The kooky housing supply filtering debate
People must relocate to fill the stock of homes. Filtering must happen. So what is the controversy about?
Filtering is a concept in housing analysis that describes the fact that people occupying new homes come from other homes. Sometimes these relocations are described as moving chains.
For an outsider, the attention to such a niche concept might seem a little strange. After all, where else do people come from to occupy new homes if not existing ones?
Yet the concept of filtering attracts a lot of attention, particularly in the YIMBY movement. Here’s a typical headline about new research on filtering: “Building new units is proven to push rents down – but not for the reasons you may think.”
So revered is this filtering concept that people will claim that building one home adds more than one home to supply because of the relocations that happen.

I think part of the confusion arises because one side of the housing debate wants to use filtering as evidence that faster new housing production will automatically happen when there are no pesky regulations.
But instead of directly making this claim, they revert to a slightly different claim, which is that adding any new housing, even luxury homes, makes rents and prices cheaper across the whole market.
In other words, if regulations on density were removed, more new luxury apartments would be built in some locations, and the quantity effect on price from adding more luxury homes would be felt by all parts of the market, including budget homes, and across many substitute locations.
In case you missed it, check out my explanation of the problems behind the common assumption that more density equals more homes.
I suspect that the other side of the debate observes that property markets never seem to over-build and crash prices, nor solve unequal access to housing in prime locations on their own. But they can’t explain why in economic terms. So they argue that filtering doesn’t happen and so the quantity effect on rent and price can’t apply to the whole market.
They claim that adding more luxury homes doesn’t make housing cheaper, so filtering can’t be important. They say things like “Building more luxury apartments doesn’t help poor people.”
It is all very confusing. And I think the reason for this confusion is the inconsistency of the application of the filtering concept.
Consider the following.
First, filtering must exist, just as it does in car markets and other durable goods markets. So showing that it does by tracking households moving into other homes doesn’t reveal new information. It also doesn’t help explain at all the choice of how quickly new homes are built, which, for me, seems to be the main issue in housing supply debates.
But the next points really show how kooky the debate has become.
Second, if filtering is a powerful force that frees up housing in other locations and markets, then this logically implies that it doesn’t matter where new homes are built or what type of homes are built. After all, people will filter into those new homes from other locations and housing types.
Even if you think filtering is important, building the same number of extra budget homes is better than more luxury homes, as both generate the same filtering effects, but with more budget housing there are many cheaper options in the market.
Third, what about filtering on the production side? If people filter to where the housing is, then it doesn’t matter where homes are built, and housing developers can simply build new housing where it is planned. If regulations or heritage protections prohibit certain dwelling types in certain locations, then property owners will produce homes elsewhere in substitute locations, and they will filter across the market too.
The logic of filtering says it doesn’t matter what type of new housing is built or where. But the flip side of this logic is that it also implies that regulations on the location of new housing don’t have quantity effects because producers can filter to other locations and residents will filter there too.
The inconsistency with which the concept of filtering is applied makes me suspect the filtering debate is actually a proxy debate about whether market outcomes under different land use regulations can compress the wealth and income distribution in general.
An illustration
Below is a diagram that might help clarify the degree to which filtering can help understand overall income and wealth distribution.
The top row shows the initial conditions of four dwellings occupied by twelve people in this hypothetical housing market, with three people per dwelling.
The dwellings are plotted out horizontally in terms of household income and wealth. The poorest people are on the left shaded in blue, with the richest on the right and shaded yellow. Homes are also sorted in terms of inferior or budget (left) and superior or luxury (right) quality and location.
Then, two extra homes are built to create six homes.1
I’ve shown the two new homes added at the luxury end of the market. The people who move into the new homes must come from the existing 12 people.
Rows two and three of the diagram show what I think the debate is about.
The second row has the people reorganised into new households in one way—the way I think those who think filtering is a big deal have in mind. Here, the richest six people reorganise into three households and occupy the three most luxurious homes. Then the six poorest people reorganise from two to three households and occupy the inferior three homes.
This means that by adding two luxury homes, we have in effect created one extra home for the poor households to move into—since the rich households have moved to occupy the new homes and vacated one home—as well as one extra home for the rich people.
Fine.
But is this exactly even outcome, where benefits are distributed completely across the wealth and income distribution, what usually happens?
In the third row of the diagram, I’ve shown an alternative filtering scenario. Here, the six rich people who originally occupied two dwellings reorganise to occupy both their current dwellings and the two new ones. The two poor households see no change in their housing arrangement.
Filtering occurs in both of these scenarios, as it must. They only differ in the way that income and wealth differences constrain the ability of some households to share in the added housing. In one scenario, filtering benefits households at all points in the income and wealth distribution. In the other scenario, it doesn’t, and the outcome is greater inequality.
So what?
Filtering seems to be controversial because it touches on the fact that many of the housing issues that concern us are about the distribution of income and wealth, not the quantity of homes.
If every person owned the same equal share of the housing stock and had the same income, we would have very different housing policy concerns. Rising rents are a problem for renters but not for landlords. Rising prices are a problem for future buyers but not current property owners. If we all had the same financial exposure to property, these wouldn’t be politically contentious issues.
In my new book, The Great Housing Hijack, I call this problem the symmetry of property markets.
Distributional issues are usually solved by redistributing income and wealth, whether in the form of money or housing directly.
Market outcomes don’t usually tend towards greater equality by themselves. So I suspect that this is what filtering skeptics have in mind. The market choices of property owners to build more homes, whether luxury homes or not, probably won’t reverse the inequality of homeownership or income that leads to some households having limited housing choices.
But that doesn’t mean that people don’t reallocate into the physical homes that exist. The filtering process must happen. And if you believe this process is important, then you must also believe that regulations on the location of homes aren’t important in determining the total quantity of housing.
For now, just assume that these homes are built because of reasons. Let’s not concern ourselves with peering into the black hole of theory about why homes are built. But if you do want to dig in, start here:
Great summary. Clear and concise as always.
This really nails it: "Filtering seems to be controversial because it touches on the fact that many of the housing issues that concern us are about the distribution of income and wealth, not the quantity of homes."
I wish people would reframe housing affordability as "poverty". We used to use that word. We no longer do.
As your example shows, fixing the housing market (in your example, somehow enabling two more homes) might not fix the housing problem (overcrowding).
But fixing inequality would fix it.
If fixing the market doesn't fix the problem, but fixing inequality does, it it a problem of markets or inequality?
Let me stray into political science. My guess is that poverty was reframed as a housing market problem to dupe the left into supporting reforms that work for vested interests but not for the left's traditional constituency, the poor.
I see the filtering argument being used to say, essentially,
"New luxury homes create affordable housing. We don't need affordable housing programs. Stop complaining that all the new housing is large and expensive."
e.g. https://www.huduser.gov/portal/periodicals/cityscape/vol25num3/ch6.pdf