kvlak limn

Field notes · Census Building Permits Survey

Who Actually
Builds Housing

A permit is the moment a home stops being a plan and becomes a legal possibility - the earliest hard number on where America is letting housing get built. Read per resident instead of in raw counts, the map turns strange: the places approving the most homes are mid-size Sun Belt and Mountain West metros most people cannot name, while the coastal cities where a house costs a fortune permit almost nothing at all. This is the supply side of affordability, stamped one form at a time.

1.43M housing units permitted, 2025 25 metro areas tracked Full
1.43M
Housing units permitted, 2025
15.8/1k
Top metro rate (Myrtle Beach)
4.7×
Boomtown vs frozen-metro gap
32%
Of new units are apartments (5+)
A-01

Where Homes Get Approved

Units per 1,000 residents

Shade the country by supply rate - housing units authorized per 1,000 residents, not raw count - and a diagonal appears. A dark band of permitting runs from the Mountain West down through the Carolinas and Florida. The palest states are the coastal Northeast and the industrial Midwest: large, established populations clearing remarkably few new homes. Raw counts would hide this; per capita is the lens that makes a big slow state and a small fast one comparable.

Alabama: 3.7 per 1,000 (19,061 units) Alaska: 1.1 per 1,000 (848 units) Arizona: 6.8 per 1,000 (51,532 units) Colorado: 5.7 per 1,000 (33,754 units) Florida: 7.6 per 1,000 (178,297 units) Georgia: 5.5 per 1,000 (61,196 units) Indiana: 4.2 per 1,000 (29,076 units) Kansas: 3.5 per 1,000 (10,299 units) Maine: 5.0 per 1,000 (6,979 units) Massachusetts: 1.7 per 1,000 (12,477 units) Minnesota: 3.6 per 1,000 (20,947 units) New Jersey: 2.9 per 1,000 (27,661 units) North Carolina: 7.8 per 1,000 (86,167 units) North Dakota: 3.1 per 1,000 (2,440 units) Oklahoma: 3.8 per 1,000 (15,513 units) Pennsylvania: 2.0 per 1,000 (25,709 units) South Dakota: 5.7 per 1,000 (5,305 units) Texas: 6.7 per 1,000 (210,217 units) Wyoming: 3.8 per 1,000 (2,238 units) Connecticut: 2.0 per 1,000 (7,273 units) Missouri: 2.8 per 1,000 (17,474 units) West Virginia: 2.2 per 1,000 (3,962 units) Illinois: 1.5 per 1,000 (18,551 units) New Mexico: 3.6 per 1,000 (7,747 units) Arkansas: 5.0 per 1,000 (15,492 units) California: 2.6 per 1,000 (103,856 units) Delaware: 6.1 per 1,000 (6,468 units) District of Columbia: 2.3 per 1,000 (1,591 units) Hawaii: 2.6 per 1,000 (3,708 units) Iowa: 4.2 per 1,000 (13,664 units) Kentucky: 3.3 per 1,000 (14,940 units) Maryland: 2.1 per 1,000 (13,358 units) Michigan: 2.3 per 1,000 (23,075 units) Mississippi: 2.9 per 1,000 (8,405 units) Montana: 4.6 per 1,000 (5,258 units) New Hampshire: 3.6 per 1,000 (5,030 units) New York: 1.9 per 1,000 (38,667 units) Ohio: 2.8 per 1,000 (33,640 units) Oregon: 3.5 per 1,000 (14,839 units) Tennessee: 6.0 per 1,000 (43,374 units) Utah: 7.6 per 1,000 (26,775 units) Virginia: 3.8 per 1,000 (33,094 units) Washington: 4.4 per 1,000 (34,913 units) Wisconsin: 4.3 per 1,000 (25,569 units) Nebraska: 5.3 per 1,000 (10,596 units) South Carolina: 8.4 per 1,000 (45,862 units) Idaho: 9.0 per 1,000 (17,944 units) Nevada: 5.7 per 1,000 (18,540 units) Vermont: 3.5 per 1,000 (2,294 units) Louisiana: 3.1 per 1,000 (14,203 units) Rhode Island: 1.6 per 1,000 (1,738 units)
Units per 1,000 residents
  • < 3.0
  • 3.0 - 4.5
  • 4.5 - 6.0
  • 6.0 - 8.0
  • 8.0 +
U.S. avg ~4.2

Fig. A-01 · Choropleth, 51 jurisdictions, single-hue mauve density ramp. Hover a state for its rate and unit count; the ranked table below carries every value so the reading never rests on colour alone. Census BPS Annual 2025 units over Census Vintage 2024 resident population.

