Methodology

Where Our Data Comes From

HouseBuildCalc Permit Cost Index — our proprietary dataset aggregates real building permit records from municipal and county open-data portals across the United States. As of our latest refresh, the index covers 876,059 permits across 1,772 ZIP codes. We use this data as a relative local cost index (see below), not as a direct dollar-for-dollar cost figure.

U.S. Census Bureau Survey of Construction — provides national benchmark data on residential construction costs and characteristics.

BLS Producer Price Index — Bureau of Labor Statistics data on construction material and input costs, used to track cost trends over time. We pull series WPU081 (lumber and wood products), WPU0811 (softwood lumber), WPU133 (concrete products), WPU1017 (steel mill products), WPU137 (gypsum products), WPU1073 (copper and brass mill shapes), WPU0638 (plastic construction products), and WPU057303 (asphalt) directly from the BLS Public Data API. The raw annual index data we publish is available at /data/ppi.json.

NAHB Cost of Constructing a Home (published summaries) — the National Association of Home Builders' periodic survey of builder-reported construction costs, used to validate our quality-tier benchmarks.

BLS Occupational Employment and Wage Statistics (OEWS) — state-level median hourly wages for construction trades, pulled directly from the BLS Public Data API (May 2025 vintage). We compute a BLS Trade Wage Index per state as that state's median hourly wage for construction & extraction occupations (SOC 47-0000) divided by the national median. This measured wage index is published alongside — but is not identical to — the site's modeled Labor Index used in the cost formula above; the two are derived independently and may diverge. The raw data we publish is available at /data/labor-wages.json.

Zillow Home Value Index (ZHVI) — used on city pages for the "Build vs. Buy" comparison, showing the median existing-home value alongside our computed new-construction estimate. We use the single-family, mid-tier (33rd-67th percentile), smoothed, seasonally-adjusted city-level series, retrieved July 2026 (latest month: June 2026). Data provided by Zillow (zillow.com/research/data/); the filtered dataset we publish, covering only the cities on this site, is available at /data/zhvi-cities.json. Note that ZHVI reflects the value of existing homes, which includes land and an existing structure, while our construction estimates cover new-build construction costs only and exclude land — the two figures are not directly comparable without adjusting for local land and site costs.

Recent Permit Activity (city pages) — for a subset of major cities (Austin, Chicago, Los Angeles, New York City, San Francisco, and Seattle), we pull live 12-month permit counts directly from each city's own open-data portal (Socrata SODA APIs) — Austin, Chicago, Los Angeles, San Francisco, and Seattle's municipal portals, and New York City's DOB NOW system. This is a separate, display-only dataset from the Permit Cost Index above and does not feed the cost calculator. The exact date field, filters, and residential-permit definitions vary by portal since each city structures its permit data differently; the full set of filters used per city is documented in the dataset itself at /data/permit-activity.json. Where shown, median declared valuation figures carry the same caveat as our Permit Cost Index above: permit-declared costs typically understate final construction cost, since contractors often declare conservatively to reduce fees. A few large cities (Nashville, Dallas) were evaluated but excluded from this refresh because their portals either don't expose a usable Socrata API or their permit data was found to be stale (see the dataset's meta.skipped list for details).

FHFA Land Price Data — state-level residential land value and land-share-of-property-value reference figures shown on state pages come from Davis, Larson, Oliner, and Shui, "The Price of Residential Land for Counties, ZIP Codes, and Census Tracts in the United States," FHFA Working Paper 19-01, via the FHFA-published file "Land Prices — 2024 June" (Version 4.0), sheet "Cross-Section States." The land value figure is the average "as-is" land value per acre for single-family parcels in the state; land share is the aggregate as-is land value divided by aggregate as-is property value. These are pooled cross-sectional estimates standardized to a 2015 base year (the underlying sample pools data from 2012–2022) — they are not current market prices and can lag today's land markets by a decade or more in fast-moving areas. We aggregated nothing ourselves for the state figures; FHFA publishes them pre-aggregated from county-level data, weighted by 2013–2017 ACS single-family housing stock. Retrieved July 2026. The full state-level dataset we publish is available at /data/land-prices.json.

