Any interested stakeholders (policymakers, planners, developers etc) feel free to get in touch. We have also produced 350 LA specific reports accessible here: www.warwick.ac.uk/cage/whereto....

Looking at detailed evidence on planning restrictiveness, we find that Bexley, Lewisham, and Wandsworth combine high housing gaps with slow, restrictive planning systems. These areas urgently need reform. In contrast, Manchester, Leeds, and Birmingham show high demand and efficient planning.
12/n

We then turn to planning designations. 15% of the gap lies in Greenbelt areas – but it’s very concentrated. Allowing building on just 10% of England’s greenbelts with the highest gap would be a huge positive. This particularly pronounced around the North and Northwest. 11/n

To meet Britain’s housing gap, around 50% of new homes need to come from densification—especially in Wandsworth, Islington, Camden, Manchester, Bristol, Salford, Edinburgh & Portsmouth. Another 30% should be urban extensions like North Shields and South London (Richmond–Kingston).
10/n

Arguably the most important finding is that 96% of the variation is within LAs rather than between. Many LAs have areas of both high and low gaps. Croydon for example has OAs in the top 6th percentile and bottom 1st percentile. 9/n

The areas with the highest housing gaps are in trendy urban centres, all classed as densification. Those with the lowest gaps cluster near sewage plants, airports, major roads, or in places lacking basic amenities—where few people want to live. 8/n

We use 20 billion Rightmove searches and housing availability data to build our key measure of excess demand—the housing gap: the difference between how many people search for homes in an area and how many are available. We calculate this at the OA level – essentially a neighbourhood. 7/n

The proportion of densification has fallen over the past 2 decades, while urban extensions have grown along with small town extensions. We show that urban extensions are slower and more bureaucratic, with longer planning times and slower build-out rates compared to densification. 6/n

We then classify every new build by where it happened—combining settlement size, distance to the city centre, and local context:
New Rural Developments – small villages & rural areas
Small Town Extensions – around small towns
Urban Extensions – city outskirts
Densification – inner-city builds
5/n

70% of new builds have worse than average job access than the existing housing stock, and this figure is even worse when considering only public transport. 4/n

18% of all new builds are now in rural villages or small towns without a secondary school or GP up from 11% in 2010. 3/n

In the report we start by asking where has Britain built? Using data on 3.2 million builds over the past two decades we document that many new homes are built in low-demand areas with poor amenities and worse than average access to jobs. 2/n

Depending on the many small files context, you can set them up as a partitioned dataset which something like duckdb can treat as a single table.
Recently started using duckdb a lot out of necessity (20tb dataset), and can't praise it enough.

One of the big benefits of duckdb is that you don’t need to load the entire file onto ram. The smaller the file, the less the gains from using columnar storage / not loading onto ram. Small csv sizes read onto ram will be basically instantaneous.