All 51 jurisdictions, ranked (data table)
StatePer 1,000UnitsMultifamilyRate
Idaho 9.0 17,944 16%
South Carolina 8.4 45,862 11%
North Carolina 7.8 86,167 23%
Florida 7.6 178,297 35%
Utah 7.6 26,775 34%
Arizona 6.8 51,532 32%
Texas 6.7 210,217 29%
Delaware 6.1 6,468 14%
Tennessee 6.0 43,374 24%
Colorado 5.7 33,754 46%
Nevada 5.7 18,540 31%
South Dakota 5.7 5,305 34%
Georgia 5.5 61,196 25%
Nebraska 5.3 10,596 42%
Arkansas 5.0 15,492 25%
Maine 5.0 6,979 27%
Montana 4.6 5,258 36%
Washington 4.4 34,913 43%
Wisconsin 4.3 25,569 41%
Indiana 4.2 29,076 29%
Iowa 4.2 13,664 41%
Oklahoma 3.8 15,513 18%
Virginia 3.8 33,094 36%
Wyoming 3.8 2,238 17%
Alabama 3.7 19,061 17%
Minnesota 3.6 20,947 36%
New Hampshire 3.6 5,030 28%
New Mexico 3.6 7,747 16%
Kansas 3.5 10,299 23%
Oregon 3.5 14,839 33%
Vermont 3.5 2,294 38%
Kentucky 3.3 14,940 34%
Louisiana 3.1 14,203 17%
North Dakota 3.1 2,440 25%
Mississippi 2.9 8,405 11%
New Jersey 2.9 27,661 48%
Missouri 2.8 17,474 27%
Ohio 2.8 33,640 43%
California 2.6 103,856 40%
Hawaii 2.6 3,708 29%
District of Columbia 2.3 1,591 86%
Michigan 2.3 23,075 27%
West Virginia 2.2 3,962 4%
Maryland 2.1 13,358 27%
Connecticut 2.0 7,273 67%
Pennsylvania 2.0 25,709 34%
New York 1.9 38,667 66%
Massachusetts 1.7 12,477 55%
Rhode Island 1.6 1,738 29%
Illinois 1.5 18,551 34%
Alaska 1.1 848 19%
A-02

A Country That Clusters Low

States by supply rate

Line the 51 jurisdictions up by how many homes each permits per resident and the shape is lopsided: a thick stack of below-average states, then a thinning above-average tail of fast builders. 30 of 51 sit under the U.S. average; the median jurisdiction (New Mexico) permits about 3.6 per 1,000. The country's housing supply is decided by a small number of outliers.

0 4 8 12 16 U.S. avg 4.2 1-2 per 1,000: 5 states 5 1 2-3 per 1,000: 12 states 12 2 3-4 per 1,000: 13 states 13 3 4-5 per 1,000: 5 states 5 4 5-6 per 1,000: 7 states 7 5 6-7 per 1,000: 4 states 4 6 7-8 per 1,000: 3 states 3 7 8-9 per 1,000: 1 states 1 8 Housing units permitted per 1,000 residents (bin lower bound)
  • Below U.S. average
  • At or above average

Fig. A-02 · Each column counts the states whose supply rate falls in that unit-wide band. Bar caps are direct-labelled with the state count. Census BPS Annual data over Census Vintage 2024 population.