Census Building Permits Survey — 35-year history — the long-run permit chart on our Trends page is built from the Census Bureau's county-level Building Permits Survey annual files for 1990 through 2025 (co{YYYY}a.txt). Our script sums each county's 1-unit (single-family) permit counts and declared valuations into national and state annual totals. We confirmed that valuations are reported in actual dollars in every vintage by tracking a single county's average valuation per unit across the full span and checking for a smooth, discontinuity-free trend. All 36 years parsed with no gaps. National figures are the sum of all 50 states plus D.C. (Puerto Rico is not included in these county files). The series reproduces the documented housing cycle — a 2005 peak of 1,681,986 single-family permits, a 2011 trough of 418,498, and 911,903 in 2025. Valuations are nominal dollars, not inflation-adjusted. Raw data: /data/permit-history.json.

Census Building Permits Survey — city (place) level — the "New Home Permits" card on city pages uses the Census place-level annual file for 2025, matched to our city list by normalized place name and state. Census place names carry governmental-type suffixes ("Austin city", "Nashville-Davidson metropolitan government"), which we strip before matching while preserving genuinely distinct entities — a city and a same-named charter township are not merged. Of our 193 cities, 181 matched. The 12 unmatched are intentionally left blank rather than approximated: several are permitted at the county level and appear only as an "Unincorporated Area" row (Charlotte, Memphis, Louisville, Colorado Springs, Augusta, South Bend, Lafayette), three Hawaii cities are reported only by county, Fort Wayne is absent this year, and New York City is tracked as five separate boroughs rather than one place. Raw data, including the full unmatched list with reasons: /data/city-permits.json.

Census Survey of Construction — Characteristics of New Housing — our New Home Data page draws on ten Census custom tables from the 2025 survey year, covering square footage, foundation type, bedrooms, bathrooms, heating system, air conditioning, exterior wall material, stories, contract price per square foot, and sale price. Each workbook is parsed directly from census.gov; withheld or near-zero cells are left null rather than filled in. Three distinct universes appear in these tables and we keep them separate rather than blending them: physical characteristics describe houses completed; contract price per square foot describes contractor-built houses started and excludes the value of the improved lot; sale price describes spec houses sold and includes land. Raw data: /data/new-home-characteristics.json.

Permit seasonality (monthly) — the seasonality profile on our Trends page comes from 60 Census BPS State/ "current month" files spanning 2021–2025, averaged into a calendar-month share-of-year profile so no single year's anomalies dominate. We verified these are true monthly counts rather than cumulative year-to-date figures by cross-checking against the parallel YTD file series — a distinction that would otherwise manufacture a false "permits rise all year" pattern. Note that this file family reports valuations in thousands of dollars, unlike the county annual files, which report actual dollars; we normalize to actual dollars. Raw data: /data/permit-seasonality.json, with metro-level permits at /data/metro-permits.json.

BLS OEWS metro wages — city pages show construction-trades wages at the metro (CBSA) level, matched to each city using the Census Bureau's official CBSA delineation file rather than an assumed mapping. Of 193 cities, 182 have a confirmed metro; the remaining 11 sit in micropolitan areas where BLS does not publish estimates and are omitted rather than back-filled with a state figure. Wages reflect the entire metro labor market, not city limits, and are labeled with the official metro name for that reason. A subset of major metros also carries carpenter, electrician, and plumber rates. Vintage May 2025. Raw data: /data/metro-wages.json.

DOE residential energy codes — the energy-code line on state pages comes from the U.S. Department of Energy's Building Energy Codes Program, taken from DOE's published state-level workbook (the dataset behind their adoption-status map), covering all 50 states and D.C. We report DOE's own current-code label and effective date, including states with their own standards rather than an IECC edition (California, Oregon, Washington) and the nine home-rule states with no single statewide code — rather than normalizing everything into a tidier claim than the source supports. Energy code is one code among several; local jurisdictions amend and enforce their own structural, electrical, and plumbing requirements, so the general guidance to confirm with your local building department still applies. Raw data: /data/energy-codes.json.

