HSN1F: US New Home Sales Monthly SAAR Data from FRED (1963–2026)

The US New Home Sales series counts sales of newly built single-family houses, monthly since 1963 at a seasonally adjusted annual rate, from the Census Bureau and HUD. New homes are a small slice of the housing market — historically 10–15% of total sales — but the slice builders and the construction cycle depend on, and the earliest-reading demand gauge: sales are recorded at contract signing, one to two months before existing-home closings. The bubble peak was 1.39 million (July 2005), the post-crash trough 270,000 (February 2011); mid-2026 runs at 580,000.

Dataset: US New Home Sales (1963–2026) · Updated 2026-06-01

Source: US Census Bureau and HUD · FRED series HSN1F

Macro Takeaway

Dataset Overview

Dataset Variables

The CSV and Excel files contain the following columns.

A value of 580 means new homes were selling at a pace of 580,000 per year.

Download the Complete Dataset

The full dataset is available in CSV and Excel formats.

FRED Direct CSV Access

The underlying data is available from FRED under series code HSN1F:

https://fred.stlouisfed.org/graph/fredgraph.csv?id=HSN1F

Direct CSV Access — Eco3min Structured Dataset

https://eco3min.fr/dataset/new-home-sales.csv

This URL returns the complete dataset in CSV format. It can be used directly in pandas, R, curl, or any data tool.

Using the Dataset in Python

import pandas as pd

url = "https://eco3min.fr/dataset/new-home-sales.csv"
df = pd.read_csv(url, parse_dates=["date"])

print(f"Latest pace: {df['new_home_sales'].iloc[-1]:.0f}k homes/year")
print(df.tail())

Using the Dataset in R

library(readr)

url <- "https://eco3min.fr/dataset/new-home-sales.csv"
df <- read_csv(url)

tail(df)
summary(df$new_home_sales)

Both examples load the dataset directly from the URL — no download or API key required.

Methodology

A sale is recorded when a purchase contract is signed or a deposit accepted, for a new single-family house — at any construction stage (not started, under construction, or completed). Multifamily units and existing homes are excluded; existing-home sales are published separately by the National Association of Realtors on a closing basis.

Contract-basis timing makes the series lead existing-home sales by one to two months, at the cost of cancellation blindness: contracts that fall through are not removed retroactively.

This dataset is updated daily (Mon–Sat, 08:00 UTC) via automated pull from the FRED API; new observations appear with each source release.

Data Quality & Provider Notes

• Release latency. ~4 weeks after the reference month.
• Revisions policy. Heavy — three subsequent monthly revisions plus annual benchmarking; initial prints regularly move 5–10%.
• Sampling error. Confidence intervals around ±15% on monthly changes — the widest in the housing bloc; the Census itself flags most monthly moves as insignificant.
• Cancellations. Not netted out; in stress periods (2007–2008, 2022) reported sales overstate completed transactions.
• Known gaps. None; continuous since 1963.

Common Pitfalls When Using New Home Sales

1. Treating it as the housing market. New homes are ~10–15% of transactions; the existing market (NAR data) dominates volumes and can move oppositely — as it did through 2023–2026.
2. Trading the monthly print. ±15% confidence bands and triple revisions make single months unreliable by construction; quarterly averages are the readable unit.
3. Ignoring cancellations in downturns. Contract-based counts overstate demand exactly when it collapses; the 2007 prints looked better than reality for months.
4. Skipping the price-mix effect. Builders responded to the rate shock partly through smaller, cheaper product; volume stability can coexist with a shifting mix that median-price data then misreport as deflation.

Historical Regimes

1963–1990 — The 400–800k cycles. Sales oscillated with rates and demographics through repeated boom-bust swings, never durably escaping the band.

1991–2005 — The long ascent. A fourteen-year climb through the securitization era to the all-time peak of 1.39M (July 2005) — the demand crest of the bubble, two years before the financial crisis.

2006–2011 — The collapse. Sales fell 80% to the series low of 270k (February 2011) — below any level recorded since the series began, and still falling two years after the recession officially ended.

2012–2019 — The half-speed recovery. A slow rebuild to ~700k by 2019 — a decade to recover half the peak, matching the underbuilding era in starts.

2020 — The pandemic spike. Sales surged to 1.03M (October 2020) on record-low mortgage rates and space demand — the strongest pace since 2006, gone within a year.

2021–2026 — Rate shock and the builder workaround. The mortgage repricing cut sales toward 580k by mid-2026, but incentives and rate buydowns kept new-home volumes far more resilient than the frozen existing market — the lock-in era’s signature divergence.

Related Macroeconomic Datasets

New home sales are the demand gauge of the construction pipeline; supply stages and financing sit below.

Macroeconomic Dataset Hub

This dataset is part of the Eco3min macro-financial data repository.

Sources

• US Census Bureau and US Department of Housing and Urban Development — New Residential Sales
• Federal Reserve Bank of St. Louis — FRED series HSN1F

Dataset Reference