Land Banking

One especially pernicious housing meme which never seems to die is that developers are land banking. This conversation is a little difficult to wade through, because there are two competing definitions of ‘land banking’ which people slide between:

1. Developers purchase land, get permission to build on it, and sit on it, deliberately delaying or refusing to build.

2. Developers purchase land, get permission to build on it, and place it in a pipeline that they then build out at a later date.

Critics often conflate the two. (1) implies anti-competitive behaviour, price gouging, and greed; (2) is what we expect in a market subject to the constraints and realities of building homes.

Land banking is therefore often seen as a significant hindrance to solving the country’s housing crisis. The planning system, goes the argument, isn’t the problem - developers are just refusing to build the homes that the planning system offers to them.

The argument is wrong.

How many unbuilt permissions are there?

Attempts have been made to size the problem. You often hear numbers as large as 1.4 million - which, were it true, would put a good dent in the UK’s housing shortage (but by no means fix the problem.)

These numbers cannot be trusted. We don’t really know how many unused permissions there are, because we don’t have good housing data, and it’s very difficult to measure.

Some permissions need to be requested again because the scheme has changed, which often renders the original permission invalid. Schemes can change during development for many different reasons. Things get discovered as you work which can constrain where you put homes or the sorts of homes you can build. Infrastructure requirements change. Market conditions shift, so four-bed houses become less wanted in an area and the developer needs to build more flats instead. These end up with new permissions granted, which increases the number of total permissions without increasing the number of actual homes that those permissions represent.

We don’t know easily whether two applications relate to the same site – location plans are submitted as PDFs, which make comparisons difficult – so there are cases where ‘Land to the east of Longstanton’ becomes ‘Northstowe’ and the figures are counted twice. Developments which have been already permitted might be held back by sudden changes to national guidance, as happened with the introduction of nutrient neutrality rules in 2019.

Schemes might be held while critical infrastructure is built, or while pre-commencement conditions on the original planning application are completed. Or, in the case of affordable housing schemes, there simply isn’t a buyer yet: in February 2025 the HBF claimed that 31% of the 2024/5 affordable housing supply is stalled because the housing associations are not purchasing them.

There’s also the fact that larger schemes just take more time to build, so the permitted-but-not-started numbers can also be inflated by schemes that are in progress.

The people who compile these figures try to make adjustments for all these caveats, but it’s simply too difficult to do with the current state of our data collection apparatus – to the extent that the government itself says that these numbers should not be used to estimate the number of unbuilt dwellings with permission. We can’t trust the numbers; we simply don’t know.

Is this behaviour anti-competitive and greedy?

The fact that we can’t trust the numbers reported doesn’t mean that land hoarding isn’t a problem. In order to address that issue, we need to assess the logic of the claim. Under what circumstances would it be rational to hoard land, get planning permission to achieve an asset gain, and withhold development?

If a developer believes that selling or developing later will yield significantly higher profits than building now, then it would be rational behaviour to do so. More precisely, if the profits from doing so minus the costs of holding the land for that period of time are significantly higher than building now, then it would be rational to do so.

Fortunately, those costs are often much higher than the expected return to holding onto the land.

The appreciation of greenfield land with planning permission varies widely depending on location and existing use, but government estimates suggest an average value increase from £21,000 per hectare to £1.95 million. But this is the increase in land from agricultural use to land with planning permission for residential development: it represents value that could be realised by the landowner today. (This, incidentally, is the business model of land promoters: they partner with the landowner, achieve planning permission, then sell it on to a builder.)

So in order for them to not realise that value increase as soon as they can, they would need to believe that the value of the land would continue to increase, taking into account the costs of holding it, more than developing it.

As for how land values fluctuate over time, this is much more unpredictable:

Source: Knight Frank, UK Residential Land Development Index (2025)

There is always a risk that by holding the land and not realising its value immediately, the value of the asset could decrease – sometimes significantly.

