2 June 2023

Filtering, or how building expensive homes can help people on low incomes


One of the striking features of British housing discourse is how often it overlooks quite basic economics. Some examples of this are so extreme that they become entertaining. I have often heard people explain that building private housing actually makes the housing crisis worse, because market-rate housing is unaffordable, and hence building it adds to the great mountain of unaffordable housing that we were worried about in the first place. Slightly alarmingly, the people saying this were not always eccentrics on social media: several times, they have been people in positions of responsibility at leading national institutions.

Readers of CapX do not need me to explain where this goes wrong. Instead, I am going to focus on a misunderstanding that is not silly at all.

Suppose the planning system can allow a splendid house to be built in an area of very high housing demand, with a sale price of £5m, or a small house in an area of low housing demand, with a sale price of £120,000. One might, very reasonably and intuitively, think that both of these things would be good, but the low-value house is better: the high-value house will be bought by an extremely high-income family who could easily have found other accommodation, whereas the low-value house will be bought by a low-income family who might have had no decent alternative. Ideally, we would permit both, but if the politics of housing targets means we must choose, then of course we should unlock the home for the people who really need it.

As I say, this is a completely understandable chain of reasoning. It is also virtually ubiquitous: my impression is that the overwhelming majority of housing policy discourse in the UK presupposes it implicitly. But some very basic economics shows that it is seriously incomplete, and often basically wrong. It is worth rehearsing why.


There are about 28 million ‘households’ in the UK, and about the same number of homes. This is not a coincidence, because a ‘household’ is defined for statistical purposes almost exactly as just the collection of people living together in a home: it is definitionally almost impossible for the two figures to diverge except when homes are left empty. I will discuss the phenomenon of empty homes later. But for the moment, set it to one side, and suppose that there are as many homes as households in the country.

If a home is removed from the system, a household gets ‘suppressed’. This is almost just a matter of arithmetic: a household is the collection of people who live in a home, so if there is one fewer home, then there has to be one fewer household. A key conclusion follows right away from this: if we take away either the £120,000 house or the £5m house, a household gets suppressed. In the case of the £120,000 house, it might perhaps be the household who would otherwise have inhabited that very house: since they are at the lower end of the housing market, they will be especially vulnerable to being pushed out of it altogether.

In the case of the £5m house, it is extremely unlikely to be the household who would otherwise have inhabited it: they are rich enough to buy almost any house in the market, so if their top choice is taken away, they will just buy their next-best option. But when they buy their next-best option, they push out the household who would counterfactually have bought that, who will then push out another household from the next house in the chain, and so on down a long ‘moving chain’ of properties, until finally we reach the unlucky household who don’t have an acceptable alternative property, and who get suppressed.

Although relatively affluent households do get suppressed sometimes, for obvious reasons this is much less likely. So the household that gets suppressed will probably look very different to the would-be inhabitants of the £5m house. Exactly how moving chains work is empirically complicated, but it seems that they often stretch down the whole length of the market, such that the household that ultimately gets suppressed if we take away the £5m house is not so different to the household that would have bought the £120,000 one. They are still the victims of the reduction in housing supply, though now their victimisation has become invisible.

In respect of household suppression, then, the ultimate effects of the two options may be surprisingly similar – assuming for now, as I mentioned, that there is no difference in the likelihood of the home being left empty. But the £5m house has a whole host of other effects. As we have noted, if the £5m house is not built, the household that would have occupied it are displaced into the next-best house, forming the first link of a long chain of households in counterfactually worse homes, leading ultimately to the unlucky household that gets suppressed. 

The corollary of that is that if the house is built, everyone in the ‘moving chain’ gets to move one link up. A potentially huge number of people get marginally better housing: for each household, the effect will probably be small, but the aggregated value will be considerable. And the beneficiaries will tend to be distributed across the housing market, from the households close to the £5m house, to the people close to the £120,000 one.

This is the basic reason that, counterintuitively, building high-value housing can help a lot of people on low incomes. The rather unappealing technical term for this is ‘filtering’. The appeal of filtering is simple: adding homes that are better quality than the existing stock allows people to move out of the existing stock into better homes, and frees up existing stock for suppressed households.

Adding houses at the bottom of the market, i.e. that are worse quality than existing homes, still creates homes for suppressed households, but is merely neutral for everyone in existing dwellings. In the long run, a country that adds its housing in this way will end up with a low-quality housing stock, an outcome that is worse for virtually everyone. This actually has happened to some extent in various times and places, such as in the Eastern Bloc in the second half of the 20th century.


So, is it always better to build the best housing possible? 

The answer is no: the situation is more complicated. First, the assumption I have made so far, that the number of households corresponds to the number of homes, is false: there are a substantial number of empty homes in the UK. Many of these are of limited relevance to filtering (e.g. abandoned homes in declining neighbourhoods, or unsold homes of elderly people who have moved into palliative care facilities).

But one group is of great importance, namely second homes. If the £5m house becomes a second home, then it does not necessarily precipitate the chain of events described above: it could happen that no household is spared from suppression, and hence that no chain of intermediate households enjoy better housing quality. Building second homes is still of course a good thing, all else being equal, but it does not necessarily alleviate the wider housing shortage.

