Housing Starts, Permits and Home Sales Data

Housing starts, permits and home sales data measure different events in the housing market. Permits measure how many new units local authorities approved, starts measure how many began construction, and sales measure transactions at a defined stage. Completions show how many units finished construction. For a quant tracking the business cycle, the distinction helps separate planned building, work underway, new supply and buyer activity. U.S. Census Bureau New Residential Construction

Construction coverage is wider than new-home sales by structure type. It still excludes publicly owned housing and HUD-code manufactured homes. The construction series come from the Building Permits Survey and the Survey of Construction, which HUD helps fund. Census construction definitions · Survey of Construction methodology · NAR Existing-Home Sales release notes

A builder can sell a new house from plans before a permit is issued or excavation begins. The same house can therefore appear in the sales series in one month, permits in another, and starts and completions later. An existing home under contract in September and closed in October counts in October's existing-home sales. The two sales series can move in opposite directions because they cover different homes and date sales at different events. Census Survey of Construction definitions · NAR Existing-Home Sales release notes

Permits show a potential pipeline. Starts can draw from permits issued in earlier months, and some authorisations may stay unstarted. Completions can rise while starts fall as builders finish older projects. To estimate how quickly permits turn into starts, follow units by month of authorisation. Dividing this month's total starts by this month's permits mixes groups of homes and can also mix coverage. Census starts rates by authorisation month · Survey of Construction methodology

The Census headline commonly uses a seasonally adjusted annual rate (SAAR). Seasonal adjustment estimates and removes recurring seasonality, such as a typical winter slowdown. An unusually severe winter can still affect the adjusted series. Annualisation then expresses that one adjusted month's pace in yearly units. Census definition of SAAR

Suppose a fictional series has 100 starts in a month after seasonal adjustment. Its headline annual rate is

The 1,200 expresses how many starts there would be over a year at that month's pace. An actual annual total comes from the activity across the year, and a forecast would require assumptions about future months. If a table is in thousands of units at an annual rate, it would show $1.2$ for the same example. Census definition of SAAR

Keep the reference month of the activity separate from the release timestamp and the data vintage. The Census construction and new-home sales reports have separate release dates. A September starts estimate first published in October becomes usable when the release reaches the research system. Census revises preliminary estimates and can update earlier seasonally adjusted values when it recalculates seasonal factors. Census release schedule · Survey of Construction release and revision schedule

For a historical trading decision, use the latest version already available at that decision time. Putting a revised September value into a backtest row dated September gives the strategy information from the future. The SelfMadeQuant lesson Point-in-Time Data shows how to select eligible versions and audit revisions. The preview is public and the full lesson requires membership.