A-03

The Metro Leaderboard

Top builders, per resident

The same 25 metros, ranked by units permitted per 1,000 residents, each bar split into single-family and multifamily (5+ units) so rank and building shape read at once. The order is the whole argument: the leaders are mid-size Sun Belt and Mountain West metros, and the biggest, costliest coastal names sink to the bottom. Rank runs opposite to price.

  1. 01 Myrtle Beach, SC 15.8
  2. 02 Boise City, ID 11.2
  3. 03 Raleigh-Cary, NC 11.2
  4. 04 Austin-Round Rock, TX 10.7
  5. 05 Provo-Orem, UT 9.6
  6. 06 Orlando-Kissimmee, FL 9.5
  7. 07 Nashville-Davidson, TN 8.9
  8. 08 Houston-The Woodlands, TX 8.3
  9. 09 Dallas-Fort Worth, TX 7.9
  10. 10 Charlotte-Concord, NC 7.7
  11. 11 Phoenix-Mesa, AZ 7.7
  12. 12 Jacksonville, FL 7.6
  13. 13 Tampa-St. Petersburg, FL 6.6
  14. 14 Atlanta-Sandy Springs, GA 5.3
  15. 15 Denver-Aurora, CO 5.2
  16. 16 Minneapolis-St. Paul, MN 3.9
  17. 17 Seattle-Tacoma, WA 3.8
  18. 18 Miami-Fort Lauderdale, FL 3.6
  19. 19 Washington-Arlington, DC 2.8
  20. 20 New York-Newark, NY 2.2
  21. 21 Los Angeles-Long Beach, CA 2.1
  22. 22 Boston-Cambridge, MA 2.0
  23. 23 Philadelphia-Camden, PA 2.0
  24. 24 San Francisco-Oakland, CA 1.7
  25. 25 Chicago-Naperville, IL 1.7

Fig. A-03 · Bar length is the supply rate; the mauve share is multifamily. Full counts in the table below. Census BPS Annual 2025 units over Census Vintage 2024 population.

Full metro table (units, single, multifamily)
MetroPer 1,000Total unitsSingle-famMultifamilyMulti %
Myrtle Beach, SC 15.8 6,511 6,418 77 1%
Boise City, ID 11.2 9,438 7,970 1,218 13%
Raleigh-Cary, NC 11.2 17,543 11,760 5,542 32%
Austin-Round Rock, TX 10.7 27,322 14,810 11,749 43%
Provo-Orem, UT 9.6 7,304 5,395 1,695 23%
Orlando-Kissimmee, FL 9.5 27,845 14,198 13,463 48%
Nashville-Davidson, TN 8.9 19,203 13,267 5,827 30%
Houston-The Woodlands, TX 8.3 65,075 46,219 16,385 25%
Dallas-Fort Worth, TX 7.9 66,179 39,790 24,607 37%
Charlotte-Concord, NC 7.7 22,233 16,242 5,820 26%
Phoenix-Mesa, AZ 7.7 39,699 23,755 14,800 37%
Jacksonville, FL 7.6 13,455 9,512 3,700 27%
Tampa-St. Petersburg, FL 6.6 22,747 13,596 8,771 39%
Atlanta-Sandy Springs, GA 5.3 33,670 22,065 11,052 33%
Denver-Aurora, CO 5.2 15,948 7,477 8,195 51%
Minneapolis-St. Paul, MN 3.9 14,516 8,868 5,342 37%
Seattle-Tacoma, WA 3.8 15,659 6,136 8,199 52%
Miami-Fort Lauderdale, FL 3.6 23,453 5,873 16,699 71%
Washington-Arlington, DC 2.8 17,990 10,168 7,392 41%
New York-Newark, NY 2.2 44,444 10,477 31,550 71%
Los Angeles-Long Beach, CA 2.1 27,242 12,883 12,525 46%
Boston-Cambridge, MA 2.0 10,072 3,971 5,695 57%
Philadelphia-Camden, PA 2.0 12,900 7,256 4,935 38%
San Francisco-Oakland, CA 1.7 7,750 2,713 4,884 63%
Chicago-Naperville, IL 1.7 15,778 9,497 5,324 34%
A-04

Bigger Metros Build Less

Population vs supply rate

Plot every metro by population and by supply rate and the dots drift down as they drift right: the largest metros - the New Yorks and Los Angeleses where housing is scarcest and dearest - sit along the floor, while the fastest permitting happens in the smaller Sun Belt boomtowns up the left side. Size is not the constraint. The places with the most people and the most demand authorize the fewest homes per resident.