We should be careful about jumping to conclusions here, though. If we never build any second homes, second homes won’t cease to exist: to a substantial extent, the would-be second-home buyers will just buy existing homes instead. So, counterintuitive though it may seem, purpose-built second homes often will still generate filtering effects. They only fail to do this if the second-home buyer would otherwise not buy a second home at all.

It should also be acknowledged that second homes are not all that common. According to the English Housing Survey, English people own about 500,000 second homes in the UK, a little under 2% of all UK homes. Scottish, Welsh and Northern Irish people presumably own second homes at similar rates. The number of second homes owned by people resident abroad is unknown: the Centre for Public Data found that about 250,000 British homes are owned by absentee foreigners in total, but most are certainly rented out or inhabited by the absentee foreigners’ children or other relations. We might thus conjecture an overall rate of 2.5% second homes. This will be higher in certain bits of the country, like the Lake District or Chelsea: maybe they will be a really substantial share of the expensive high-rise developments around Vauxhall or Old Street. But in most cases, second homes are actually a surprisingly minor consideration when we are thinking about filtering.

Here is a second complication: sometimes building homes has positive side effects that are not factored into their price. If we build attractive houses on a derelict site in a run-down town centre, we might improve the streetscape, help to renew the high street, make the neighbourhood feel safer and more vibrant, and foster more civic pride in the town as a whole. These are powerful reasons for building those houses, but they will not be reflected in the price to the extent that they are benefits to the whole community, not to the buyer individually. It could easily happen that they make it preferable to build those houses rather than to build houses with a higher market price elsewhere.

I believe this argument is absolutely correct. Indeed, in my view, such ‘externalities-based’ arguments are probably decisive far more often than the more familiar second-home-based arguments mentioned above. But they should be used discriminatingly. They are seldom relevant for greenfield developments, which do not normally have major positive side-effects in the same way. In the case of brownfield developments, positive side effects must be balanced against negative ones (e.g. pollution).

And we must retain a grip on scale. The £120,000 house might have £20,000’s worth of positive side effects. Maybe, in some extreme cases, it might have £100,000’s worth. But even then, it won’t necessarily outweigh the value of the long chain of improved housing that the £5m house generates.

Perhaps the trickiest complication surrounds the effect of new housing on immigration. High housing costs are a major deterrent from immigrating to the UK. When we build new homes, they marginally ease those costs, encouraging more people to immigrate. So when we build a new home, there is a chance that it attracts a new household to Britain rather than accommodating an existing one. Of course, there will still be a filtering effect in the immigrant’s home country, so given certain universalist premises, it is no less impactful. But from the point of view of the British government, concerned primarily with addressing Britain’s housing needs rather than those of other countries, it is liable to be.

Immigration effects certainly blunt the effect of new housing on easing housing scarcity. But the place of specifically high-value housing here is complicated. Around 42% of the 1.5 million non-visitor visas issued in the year to March 2023 were for students, almost all of whom will enter the lower end of the housing market. Some 18% were for refugees, who are also likely to seek cheaper accommodation. Some 32% of arrivals came on work visas, a total of around 300,000 people, together with their dependents. Of these, about 75,000 visas were granted to temporary workers, mostly in seasonal employment like agriculture; 102,000 were in the Health and Care category, mostly nurses and carers – both groups likely to be at the lower end of the housing market. Just 69,000 visas were issued in the ‘skilled worker’ category – about 8% of the total – and their minimum salary was set at £26,000, leaving many of them still clearly in the lower half of the housing market.

The upshot of this is that only a very small minority of immigrants enter the upper half of the housing market. For the great majority of potential immigrants, therefore, greater affordability lower down the market will do more to make immigrating to Britain appealing than greater affordability at the top will, since expensive properties are out of their reach anyway. Now, due to the filtering effect explained above, building at the top of the market generates affordability across the whole spectrum, so high-value housing is still likely to have a slightly greater overall immigration effect. But the difference is a small one.


Our intuitions about housing are not always wrong: there are certainly cases in which it is right to prioritise building low-value homes over high-value ones. I have explored some of these here, and there are surely others too. But even so, these cases may be somewhat exceptional. Our default assumption should probably be that high-value homes help a lot more people, both on low incomes and high ones.

If you want to defend prioritising low-value housing in some given case, you might be right. But you should reflect on which of these special circumstances might obtain, rather than just assuming that it must be better on account of being directly targeted at the groups most in need.

Britain has a planning system that profoundly reinforces inequalities, generating huge rents for landlords, landowners and a minority of homeowners, and imposing huge burdens on renters and first-time buyers. It is absolutely right to prioritise the worst-off when we think about how to reform it. But we must be thoughtful in the way we do this: intuitively appealing but poorly thought-through policies can have grave consequences, as the current system shows. The idea that we should build expensive houses to help low-income people may sound like a frivolous neoliberal provocation. But it isn’t. In many cases, it deserves very careful attention.

Click here to subscribe to our daily briefing – the best pieces from CapX and across the web.

CapX depends on the generosity of its readers. If you value what we do, please consider making a donation.

Samuel Hughes is Head of Housing at the Centre for Policy Studies.

Columns are the author's own opinion and do not necessarily reflect the views of CapX.