Housing starts are among the most influential collections of data in the world of real estate, but a Washington Post blog has laid bare some troubling flaws with the data’s collection.
Here are the five most notable inconsistencies with the statistic:
- Housing starts data came out last week, and amidst the promising increases (which we, like many other outlets, reported on), was this fact – starts rose 2.6 percent, but they could have fallen by 12.2 percent or risen by 17.4 percent. Why? The Census Bureau’s margin of error was 14.8 percent.
- As the Chief of the Bureau’s Residential Construction Branch, Raemeka Mayo, explained to the Post, the Bureau follows a process to collect its data. First, it gets fresh building permit info from 900 of around 19,000 offices that issue permits.
- Then, after identifying the properties to track, they follow all projects with five or more housing units, but only 1 out of every 50 other projects.
- How they follow the construction is similarly scattered. Once construction begins, the Bureau visits the sites or chats with the builders/property owners, and follow up each month until the project is completed and becomes part of the Bureau’s housing completions total.
- The reported margins of error for the past three months were the following: 15.6 percent in March, 15.9 percent in February and 16.8 percent in January.
The Bureau admits in the fine print, “The Census Bureau does not have sufficient statistical evidence to conclude that the actual change is different from zero,” or, as Mayo, told The Post, “We allow everybody to come up with their own interpretation.”
Housing starts is seen as an important signal for construction activity; the stock market rises and falls on its projections. But what does it say that we put our faith in such a shaky measurement?