On June 10, 2025, the Monetary Policy Committee (MPC), through its chairman, Central Bank Governor Kamau Thugge, announced that the gross non-performing loans (NPLs) ratio in Kenya’s banking sector had climbed to 17.6 percent in April, up from 17.2 percent in February.
This upward trend in loan defaults, particularly in real estate, personal and household, trade, construction, and manufacturing sectors, has raised alarms across the financial ecosystem.
Despite this deterioration, the Central Bank of Kenya emphasised that the sector remains sound, supported by liquidity and capital buffers, and banks continued to make adequate provisions. In the same meeting, the MPC also reduced the Central Bank Rate by 75 basis points to 10.00 percent to boost private sector lending amid easing inflation and a stable exchange rate environment.
As someone deeply involved in the credit information space, these developments trouble me. We have robust data systems and legal frameworks in place—yet defaults are rising.
This paradox led me to undertake a research project last year to explore a narrower but revealing aspect of the problem: whether significant differences in credit risk exist across Kenya’s eight former provinces, and how these differences influence Loss Given Default (LGD) among micro, small to medium enterprises.
The findings were compelling. Credit scores and PD values differed significantly by region and enterprise size. This indicates that two businesses with similar profiles may have very different risk levels depending on where they operate. In contrast LGD showed variation across groups but was not statistically significant when controlling for business size, suggesting that loss severity is more closely linked to the scale of the business than to geography.
These findings have clear implications. Lenders must move beyond generic risk assessments and develop customer-centric, data-driven products tailored to regional and sectoral nuances. Regulators and accountants can also use this insight to fine-tune Expected Credit Loss models under IFRS 9, ensuring more accurate provisioning. For policymakers, the study raises an important question: could some well-meaning policies, such as blanket debt relief, unintentionally encourage strategic default?
Metropol unveiled a new tool within its analytics platform, Metropol Analytics Platform (MAP), to help banks and other lenders manage their loans more easily. It has been designed to make life simpler for everyone involved in lending, from bank managers to loan officers.
One big problem banks have been facing is figuring out which borrowers might not pay back their loans. The old way of doing this often missed early signs of trouble. Metropol uses up-to-date information to sort borrowers into groups from low risk, medium risk, or high risk, with the new tool to help banks step in early if something looks off.
But if Kenya is to support financial inclusion and build a more stable credit market, we must move beyond compliance checklists and ask harder questions about borrower behaviour, motivation and context. We must try and deal with why banks are increasingly favoring clients with strong tract records, collateral, and credit scores, instead of leveraging data and analytics in their decision making.
Similarly, at the service level, why should they continue to tax customers with bureaucratic lending procedures and extensive risk assessments, when they are convenient and more predictive approaches to risk management? I hope that data and analytics, coupled with objective scientific research can be embraced more in policy formulation by both government and private sector towards a more resilient and responsible financial ecosystem.