Mortgage rate — the 30-year fixed rate used in payment estimates is a build-time snapshot of Freddie Mac's Primary Mortgage Market Survey, pulled from FRED's public MORTGAGE30US series. Because this is a static site, no rate is fetched in your browser: the figure shown is the one captured at the last rebuild, and every rate on the site is displayed with its PMMS observation date so it is never presented as more current than it is. Raw data, including 52 weeks of history: /data/mortgage-rate.json.

How Estimates Are Computed

Every estimate on HouseBuildCalc starts from the same formula:

Total cost = Square footage × Base cost per sq ft [quality tier] × Style multiplier × State multiplier

We calibrate three national base rates per square foot: $139/sq ft for basic (builder-grade materials and standard finishes), $188/sq ft for mid-range (the national median for custom residential construction), and $256/sq ft for premium (high-end custom finishes throughout). These rates are set from Census Bureau and NAHB benchmark data.

How those rates compare to the federal benchmark. The Census Bureau publishes its own measure of what new homes cost to build: contract price per square foot for contractor-built single-family houses, which — like our estimates — excludes the value of the lot, making it a like-for-like comparison. For the 2025 survey year, the Census median was $171.57/sq ft and the average was $198.33/sq ft. Our modeled median tier of $188/sq ft sits between those two figures, and the federal median falls between our basic and mid-range tiers. We publish this comparison rather than only our own numbers so you can judge the model against an independent source. Regional Census figures for 2025: Midwest $194.09, Northeast $179.30, West $176.87, South $157.30 per square foot.

A style multiplier then adjusts for architectural cost differences — for example, a two-story home typically costs less per square foot than a highly detailed modern design because it has less roof and foundation area relative to living space.

Finally, a state (or ZIP-level) multiplier calibrates the estimate to local market conditions. Where we have permit data for a ZIP code, we compute the ratio of that market's permit-declared median cost to the national permit median and apply it as a relative multiplier on top of our calibrated base rates. Permit-declared values typically understate final construction cost — contractors often declare conservatively to reduce permit fees — so we never use permit values as a direct price. Instead we treat them purely as a relative index: a ZIP with permit values running 2× the national permit median is treated as roughly a 2× cost market, applied on top of our nationally calibrated base rates.

Update Cadence

Our underlying data is reviewed quarterly. The current dataset is dated June 2026. When base rates, state multipliers, or the permit index are refreshed, we update this page and the site's structured data accordingly.

Limitations

• Estimates carry a typical variance band of ±12–15% around the figures shown. That band is no longer a round number we picked: we measured the real spread against 2,706 contractor-built house records in the Census Survey of Construction 2022 microdata. For a like-for-like build — comparable quality level, same census division — the federal interquartile spread is -8% to +8.7%, so our band sits slightly wider than the middle half of comparable homes. Across the whole market, ignoring quality and geography, the spread is far larger: $110.34 to $197.5 per square foot for the middle half. See the full analysis.
• Our model applies a flat rate per square foot — cost scales linearly with floor area. We tested that against the same federal microdata: median contract price per square foot varies by only 7 percentage points across every size bracket from 740 to 8,400 sq ft, with no consistent direction, so a flat rate is defensible over that range. Below 740 sq ft we are extrapolating past the data, and detached accessory dwelling units carry site-access and utility-connection costs this survey universe does not capture.
• All figures cover construction costs only — land purchase, site prep, and financing costs are excluded. Where shown, state-level land value reference figures are FHFA research estimates standardized to a 2015 base year, not current market prices — use them only as a rough order-of-magnitude gauge.
• Permit-declared values routinely understate final, as-built construction cost; we use them as a relative market index, not an absolute price.
• Local jurisdictions vary significantly in permitting requirements, labor availability, and code complexity, which can shift actual costs beyond our modeled range.
• These numbers are for planning purposes only. Always get 2–3 bids from licensed general contractors in your area before committing to a budget.

Corrections

If you believe any figure on HouseBuildCalc is inaccurate or out of date, please contact us — we review corrections as part of our quarterly data refresh.