Planning matters

There are few topics that can command the same media attention as housing. One might think that the level of interest, scrutiny and desire to overcome the challenge would mean new reports and analysis would reflect how far the conversation should have come, with ‘nuance’ as the watchword. But, alas, it appears not.

One story that does the rounds every year or two is what the Local Government Association calls ‘unimplemented permissions’. In its latest iteration, the LGA claims that more than 400,000 homes have been granted permission but have been left unbuilt and that they are taking longer to build those homes too. The implication to this is threefold: 1. planning is not the problem in housing delivery; 2. housebuilders are likely to be landbanking; and 3. the government may need to take a more interventionist role.

But in the name of nuance – and before we rush to designing policy solutions – we should understand a bit more about the analysis.

The data the LGA uses is from Glenigan, a firm that provides planning data on all development in the UK. Importantly, Glenigan mostly tracks schemes through the planning system and have a network of stakeholders and partners but what they do not do is visit every site to see precisely how many have been built. Back in 2016 – the last time this analysis was undertaken and released – it meant that units on sites that were not fully complete would be classed as ‘unimplemented’ i.e. a site of 1,000 units that was 99% complete would have all units counted as ‘unimplemented’.

This year, they have estimated the number of ‘unimplemented units’ on site by using the median construction time of projects with similar characteristics (project size, type and region) completed in 2015/16 and 2016/17, assumed that no units were completed during the first 16 weeks of a project and that unit completions were evenly distributed across the remainder of the construction phase. This is certainly an improvement to the methodology – which is welcome – but the analysis still doesn’t highlight key considerations which are vital if we are to create useful and effective policy.

Our extensive analysis from 2017 on this topic drew two key conclusions. First, even if we set aside the specific and detailed issues the LGA has not considered, it is still a relatively small number of units considering three factors: 1. the number of homes built each year; 2. the length of time the different sizes of sites take to build out; and 3. the number of sites in the pipeline required to signal to investors that developers are worth investing in.

Secondly, given that differently-sized sites take different timescales to plan and build, and that every site has unique ‘build profile’ that extends into the future - i.e. not all units on site are built in the year they are granted permission – it is entirely understandable that there would be ‘unimplemented units’ and, indeed, in a period when output is increasing after a slump the ratio between permissions and dwellings completed each year will naturally increase.

Recognising this build-out profile, our analysis uses two different housebuilding scenarios to show that there needs to be a stock of units with permission to sustain housebuilding in the future and this stock would initially increase at a faster rate than supply in order build the pipeline necessary to match increased targets. With this in mind, our modelling showed that a stock of between 0.9 and 1.1 million units would be required to ramp up future housebuilding to hit 300,000 homes per year by the mid-2020s (Figure 1) – a target subsequently adopted by the Chancellor.

This leads one to conclude the LGA figure of 423,000 unimplemented units is a sign not of landbanking but of an urgent need for more permissions.