0 4 8 12 16 500k 1M 2M 5M 10M 20M Resident population (log scale) Units permitted / 1,000 residents Myrtle Beach, SC: 15.8 per 1,000, pop 413,391 Myrtle Beach Boise City, ID: 11.2 per 1,000, pop 845,877 Raleigh-Cary, NC: 11.2 per 1,000, pop 1,562,009 Austin-Round Rock, TX: 10.7 per 1,000, pop 2,550,637 Provo-Orem, UT: 9.6 per 1,000, pop 760,531 Orlando-Kissimmee, FL: 9.5 per 1,000, pop 2,940,513 Nashville-Davidson, TN: 8.9 per 1,000, pop 2,150,553 Houston-The Woodlands, TX: 8.3 per 1,000, pop 7,796,182 Houston Dallas-Fort Worth, TX: 7.9 per 1,000, pop 8,344,032 Dallas Charlotte-Concord, NC: 7.7 per 1,000, pop 2,883,370 Phoenix-Mesa, AZ: 7.7 per 1,000, pop 5,186,958 Phoenix Jacksonville, FL: 7.6 per 1,000, pop 1,760,548 Tampa-St. Petersburg, FL: 6.6 per 1,000, pop 3,424,560 Atlanta-Sandy Springs, GA: 5.3 per 1,000, pop 6,411,149 Denver-Aurora, CO: 5.2 per 1,000, pop 3,052,498 Minneapolis-St. Paul, MN: 3.9 per 1,000, pop 3,757,952 Seattle-Tacoma, WA: 3.8 per 1,000, pop 4,145,494 Miami-Fort Lauderdale, FL: 3.6 per 1,000, pop 6,457,988 Washington-Arlington, DC: 2.8 per 1,000, pop 6,436,489 New York-Newark, NY: 2.2 per 1,000, pop 19,940,274 New York Los Angeles-Long Beach, CA: 2.1 per 1,000, pop 12,927,614 Los Angeles Boston-Cambridge, MA: 2.0 per 1,000, pop 5,025,517 Philadelphia-Camden, PA: 2.0 per 1,000, pop 6,330,422 San Francisco-Oakland, CA: 1.7 per 1,000, pop 4,648,486 San Francisco Chicago-Naperville, IL: 1.7 per 1,000, pop 9,408,576 Chicago
  • Sun Belt boomtown
  • Steady
  • Supply-frozen

Fig. A-04 · One mark per metro; shape and colour both encode archetype, so the groups read under colour-blindness and in greyscale. Extremes direct-labelled; hover any dot for its figures. Census BPS Annual 2025 units over Census Vintage 2024 population.

A-05

Houses or Apartments

Share of units by building type

Now hold volume constant and ask only about shape. Each bar is one metro's permitted units normalized to 100% and split into single-family, the 2 to 4 unit "missing middle", and multifamily (5+), sorted from most detached at the top to most apartment at the bottom. Sun Belt sprawl and Texas subdivisions anchor one end; the supply-frozen coasts, where almost every new unit is an apartment, anchor the other. Two different housing economies, drawn to the same width.

