Housing starts are the number of new privately owned housing units on which construction began during a period. In the United States the figure is published every month and is one of the most closely followed measures of activity in residential construction. A start is counted when excavation begins for the footings or foundation of a building, and each apartment in a multi-unit building counts as one unit.
The statistic is treated as one of the leading indicators of the economy. Home building is sensitive to interest rates and confidence, tends to turn down before a recession and to recover early, and feeds demand for materials, appliances, furniture and labour. It is published together with building permits, which record authorisations to build and so look a little further ahead, and housing completions.
Publication and measurement
The data appear in the New Residential Construction report, issued jointly by the US Census Bureau and the Department of Housing and Urban Development. The report is released at 8:30 a.m. Eastern time around the middle of the month and covers the previous month. It is a fixed entry in the economic calendar.
Starts are estimated from a sample survey in which field representatives follow a selection of permits and visit sites to find out whether building has begun. The results are broken down in two ways:
- By type of structure. Single-family houses are reported separately from buildings with two to four units and those with five or more. Single-family starts are steadier and are considered the better guide to underlying demand. Multi-family starts swing widely from month to month.
- By region. Figures are given for the Northeast, Midwest, South and West. The South accounts for the largest share.
Annualised rate and worked example
The headline number is a seasonally adjusted annual rate. Building is far busier in summer than in winter, so raw monthly counts are first corrected by seasonal adjustment and then multiplied by twelve to show how many homes would be started in a year if that month's pace continued:
- is the unadjusted number of units started in the month.
- is the seasonal factor for that month, above 1 for busy months and below 1 for slow ones.
If builders actually began 126,500 homes in June and the seasonal factor for June is 1.10, the adjusted monthly figure is and the annual rate is , reported as 1.38 million. If the previous month's rate was 1.42 million, starts fell by percent. The annualized rate magnifies any error in a single month by the same factor of twelve.
How markets read the report
Traders compare the release with the consensus estimate of economists. A result well above expectations points to stronger growth. It tends to push up yields in the bond market and to support the shares of home builders, building-materials suppliers and home-improvement retailers, as well as the prices of lumber and copper. A weak result has the opposite effect. The reaction is usually smaller than for employment or inflation data, because the series is erratic.
Housing is one of the main channels through which monetary policy acts on the economy. When the Federal Reserve raises rates, the cost of a mortgage rises, affordability falls and starts decline within months. Analysts therefore watch the trend in single-family starts and permits for early evidence that policy is biting or easing. Residential investment is also a direct component of gross domestic product, so the monthly figures feed into forecasts of quarterly growth.
Over longer periods the level of starts is judged against household formation and the loss of old homes. Building well above that level for several years suggests oversupply, and building below it suggests a shortage that will support prices and rents.
Historical range
The series has been published since 1959. Starts climbed to an annual rate above two million at the height of the housing boom in early 2006, then collapsed to around half a million in 2009, the lowest on record, as the mortgage market broke down. The recovery took more than a decade, and the long period of low building that followed the crash is widely cited as one cause of the housing shortage of later years.
Limitations
The monthly estimate has a wide margin of sampling error, often larger than the change being reported, so a single month's rise or fall is frequently not statistically significant. Figures are revised in the following two months and again each year. Weather distorts the data: a mild winter brings starts forward and a harsh one delays them, and seasonal factors cannot fully correct for unusual conditions. Analysts therefore prefer a three-month average and give more weight to permits, which are based on a far larger sample and are less affected by weather.
A start is also not a finished home. Labour shortages, delays in materials or financing problems can lengthen the time between start and completion, so the number of homes under construction and the number completed are needed to see how much new supply is reaching the market.