How to Chart US Housing Permits Since 1990 in Python

On this page

Problem statement

How much housing does the US build now compared with the boom of the mid-2000s, and has it recovered from the crash? The Census Bureau's county permit files go back to 1990, one file a year, and stacked together they give a 36-year series by county and by size of building. This page builds that series, measures the 2005 peak and the 2009 trough, separates houses from apartments, and finds the states and counties that have and have not returned to their boom-year building.

Quick answer

Read the 36 annual files with permits(), stack them with a year column and sum:

years = pd.concat([permits(f"co{y}a.txt").assign(year=y) for y in range(1990, 2026)], ignore_index=True)
total = years.groupby("year").units.sum()
print(f"peak {total.idxmax()}: {total.max():,}; trough {total.idxmin()}: {total.min():,}; 2025: {total[2025]:,}")

peak 2005: 2,155,316; trough 2009: 582,963; 2025: 1,431,616

Permits peaked at 2,155,316 units in 2005 and fell to 582,963 in 2009, a quarter of the peak. The 1,431,616 units of 2025 are about two thirds of 2005.

Stacked area chart of US housing units permitted from 1990 to 2025 by size of building, rising to 2,155,316 in 2005, collapsing to 582,963 in 2009 and recovering to 1,431,616 in 2025, with the five-plus layer much thicker after 2010. — US housing permits by size of building, 1990-2025.

Stacking the 36 files

1. Units by size of building, every year

kinds = {"1 unit": "est1_units", "2-4 units": ["est2_units", "est34_units"], "5+ units": "est5_units"}
ts = pd.DataFrame({k: years[v if isinstance(v, list) else [v]].sum(axis=1).groupby(years.year).sum() for k, v in kinds.items()})
ts["5+ share"] = (ts["5+ units"] / ts.sum(axis=1)).round(3)
print(ts.loc[[1990, 1995, 2000, 2005, 2009, 2015, 2020, 2022, 2025]].to_string())

       1 unit  2-4 units  5+ units  5+ share
year                                        
1990   801071      54576    262904     0.235
1995   997829      63764    271549     0.204
2000  1212124      65177    335057     0.208
2005  1681986      83996    389334     0.181
2009   441148      20690    121125     0.208
2015   697372      32121    454678     0.384
2020   979360      47242    444539     0.302
2022   973851      55215    651302     0.388
2025   911903      55741    463972     0.324

The recovery is almost all apartments. Single-family permits reached 1,681,986 in 2005 and were 911,903 in 2025; buildings of five or more units went from 389,334 to 463,972, and their share of all units from 18.1% to 32.4%, after a high of 38.8% in 2022.

Two line charts: the share of new units in buildings of five or more rising from about 20% before 2008 to over 30% after 2012, and the regional shares with the South rising past half while the other regions fall. — Apartments and the South took a growing share of new housing.

2. The bust, measured

fall = 1 - ts.loc[2009] / ts.loc[2005]
back = ts.loc[2025] / ts.loc[2005]
print(f"2005 to 2009, fall: single {fall['1 unit']:.1%}, 5+ {fall['5+ units']:.1%}")
print(f"2025 as a share of 2005: single {back['1 unit']:.1%}, 5+ {back['5+ units']:.1%}")

2005 to 2009, fall: single 73.8%, 5+ 68.9%
2025 as a share of 2005: single 54.2%, 5+ 119.2%

Both kinds collapsed by about the same amount between 2005 and 2009: single-family by 73.8%, five-plus by 68.9%. But single-family building in 2025 is 54.2% of its 2005 level, while five-plus building is 119.2%, above the peak.

3. Where the building moved

REG = {1: "Northeast", 2: "Midwest", 3: "South", 4: "West"}
reg = years.groupby(["year", "region"]).units.sum().unstack().rename(columns=REG)
share = reg.div(reg.sum(axis=1), axis=0)
print(share.loc[[1990, 2005, 2025]].round(3).to_string())

region  Northeast  Midwest  South   West
year                                    
1990        0.114    0.210  0.386  0.291
2005        0.095    0.164  0.482  0.259
2025        0.089    0.147  0.539  0.225

The South's share of all units rose from 38.6% in 1990 to 48.2% in 2005 and 53.9% in 2025. Every other region lost share, the West most since 2005, from 25.9% to 22.5%. region is the Census region code in the file (1 Northeast, 2 Midwest, 3 South, 4 West).

States that recovered and states that did not

The data used on this page

The US Census Bureau's Building Permits Survey county files (annual 1990–2025 and year-to-date 2025–2026), its Northeast place files for 2020–2025, the Vintage 2025 county population estimates and the 2024 cartographic boundary counties, all public domain. They were downloaded on 16 October 2026 and saved on this site in one archive, because Census revises recent months as late reports arrive.

from pathlib import Path
import zipfile

import geopandas as gpd
import numpy as np
import pandas as pd
import requests

Path("data").mkdir(exist_ok=True)
# US Census Bureau Building Permits Survey: annual county files co1990a.txt-co2025a.txt, year-to-date county files
# co2501y.txt-co2512y.txt and co2601y.txt-co2608y.txt, Northeast place files ne2020a.txt-ne2025a.txt; the Vintage 2025
# county population estimates; 2024 cartographic boundary counties. Downloaded 16 October 2026, saved here unchanged.
# Live: https://www2.census.gov/econ/bps/County/co2025a.txt (and the same folder for the others),
# https://www2.census.gov/econ/bps/Place/Northeast%20Region/ne2025a.txt,
# https://www2.census.gov/programs-surveys/popest/datasets/2020-2025/counties/totals/co-est2025-alldata.csv,
# https://www2.census.gov/geo/tiger/GENZ2024/shp/cb_2024_us_county_20m.zip
ARCHIVE = "census-building-permits-2026-10-16.zip"
if not Path(f"data/{ARCHIVE}").exists():
    Path(f"data/{ARCHIVE}").write_bytes(requests.get(f"https://data.spatialworkflow.io/{ARCHIVE}", timeout=600).content)
Z = zipfile.ZipFile(f"data/{ARCHIVE}")

