

The world is full of problems that seem intractable. How many are actually unsolvable? How many simply hinge on something mundane, like an ID system no one thought to question?

Remember that seven percent. Only seven percent of women-owned small and medium enterprises in Kenya had accessed formal finance. That’s not just a statistic. That’s people like Jane Muthoni, waiting for an ID card. Thousands of entrepreneurs did everything right, only to be stopped by a bottleneck almost no one had connected.

If a binding constraint on women’s entrepreneurship in Kenya turned out to be something as ordinary as ID cards, what else are we missing? What other massive problems hide small, fixable causes buried in all that complexity?

The data scientists mapped the bottleneck. Now it’s time to find the next one and remove it for Jane, so that nothing gets in her way.

Endnotes

[1] Mount Kenya University, "How to apply for HELB 2023/2024 Scholarships and Loans through the HEF Portal," August 28, 2023. Multiple university sources confirm HELB requires "A copy of the Applicant's national ID Card" as mandatory documentation.

[2] Republic of Kenya, Proceeds of Crime and Anti-Money Laundering Act, 2009 (POCAMLA). The Act mandates customer identification using reliable source documents for all financial transactions. See also: Financial Reporting Centre, "Compliance Requirements," available at: https://www.frc.go.ke/?page_id=25

[3] Eastleigh Voice, "Thousands of students at risk of missing out on HELB loans after halt on ID card printing," July 31, 2024. Available at: https://eastleighvoice.co.ke/maisha-namba/63577/

[4] IDinsight, "Women Enterprise Fund Learning Partnership," March 15, 2024. "As of 2022, the fund has disbursed loans to over a million women in Kenya." Available at: https://www.idinsight.org/project/women-enterprise-fund-learning-partnership/

[5] International Finance Corporation, "Bridging the Finance Gap for Women Entrepreneurs in Kenya," 2024. "With only seven percent of women-owned micro, small and medium enterprises (MSMEs) in Kenya estimated to have formal access to finance, these businesses often struggle to grow or survive." Available at: https://www.ifc.org/en/stories/2024/bridging-the-finance-gap-for-women-entrepreneurs-in-kenya

[6] Keesing Platform, "Kenya Kicks Off Digital Identity Pilot Program," February 27, 2024. Available at: https://platform.keesingtechnologies.com/kenya-kicks-off-digital-identity-pilot-program/

[7] Multiple sources confirm the December 5, 2023 court order. See: ITWeb Africa, "Kenya's High Court lifts injunction on new digital IDs issuance," February 26, 2024; Capital News, "High Court Lifts Injunction on Maisha Namba Printing," February 23, 2024.

[8] The Star, "Maisha Card issuance to resume immediately – PS Bitok," February 23, 2024. The injunction was issued December 5, 2023 and lifted February 23, 2024 (approximately 11 weeks). Available at: https://www.the-star.co.ke/news/2024-02-23-maisha-card-issuance-to-resume-immediately-ps-bitok

[9] The Star, "High Court suspends rollout of new digital IDs," July 25, 2024. Available at: https://www.the-star.co.ke/news/africa/2024-07-25-high-court-suspends-rollout-of-new-digital-ids

[10] Kenyans.co.ke, "High Court Approves Maisha Card Rollout in Latest Ruling," August 12, 2024. The July suspension lasted from July 25 to August 12, 2024 (approximately 2.5 weeks). Available at: https://www.kenyans.co.ke/news/103549-high-court-approves-maisha-card-rollout-latest-ruling

[11] The Star, "Maisha Card issuance to resume immediately – PS Bitok," February 23, 2024. Immigration PS Julius Bitok stated the decision "brings relief to more than 600,000 Kenyans whose applications for new and duplicate ID cards were put on hold."

[12] The Star, "Over 470,000 ID cards remain uncollected countrywide – PS," August 27, 2024. "By close of business on August 26, 2024, 476,167 printed National ID cards were lying uncollected in the NRB registration offices and Huduma Centres across the country." Available at: https://www.the-star.co.ke/news/2024-08-27-over-470000-id-cards-remain-uncollected-countrywide-ps