GovSpendTracker

← All articles

Major U.S. cities are not building the housing the shortage requires

The national gap is still 3.7 million homes. New York and Chicago authorize a fraction of the homes per person that Dallas and Houston do.

October 7, 2026 · 8 min read · Drafted by an AI agent, reviewed before publishing

Freddie Mac still puts the U.S. housing shortage at 3.7 million units as of the third quarter of 2024, almost unchanged from 3.8 million at the end of 2020, because new households absorbed the 5.8 million units added in between. In 2024 the New York metro authorized 2.9 homes per 1,000 residents and Chicago 1.9, against 8.6 in Dallas–Fort Worth and 8.4 in Houston. Auckland's 2016 citywide upzone, which roughly tripled capacity on about three-quarters of residential land, is the measured version of the reform those cities have not made.

The shortage did not close

The United States is still short about 3.7 million homes. That is Freddie Mac's estimate through the third quarter of 2024, from the Census Housing Vacancy Survey: the stock is below what the current population would need if vacancy returned to its 1994–2003 rate and inflation-adjusted housing costs returned to their historical level. In late 2020 the same method put the gap at 3.8 million homes. Since then the country added 5.8 million housing units. Households grew by almost as much. The shortage barely moved. Freddie Mac, Housing supply: still undersupplied

A functioning market needs vacant homes. Freddie Mac's benchmark vacancy rate is 11.7 percent. The actual rate in the third quarter of 2024 was 10.1 percent. The missing vacancies, plus the households that high costs are keeping from forming, are the shortage above.

Where the permits actually are

In 2024, local offices authorized 1,478,000 homes, 2.2 percent fewer than in 2023. Texas authorized 225,756 homes. California authorized 101,545 homes. New York State authorized 46,549 homes. Census Bureau, 2024 building permits highlights

The metro comparison is the policy fact. Populations are the Census Bureau's July 1, 2024 estimates. Permits are homes authorized in calendar 2024. The rate is permits divided by that population.

MetroPopulationHomes authorized, 2024Per 1,000 residents
Dallas–Fort Worth8,344,032 people71,788 homes8.6
Houston7,796,182 people65,747 homes8.4
Seattle4,145,494 people17,920 homes4.3
New York19,940,274 people57,929 homes2.9
Chicago9,408,576 people18,064 homes1.9

Sources: Census Bureau, metro population estimates, July 1, 2024; Census Bureau, 2024 building permits highlights. Rates divide 2024 authorizations by the July 1, 2024 population.

New York has more than twice Dallas–Fort Worth's population and authorized fewer homes. At Dallas's 2024 rate, the New York metro would have authorized 171,555 homes, against the 57,929 homes that were issued. Chicago is larger than Houston and authorized about a quarter as many. Seattle, a high-price coastal metro, permits at about half the Dallas and Houston rate and well above New York and Chicago. Los Angeles authorized 26,781 homes, behind Dallas, Houston, New York, Phoenix, Atlanta, and Austin.

The reform that has been measured

In 2016 Auckland upzoned about three-quarters of its residential land and roughly tripled how many homes that land could hold. A synthetic-control study finds that new-dwelling permits per thousand residents about doubled within five years, from just under 6 to 12.2, against 5.9 in the comparison group. Over the seven years after the plan, about 52,200 permits — a bit under half of those issued — are attributed to the reform, roughly an 87 percent increase over the counterfactual. Greenaway-McGrevy and Phillips, Auckland upzoning (EPC-WP017)

That is a large change in what could be built, on most of the city's residential land. It is not a rezoning of one corridor.

Ending single-family zoning as a label, the way Minneapolis did in its 2040 plan, is a smaller reform, and the studies of modest upzones often find no clear jump in construction. The Auckland result is the one at the scale of a multi-million-home national gap: most residential land, much more capacity, and permits that follow.

What a major city would change

The Sun Belt metros in the table are not short of demand. They lack the rules that keep a Dallas-sized permit flow out of New York and Chicago. Three rules match what was actually measured:

  • Apartments are legal on most residential land, not on a map of exceptions.
  • Approval is by right when the building matches the rule. A hearing is a cap.
  • The capacity increase is large. A duplex on a few lots will not move the Census permit totals the way a citywide tripling of capacity did.

Interest rates slowed starts after 2022. They did not create the shortage. Freddie Mac's gap was already 3.8 million homes at the end of 2020, before the rate spike. Freddie Mac, Housing supply: still undersupplied

What this is not saying

It is not saying Houston's deed restrictions and floodplain are a zoning code to photocopy. It is saying the metros with the worst prices authorize homes at a fraction of the rate of metros that are cheaper and still growing.

It is not saying demand is fake. Freddie Mac's method counts households that high costs keep from forming. Building only as fast as new households appear is what the last four years did, and the gap stayed.

Sources & further reading