Value of Construction Put in Place (VIP) — US Construction Spending Data

The rows above are lifted from the current monthly release tables and the historical annual workbooks, exactly as they land in a delivery:

type_of_construction: Total Construction   jun_2026p: 2166539   may_2026r: 2168496
type_of_construction: Nonresidential       jun_2026p: 1277174   may_2026r: 1276048
type_of_construction: Residential          jun_2026p: 889365    may_2026r: 892448
unit: millions of dollars, seasonally adjusted annual rate

year: 2025   series: Total Private Construction   value_millions: 1707180

Three things to notice. The suffixes do real work - p marks a preliminary estimate, r a revised one - so a trend you compute on the latest column is provisional until the revisions settle. The hierarchy is one tree: Total Construction resolves into Residential and Nonresidential, and Nonresidential resolves again into lodging, office, commercial, health care, educational, religious, public safety and the communication/power/highway residual. And the same category spine repeats across sector cuts - private, public, federal, state and local - which is what makes a single join key viable across the whole program.

Get a sample of this dataset and we route the category cuts, states and year range you name, the same day.

Nine documented fields cover the release tables and the historical workbooks; everything else folds out on request. Definitions below are verified against the publisher's own technical documentation.

Geography - United States national totals as the headline series, with Census region and division breakouts and all fifty states in the historical workbooks. A monthly subnational experimental series extends coverage for private manufacturing construction.

Temporal - the current release carries the most recent months, revised backward as estimates mature; historical workbooks hold multi-decade annual and monthly series, with geographic detail published through 2025 per the August 2026 announcements. Those same announcements added new Length of Time data for Private Nonresidential, Private Multifamily, and State and Local construction.

Granularity - monthly and annual dollar estimates by sector (private, public, federal, state and local) and by detailed structure type. Seasonally adjusted values exist specifically so month-to-month comparisons don't lie; not-seasonally-adjusted values are the ones that sum to a calendar year.

More on the industry this serves lives on our construction engineering data hub.

API, files, or your warehouse. Daily, weekly, or hourly.

Name the fields, the geography cut and the cadence when you request the sample; the sample ships first either way, with the full field dictionary attached so your engineers can validate before anything recurring starts.

• Market sizing for client decks - consultants size the US construction market in dollars by private versus public sector and structure type, then slice the same tree to a state or division for regional chapters. The workflow is written up on our market researchers use cases page.
• Macro factor backtesting - investors and quants treat the monthly construction-spending print as a demand-side macro factor, with enough history behind each series to survive a proper backtest; see our investors use cases.
• Construction-demand modeling - data scientists feed the monthly dollar series, split by sector and detailed structure type, into models that predict materials demand, equipment utilization and labor needs downstream.
• Citable official figures - journalists and academics cite the government's own dollar numbers on infrastructure investment rather than a think tank's reconstruction of them.
• Territory planning - sales teams set contractor territory quotas against addressable construction spend by state and sector before the fiscal year locks.
• Competitive framing - product and strategy teams frame a rival's growth against segment-level construction spend trends instead of anecdote.

Market researchers and consultants, investors and quant researchers, and journalists, academics and students all rate this their top relevance: one group sizes markets in dollars, one backtests the monthly print, one cites it. Data scientists and ML engineers rate it close behind for construction-demand modeling - budgeting a parsing step is the only tax, since the native distribution leans toward spreadsheet formats rather than CSV. Sales and growth teams and competitive intelligence teams use it as the denominator layer under account lists, sizing territories by where the dollars actually flow.

It is the dollar anchor; most siblings count things. New Residential Construction, its sibling release from the same statistical agency, counts housing units permitted, started and completed - pair them when you need units and dollars together. Chicago Building Permits runs about 845,000 permit rows since 2006 for city-level pipeline work, and NYC DOB Job Application Filings exposes roughly 2.72 million job applications across 95 columns - both go far deeper than any national series on geography, neither prices the work. For cross-country panels, Eurostat Construction Statistics & Building Permits harmonizes EU permits and output indices back decades, and Building Construction Investment by Type of Structure supplies the Canadian quarterly equivalent. Where this dataset stands alone is combining dollar valuation, public-sector inclusion and state-level depth in one hierarchical spine - the head-to-head with the Canadian series is worked through on our comparison page.

Three caveats worth having upfront.

First, preliminary means provisional. Each release adds the current month and revises prior ones - a Jun p column will be a Jun r column next month - so freeze your extract dates if a report must reproduce months later.

Second, pick the adjustment to match the question: seasonally adjusted values for month-over-month movement, not-seasonally-adjusted values for calendar-year totals. Mixing them is the most common error we see in samples sent back with questions.

Third, whether every long-run historical workbook reaches the survey's earliest years wasn't exhaustively verified at review - if your analysis depends on a specific deep decade, say so in the sample request and we confirm the exact window before delivery.