SIZES = ["1", "2", "34", "5"]  # buildings with 1 unit, 2 units, 3-4 units, 5 or more units
COLS = (["survey", "state", "county", "region", "division", "name"]
        + [f"{k}{s}_{m}" for k in ("est", "rep") for s in SIZES for m in ("bldgs", "units", "value")])


def permits(name):
    """One county file: two header rows and a blank line, then a row per county. est = estimate with
    imputation for places that did not report, rep = reported only."""
    d = pd.read_csv(Z.open(name), skiprows=3, header=None, names=COLS, skipinitialspace=True,
                    dtype={"survey": str, "state": str, "county": str, "name": str})
    d["survey"], d["name"] = d.survey.str.strip(), d.name.str.strip()
    d["state"], d["county"] = d.state.str.strip().str.zfill(2), d.county.str.strip().str.zfill(3)
    return d.assign(GEOID=d.state + d.county,
                    units=d[[f"est{s}_units" for s in SIZES]].sum(axis=1),
                    reported=d[[f"rep{s}_units" for s in SIZES]].sum(axis=1),
                    value=d[[f"est{s}_value" for s in SIZES]].sum(axis=1))


pop = pd.read_csv(Z.open("co-est2025-alldata.csv"), encoding="latin-1", dtype={"STATE": str, "COUNTY": str})
pop = pop[pop.SUMLEV == 50].assign(GEOID=lambda p: p.STATE + p.COUNTY)
Path("data/cb_2024_us_county_20m.zip").write_bytes(Z.read("cb_2024_us_county_20m.zip"))
counties = gpd.read_file("data/cb_2024_us_county_20m.zip")
counties = counties[~counties.STUSPS.isin(["AK", "HI", "PR"])].to_crs(5070)
y25 = permits("co2025a.txt")
print(f"2025 county file: {len(y25):,} rows, {y25.units.sum():,} units permitted; {len(pop):,} counties in the estimates")

2025 county file: 3,029 rows, 1,431,616 units permitted; 3,144 counties in the estimates

Example 1: 2025 against the 2005 peak, by state

names = pop.drop_duplicates("STATE").set_index("STATE").STNAME
st = years[years.year.isin([2005, 2025])].groupby(["state", "year"]).units.sum().unstack()
st["ratio"] = st[2025] / st[2005]
st.index = st.index.map(names)
st = st.dropna().sort_values("ratio")
print("lowest:", ", ".join(f"{k} {v:.2f}" for k, v in st.ratio.head(5).items()))
print("highest:", ", ".join(f"{k} {v:.2f}" for k, v in st.ratio.tail(5).items()))

lowest: Illinois 0.28, Alaska 0.29, Hawaii 0.38, Nevada 0.39, Maryland 0.44
highest: South Dakota 0.93, Utah 0.96, Texas 1.00, Nebraska 1.07, Montana 1.09

Montana, Nebraska and Texas permitted as much in 2025 as in 2005 or more (ratios of 1.00 and above), and Utah and South Dakota nearly so. Illinois and Alaska built under 30% of their 2005 numbers, Nevada and Hawaii under two fifths.

Bar chart of each state's 2025 housing permits as a percentage of its 2005 total, from Illinois and Alaska below 30% to Texas, Nebraska and Montana at or above 100%. — 2025 permits against the 2005 peak by state: only Texas, Nebraska and Montana are back.

Example 2: the counties behind the peak

c = years[years.year.isin([2005, 2025])].pivot_table(index="GEOID", columns="year", values="units", aggfunc="sum")
c = c.join(years[years.year == 2025].set_index("GEOID").name)
print(c.nlargest(5, 2005)[["name", 2005, 2025]].to_string())
print(f"counties that permitted more in 2025 than in 2005: {(c[2025] > c[2005]).sum():,} of {c[[2005, 2025]].notna().all(axis=1).sum():,}")

                   name     2005     2025
GEOID                                    
04013   Maricopa County  50823.0  31651.0
48201     Harris County  41506.0  30909.0
32003      Clark County  39237.0  14466.0
06065  Riverside County  34373.0   9697.0
12071        Lee County  29330.0  13547.0
counties that permitted more in 2025 than in 2005: 758 of 2,988

The counties that led the boom have not returned to it: Clark County, Nevada (Las Vegas) permitted 39,237 units in 2005 and 14,466 in 2025, Riverside County, California 34,373 and 9,697. Only 758 of 2,988 counties with rows in both years permitted more in 2025.

Internal links

Read the permit files — the reader used here
This year's permits month by month — the series after 2025
The apartment share of new housing — the 5+ share by county
Reported against imputed units — why 2003 behaves differently

FAQ

How many housing units were permitted in the US in 2025?

1,431,616 units in the Census Bureau's 2025 county file: 911,903 single-family and 463,972 in buildings of five or more units. That is about two thirds of the 2005 peak of 2,155,316.