Blog Prices/Rents

Movin’ On Up: How Costly New Homes Create Affordable Old Homes

January 22, 2024

In housing policy, a common theory of how to provide the maximum number of affordable homes to the greatest number of people is known as “filtering:” the process through which aging homes depreciate, and become less costly as higher-income residents purchase new homes, and move into them from the older, depreciated homes. 

Under filtering theory, people concerned with housing affordability should support the construction of new homes to ensure a robust supply of high quality but lower priced “used” homes in the future.

Unfortunately, particularly in markets where it is onerous, costly, and/or time-consuming to build a lot of new homes, filtering can take a long time. While a used car might lose half its value in three years, a used home in a market with slow housing growth can take years to become affordable to middle class households.In this paper, “Does new housing for the rich benefit the poor? On trickle-down effects of new homes,” Gabriella Kindström and Che-Yuan Liang “use microdata on the Swedish population and housing stock (1990-2017) to investigate how building new homes affects the housing distribution across income groups.”

Key Takeaways

  • Microdata proves that filtering is real: while new homes are mostly for the rich, “poor people are overrepresented among in-movers to vacated homes.”
  • New homes take about 30 years to filter down to an even income distribution.
  • New housing benefits everyone by giving every income group better access to newer, more spacious housing.

Without robust new housing construction, old homes can reverse filter and become more expensive, as is the case in older cities like San Francisco and New York, where a dearth of new housing means old houses are routinely renovated into modern luxury homes.

However, new housing still does not exist in a vacuum, and most people who move into new homes move out of existing, less-expensive homes in the same city. As such, new housing can theoretically free up inexpensive units at the low end of the market by inducing what researchers call “moving chains:” As people “move on up” into newer and nicer housing, they create vacancy in the older, less-expensive homes they are leaving. Go far enough down the chain and it’s possible that building a new high-end home pretty quickly frees up an existing home that’s affordable to a working class family.

In 2019, housing researcher Evan Mast used data from a mailing list company to document moving chains for the first time, finding that new market rate housing does in fact free up older, less expensive homes. Other researchers using different data sets have followed suit and found similar results.

For their contribution to the literature on filtering and moving chains, Kindström and Liang leverage data from the GeoSweden database, which covers the entire population of the country and includes information on income, taxes and transfers, and the entire Swedish housing stock—including geographical location, building type, construction year, size, assessed value, ownership, and renovation—to follow how individuals move between homes.

The headline finding is consistent with Mast’s earlier findings, and affirms that new market rate housing starts a moving chain that quickly moves down the income ladder to low-income households. While the initial occupants of new-build homes have higher incomes, the third round of the migration chain reaches households making 60% of the average income. Put differently, new high-end housing frees up older homes for low-income households.

Filtering works fastest in rental apartments, followed by owned apartments (what we call condos here in the US), and finally detached single family homes. Further, rental apartments helped low-income households afford the most additional floorspace. 

Finally, the researchers found that new housing supply appears to lead to newer, larger housing for every income group.(For additional detail, see the land economist Stephen Hoskins’ annotated graphs from the study here.)