Data Dig : Single-Family Housing Starts March 2026
Anticipation of Warmer Weather and of Mortgage Rate Decline Likely Caused Single-Family Starts to Jump in March 2026, but the Trend May Not Continue
Each month, the United States Census Bureau publishes new housing construction permits, starts, under-construction, and completions data collected by survey. The data series reaches back to 1959. Month-to-month changes may be curious but typically are not instructive. The trends do tell stories. When the data are released, Shovel to Keys goes data digging.
First a brief comparison of March 2026 data to the prior month and prior year. (All data are seasonally adjusted by the USCB.1)
Single-family housing starts climbed to 1,032,000 units from 941,000 in February 2026 and 948,000 in March 2025.
The March 2026 number was the highest since February 2025 when the starts number was 1,098,000.
The twelve-month trailing average in March was 927,000 down from 1,003,000 for the twelve months ending March 2025.
The March 2026 starts number is noticeably above the twelve-month average and nearly mid-way between the peak and trough since January 2000. The jump from February likely is attributable to the warming of the weather (whether or not attributable to the climate crisis) from a harsh winter especially in the South and the declining trend in mortgage rates, which briefly excited home building and sales companies before being predictably interrupted by a spike in mortgage rates caused by the Iran War.
Putting the March data in the context of housing affordability and availability, the chart above suggests one reason, at least, for the purported under-supply of housing about which much has been and is being said.2 The peak and trough give us the historical context of the boom and bust just before and after the Great Financial Crisis that arose roughly between the end of 2006 and early 2008. Note that the count of monthly single-family housing starts since 2008 has been below the pre-GFC starts count for nearly all months, except briefly during the very low interest rate anomaly of the Covid pandemic.
The historical twelve-month trailing average single-family starts data shows that new construction consistently has struggled to reach the heights of the years preceding the GFC. The volume of production of single-family housing has not recovered from the systemic shock of the GFC. The twelve-month trailing average dropped to just above 500,000 in 2009 and remained below the 25-year median until 2020. These data leave little room to question the proposition that any 2026 shortage in housing is the long tail effect of the pain experienced in the housing industry from the GFC. The reasons will be the subject of more than one future Shovel to Keys post.
The April 2026 print of housing starts from the USCB will bear watching for the impact of the Iran War. Mortgage rates followed the 10-year Treasury upward over 500 basis points after the War began. The March data likely does not reflect the impact because starts decisions are made by builders at least thirty days in advance, if not longer.
In our next post, we’ll look at new housing under construction. On May 5, the Census Bureau will release new home sales data for February and March (delayed because of the most recent federal government shutdown). These data collectively provide useful insights into the new home construction industry.
The Census Bureau adjusts data each month to account for the effects of predictable, seasonal patterns, such as weather-caused increases or decreases in home construction. The Bureau explains in its frequently asked questions. The adjustment attempts to reveal data patterns that would be masked by seasonal effects. The seasonally adjusted numbers are annualized.
For example, by the U.S. Chamber of Commerce, the sages in Congress, the housing industry, and Wall Street.