糖心TV

• A three-year study of 19 million planning applications is changing our understanding of England鈥檚 housing pipeline.
• The new dataset reveals fewer distinct proposed homes than government figures suggest, alongside a much stronger upturn in 2025.
• The increase follows a decade of decline from a peak in 2014 and comes after changes to England鈥檚 national planning rules in December 2024.
• While the findings do not establish that the reforms caused the rebound, they suggest a stronger upturn in proposals than previously reported. 

Researchers at the University of 糖心TV today release the first findings from Housing Pipeline, a new dataset that connects applications, revisions and phases belonging to the same development.  

The project is needed because official planning figures are not designed to track the number of homes entering the planning system over time, and this has substantial implications for how we understand future supply. 

By following individual projects across years, Dr Nikhil Datta and Dr Amrita Kulka鈥檚 dataset can distinguish new housing proposals from repeated applications for the same homes. 

The first findings, released today in a briefing note, show that government figures for homes proposed through outline applications were, on average, 43% higher than the researchers鈥� estimates over 2022–2025. 

The new data also reveals a sharper recent recovery: proposed homes rose by 107% in 2025, compared with 59% in the government series. 

The briefing note published today is the first in a series of studies using the new dataset. Dr Datta and Dr Kulka say their next steps are to examine where delays occur, how developments change as they move through the planning system, and how often planning permissions translate into completed homes.

Dr Kulka comments:

鈥淕iven the severity of the housing crisis there is a need to understand which policies move housing supply. Yet, counting how many homes are being applied for is difficult due to the complexity of planning records. We have spent over three years collecting new data and developing methods to accurately measure applied-for units which are the foundation of future homes."

Dr Datta adds:

鈥淧lanning records were created to administer applications, not for tracking housing statistics and turning them into something reliable requires substantial work. We spent three years combining state-of-the-art machine learning and AI tools with thousands of hours of human checking to reconstruct and track individual developments from start to finish.鈥�

If we cannot reliably count the distinct homes entering the planning system, it becomes much harder to judge whether reforms are working or understand where housing delivery is falling short. Reliable measurement needs to be the starting point.鈥�

•  

  Edoardo Badii, Nikhil Datta and Amrita Kulka. CAGE Policy Brief, October 2026,