Building finance, sectoral bias, and housing stock accumulation in Kenya, a nonlinear ARDL analysis

This paper examines whether Kenya’s building finance is subject to sectoral bias-flowing disproportionately toward non-residential rather than residential building stock, and what this implies for the long-run relationship between building finance and building stock accumulation. Despite an increase in credit proportions going to the building sector between 2010 and 2023, the country’s housing deficit has continued to widen, raising the question of whether this finance is actually reaching residential housing supply or is instead absorbed by the commercial real estate sector. We use annual data from 1970 to 2024 to estimate a non-linear autoregressive distributed lag framework that allows for asymmetric short and long run effects of building finance while distinguishing between residential housing stock and non-residential stock. The results show a limited transmission of building finance to physical housing stock. A sectoral asymmetry is observed, where building finance is associated with short-run adjustments in non-residential stock but no significant effect on residential housing stock. Residential stock accumulation is instead driven primarily by income growth and construction costs. The findings suggest that building finance deepening in Kenya has had limited sustained effect on residential housing stock expansion, raising important questions about the structure of credit allocation within the housing sector. Kenya has deepened its housing finance system considerably over the past decade. Outstanding mortgage credit grew from under KSh 100 billion in 2010 to about KSh 262 billion in 2023, yet the formal housing supply has remained at roughly 50,000 units a year against an estimated annual demand of about 250,000. This study asks why financial deepening has not translated into more homes. Using annual data for 1970–2024 and a nonlinear autoregressive distributed lag (NARDL) framework, it examines how positive and negative changes in housing finance affect residential and non-residential housing stock separately, in both the short and the long run.The results show that housing finance has no statistically significant long-run effect on residential housing stock. Residential accumulation is instead anchored to income growth, with a long-run elasticity of about 3.2, and to construction costs. Non-residential stock behaves differently, positive credit shocks are associated with short-run expansion in commercial construction, but the sector is not cointegrated with income and cost fundamentals. These findings point to a sectoral bias in which credit reaches the property sector but flows disproportionately towards commercial real estate rather than towards homes.The work is significant in three ways. First, by disaggregating housing stock by sector, it reveals a pattern that aggregate analyses conceal, the weak link between finance and housing supply in Kenya reflects the allocation of credit, not simply its scarcity. Second, it challenges the premise underlying credit-led housing policy, including the Kenya Mortgage Refinance Company, that expanding mortgage availability will close the housing deficit. Because most Kenyan homes are built incrementally on informally held land outside the formal credit system, the study argues that policy should focus on connecting formal finance to incremental building, for example through SACCO-linked construction finance, staged plot-development mortgage products, and the channeling of Housing Levy funds towards these models, and on differentiating residential from commercial credit allocation. Third, the study offers a transferable diagnostic for other Sub-Saharan African economies, where similar finance–housing disconnects have been documented without a sectoral distinction.The findings are relevant to policymakers, central banks, development finance institutions and housing researchers seeking to design finance interventions that translate into real additions to the residential housing stock.