Myrtle Beach, SC 99%
Boise City, ID 84%
Provo-Orem, UT 74% 23%
Houston-The Woodlands, TX 71% 25%
Charlotte-Concord, NC 73% 26%
Jacksonville, FL 71% 27%
Nashville-Davidson, TN 69% 30%
Raleigh-Cary, NC 67% 32%
Atlanta-Sandy Springs, GA 66% 33%
Chicago-Naperville, IL 60% 34%
Minneapolis-St. Paul, MN 61% 37%
Dallas-Fort Worth, TX 60% 37%
Phoenix-Mesa, AZ 60% 37%
Philadelphia-Camden, PA 56% 38%
Tampa-St. Petersburg, FL 60% 39%
Washington-Arlington, DC 57% 41%
Austin-Round Rock, TX 54% 43%
Los Angeles-Long Beach, CA 47% 46%
Orlando-Kissimmee, FL 51% 48%
Denver-Aurora, CO 47% 51%
Seattle-Tacoma, WA 39% 52%
Boston-Cambridge, MA 39% 57%
San Francisco-Oakland, CA 35% 63%
New York-Newark, NY 24% 71%
Miami-Fort Lauderdale, FL 25% 71%

Bars normalized to 100%. Small segments omit their label; the middle band is the 2 to 4 unit class.

A-06

Fifteen Years of Supply

U.S. units permitted, 000s

National housing units authorized each year, stacked by building type. The recovery from the 2008 collapse was carried by multifamily nearly as much as single-family - the apartment share crested around 2022, at roughly 39% of all new units. The 2-to-4-unit "missing middle" stayed a rounding error the entire stretch, the housing type America stopped building.

0 250 500 750 1,000 1,250 1,500 1,750 2010: 610k units (single 452k, 2-4 22k, multi 136k) '10 2011: 628k units (single 420k, 2-4 22k, multi 186k) '11 2012: 830k units (single 519k, 2-4 26k, multi 285k) '12 2013: 991k units (single 621k, 2-4 29k, multi 341k) '13 2014: 1,050k units (single 637k, 2-4 30k, multi 383k) '14 2015: 1,186k units (single 698k, 2-4 33k, multi 455k) '15 2016: 1,210k units (single 753k, 2-4 35k, multi 422k) '16 2017: 1,282k units (single 820k, 2-4 37k, multi 425k) '17 2018: 1,332k units (single 857k, 2-4 40k, multi 435k) '18 2019: 1,389k units (single 864k, 2-4 43k, multi 482k) '19 2020: 1,471k units (single 979k, 2-4 47k, multi 445k) '20 2021: 1,737k units (single 1115k, 2-4 53k, multi 569k) '21 2022: 1,680k units (single 974k, 2-4 55k, multi 651k) '22 2023: 1,511k units (single 920k, 2-4 55k, multi 536k) '23 2024: 1,478k units (single 982k, 2-4 54k, multi 442k) '24 2025: 1,432k units (single 912k, 2-4 56k, multi 464k) '25
  • Single-family (1 unit)
  • 2 to 4 units
  • Multifamily (5+ units)

Fig. A-06 · Thousands of units authorized nationwide; segment heights are the three building classes. 2025 totalled 1,432k. Source: Census BPS annual state files, summed to national.

A-07

The Apartment Share, Rising

Multifamily % of U.S. units

Read as a rate rather than a volume, the same national data tells a structural story: the multifamily share of everything permitted climbed from about 22% in 2010 to a peak near 39% in 2022, before easing to 32%. The country is not just building - it is quietly building denser, one approval at a time.

0% 10% 20% 30% 40% '10'12'14'16'18'20'22'24 39% (2022) 32%

Fig. A-07 · Multifamily (5+ unit) units as a share of all units authorized nationwide each year. Single series; peak and latest year direct-labelled. Source: Census BPS annual state files, summed to national.

A-08

Three Tempos of Building

Permits indexed, 2010 = 100

Group the metros by archetype and index each group to its own 2010 level - the bottom of the post-crash trough - and the gap becomes a matter of slope. Every cohort recovered, but not evenly: Sun Belt boomtowns ran to roughly three and a half times their 2010 pace and kept climbing; the steady metros and the supply-frozen coasts each about doubled, then stalled and gave ground after the 2021 peak. Same demand, three different tempos of supply - and the boomtowns pulled decisively away.

Sun Belt boomtowns
100 200 300 400 '10 '25 351

Led the recovery out of the 2010 trough and never let up - roughly three and a half times their 2010 pace by 2025.