And, of course, there are costs to holding the land. Land is generally not bought with cash that already sits on a company’s balance sheet – the money is borrowed, and therefore has a direct cost of capital. (Even cash sitting on a balance sheet has a cost of capital, of course, which is the inflation rate.) The average development loan costs 5.2%, and can range from 4-16%; it is also driven by broader interest rates, so when the rates increase, so does the cost of future loans. The chart above shows that in only nine quarters since 2016 did the value of greenfield land increase at all – and in only four of those quarters did that increase exceed 5%. For every other quarter, it would have been more expensive to hold onto the land than to sell it.

And that’s just the cost of capital. Homebuilders have other costs too. They have wage bills they need to pay, supply chains they need to maintain, the costs to gain planning permissions for future projects, insurance, site security, sales and marketing costs, contributions to government. Even when some of these costs get assigned to specific projects, there’s a cashflow issue - they need to keep paying them.

In short, it is rarely rational for a homebuilder to hoard land and not build houses. Hoarding land is a costly endeavour. They act greedier when they build homes, not when they don’t.

Is the planning system to blame?

It takes, on average, 2.8 years for schemes between 100 and 500 houses to receive planning permission. Just to achieve outline permission – which establishes the principle of development, something which is generally not needed in a zoned system – it takes a year.

The planning system is supposed to be used to ration land for development, but in practice it does much more. We expect it to protect the environment, preserve our heritage, reduce noise, increase biodiversity, promote light, promote cycling, promote driving, and even fight crime.

This means that the average local authority requests 30 different document types for an outline permission, and the total variety can stretch into the hundreds:

Source: Lichfields, ‘Small builders, big burdens’ (2023)

In some cases these reports are necessary, and lead to a better decision and scheme. But in many others, developers are encouraged to do the maximum work possible in order to ‘copper plate’ the application and reduce the likelihood of a challenge. This builds extra cost and time into the process where it is not needed.

And this cost and time needs to be accounted for in a housebuilder’s plans. In order to run maximally efficiently, they need to be working on sites continuously, which means they need to maintain a pipeline of sites at various stages of the planning process.

This pipeline – a direct consequence of the uncertainties inherent in construction and the UK planning process – is a land bank.

This effect has been studied many times, by both industry bodies and in independent reviews. The last major investigation was by the Competition and Markets Authority last year, who say that:

observed levels of land banking activity represent a rational approach to maintaining a sufficient stream of developable land to meet housing need, given the time and uncertainty involved in negotiating the planning system

These findings were also backed up by 2018’s Letwin Review, which:

found no evidence that speculative land-banking is part of the business model for major house builders, nor that this is a driver of slow build out rates

A review in 2008 by the Office of Fair Trading suggests:

that homebuilders are not delaying building on permissioned land to an extent that would appreciably affect the rate of delivery of new homes

And that:

[h]omebuilders' landbanks appear to be largely explicable in terms of the amount of time it takes to promote land through the planning system, and the risks involved in doing so.

And in 2007, the Callcut Review opined:

'There are no doubt some individual cases where house builders hold land for longer than they need. But, in our view, the current evidence does not support the suggestion that this practice is at all widespread.'

There are several other such reports, reviews, and studies, each one of which reaches the same conclusion.

Conclusion

We can’t trust the numbers, because our data quality is poor and there are lots of reasons why a site might have permission without it being built on. It would be irrational for developers to purchase land and sit on it, because their costs of capital and operation would almost always outweigh the expected increase in the value of the land while hoarded. And there is an incredibly strong, consensus case that developers’ pipelines are a benign and rational response to the incentives created by the planning system.

If we want quicker build-out rates, and for developers to sit on less land in aggregate, we need a planning system that brings certainty, predicability, and speed to the process of assigning permissions to build. Blaming it on greed is both inaccurate, and cheap.

Thanks to Paul Smith, whose work and conversation I’ve drawn on extensively, and Ben Dunn Flores for notes on this piece.