Steady metros
100 200 300 400 '10 '25 220

Doubled off the 2010 bottom, then flattened - a milder arc than the boomtowns, without their late surge.

Supply-frozen metros
100 200 300 400 '10 '25 216

Also about doubled from the crash low, but from a far smaller base, and gave much of it back after 2021.

Fig. A-08 · Each panel indexes one cohort's total permits to 2010 = 100; the two grey lines repeat the other cohorts for scale. Cohorts are an editorial grouping of the metros; the trajectories are their real summed BPS permits.

A-09

The Supply Gap

Deviation from the U.S. average

Here is the whole argument on one axis. Each bar is a metro's distance from the national average of 4.2 units per 1,000 residents - above to the right, below to the left. The fast-permitting Sun Belt and Mountain West metros authorize about 4.7x as many homes per resident as the big coastal metros where demand and prices are highest. Where housing costs the most, the least gets built.

U.S. avg 4.2 Myrtle Beach, SC: 15.8 per 1,000 (+11.6 vs avg) Myrtle Beach 15.8 Boise City, ID: 11.2 per 1,000 (+7.0 vs avg) Boise City 11.2 Raleigh-Cary, NC: 11.2 per 1,000 (+7.0 vs avg) Raleigh-Cary 11.2 Austin-Round Rock, TX: 10.7 per 1,000 (+6.5 vs avg) Austin-Round Rock 10.7 Provo-Orem, UT: 9.6 per 1,000 (+5.4 vs avg) Provo-Orem 9.6 Orlando-Kissimmee, FL: 9.5 per 1,000 (+5.3 vs avg) Orlando-Kissimmee 9.5 Nashville-Davidson, TN: 8.9 per 1,000 (+4.7 vs avg) Nashville-Davidson 8.9 Houston-The Woodlands, TX: 8.3 per 1,000 (+4.1 vs avg) Houston-The Woodlands 8.3 Dallas-Fort Worth, TX: 7.9 per 1,000 (+3.7 vs avg) Dallas-Fort Worth 7.9 Charlotte-Concord, NC: 7.7 per 1,000 (+3.5 vs avg) Charlotte-Concord 7.7 Phoenix-Mesa, AZ: 7.7 per 1,000 (+3.5 vs avg) Phoenix-Mesa 7.7 Jacksonville, FL: 7.6 per 1,000 (+3.4 vs avg) Jacksonville 7.6 Tampa-St. Petersburg, FL: 6.6 per 1,000 (+2.4 vs avg) Tampa-St. Petersburg 6.6 Atlanta-Sandy Springs, GA: 5.3 per 1,000 (+1.1 vs avg) Atlanta-Sandy Springs 5.3 Denver-Aurora, CO: 5.2 per 1,000 (+1.0 vs avg) Denver-Aurora 5.2 Minneapolis-St. Paul, MN: 3.9 per 1,000 (-0.3 vs avg) Minneapolis-St. Paul 3.9 Seattle-Tacoma, WA: 3.8 per 1,000 (-0.4 vs avg) Seattle-Tacoma 3.8 Miami-Fort Lauderdale, FL: 3.6 per 1,000 (-0.6 vs avg) Miami-Fort Lauderdale 3.6 Washington-Arlington, DC: 2.8 per 1,000 (-1.4 vs avg) Washington-Arlington 2.8 New York-Newark, NY: 2.2 per 1,000 (-2.0 vs avg) New York-Newark 2.2 Los Angeles-Long Beach, CA: 2.1 per 1,000 (-2.1 vs avg) Los Angeles-Long Beach 2.1 Boston-Cambridge, MA: 2.0 per 1,000 (-2.2 vs avg) Boston-Cambridge 2.0 Philadelphia-Camden, PA: 2.0 per 1,000 (-2.2 vs avg) Philadelphia-Camden 2.0 San Francisco-Oakland, CA: 1.7 per 1,000 (-2.5 vs avg) San Francisco-Oakland 1.7 Chicago-Naperville, IL: 1.7 per 1,000 (-2.5 vs avg) Chicago-Naperville 1.7
  • Above U.S. average
  • Below U.S. average

Fig. A-09 · Bars measure each metro's supply rate against the national average; length is the size of the gap, direction is its sign. Census BPS Annual 2025 units over Census Vintage 2024 population.

Every metro vs the U.S. average (data table)
MetroPer 1,000vs U.S. avgTotal units
Myrtle Beach, SC 15.8 +11.6 6,511
Boise City, ID 11.2 +7.0 9,438
Raleigh-Cary, NC 11.2 +7.0 17,543
Austin-Round Rock, TX 10.7 +6.5 27,322
Provo-Orem, UT 9.6 +5.4 7,304
Orlando-Kissimmee, FL 9.5 +5.3 27,845
Nashville-Davidson, TN 8.9 +4.7 19,203
Houston-The Woodlands, TX 8.3 +4.1 65,075
Dallas-Fort Worth, TX 7.9 +3.7 66,179
Charlotte-Concord, NC 7.7 +3.5 22,233
Phoenix-Mesa, AZ 7.7 +3.5 39,699
Jacksonville, FL 7.6 +3.4 13,455
Tampa-St. Petersburg, FL 6.6 +2.4 22,747
Atlanta-Sandy Springs, GA 5.3 +1.1 33,670
Denver-Aurora, CO 5.2 +1.0 15,948
Minneapolis-St. Paul, MN 3.9 -0.3 14,516
Seattle-Tacoma, WA 3.8 -0.4 15,659
Miami-Fort Lauderdale, FL 3.6 -0.6 23,453
Washington-Arlington, DC 2.8 -1.4 17,990
New York-Newark, NY 2.2 -2.0 44,444
Los Angeles-Long Beach, CA 2.1 -2.1 27,242
Boston-Cambridge, MA 2.0 -2.2 10,072
Philadelphia-Camden, PA 2.0 -2.2 12,900
San Francisco-Oakland, CA 1.7 -2.5 7,750
Chicago-Naperville, IL 1.7 -2.5 15,778
A-11

Methodology

Notes on the Data

Everything here descends from one dataset: the U.S. Census Bureau, Building Permits Survey (BPS) (Annual 2025). BPS counts permits authorized - the moment a jurisdiction clears a home to be built - not homes finished. It is the earliest and most complete read on where America is letting housing happen. Each permit records a unit count and a building size class: single-family (1 unit), 2-unit, 3-and-4-unit, and multifamily (5 or more). Every per-capita figure divides units by Census resident-population estimates for the same place, so a small fast state and a big slow one land on one scale.

What's real, what's editorial

Every number on this page is drawn from the real BPS release. The state rates, the metro rankings, the sixteen-year national trend, and the rising apartment share all come straight from the Census BPS annual bulk files (state st2025a, metro cbsa2025a, and the metro annual series back to 2010). Per-capita figures divide those unit counts by Census Population Estimates, Vintage 2024 - the latest published - so the 2025 permits are read against 2024 population, the standard one-year lag. Two things are editorial, not source fields, and are labelled as such where they appear: which 25 metros the leaderboard tracks, and the archetype tag on each - boomtown, steady, frozen. The archetype trajectories index each cohort's real summed permits to its 2010 level. The ingest is reproducible: the bulk files live in data/raw/, and npm run data rebuilds every figure without a line of component code changing. The swap-point is documented in HANDOFF.md.

What you're not seeing

Permits are not starts and not completions - some authorized units are never built, and there is a lag of months to years before a permit becomes a home. Roughly a fifth of U.S. housing is built in areas that do not require permits or do not report to the survey, so rural and unincorporated construction is undercounted. "Units" counts dwellings, not square footage or affordability: a metro can lead on units per capita while building homes few of its residents can buy. And a permit says nothing about who the housing is for - this page measures supply volume and its single- vs multifamily shape, not tenure, price, or who ends up living there.


Generated 2026-07-10 10:55 UTC

Source: U.S. Census Bureau, Building Permits Survey (BPS)