by Calculated Risk on 8/02/2012 10:31:00 AM

It is difficult to find good and timely data on the number of household formations in the US, and also for the number of excess vacant housing units. The decennial Census is probably the best measure (and also the ACS), but those two estimates aren't consistent (the Census Bureau is looking into the reasons why). Another Census Bureau survey, the Housing Vacancies and Homeownership (HVS) is clearly flawed. The HVS indicates that the number of occupied households increase by 809 thousand over the last year - and that seems too low.

Jed Kolko, chief economist at Trulia, has been looking at Postal Service data.

From Jed Kolko at Trulia: Housing Glut or Housing Shortage? America’s Got Both

With this post, we present a new measure of vacancies, based on U.S. Postal Service (USPS) monthly data on the number of addresses that are and are not receiving mail. ... Here’s what we found.

Nationally, the number of occupied housing units – that is, those receiving mail – rose by 970,000 in the last year, from mid-July 2011 to mid-July 2012. Over the same period, the total number of housing units – those that could receive mail – rose by 760,000. The difference – 210,000 – is the reduction in the number of vacant units. That’s a 5% drop in the number of vacant units nationally. As a percentage of all units, the vacancy rate declined from 3.6% one year ago to 3.4% now.

In fact, vacancies have declined in 90 of the 100 largest metros.

There is much more in Kolko's post.

This data suggests close to 1 million household formations over the last year. I have less confidence in the count of housing units, since total completions only increased by a little more than 600 thousand last year (single family, multi-family and manufactured homes) - and there were also some demolitions.

Clearly the vacancy rate is falling - and the number of household formations exceeds the number of housing units added to the housing stock.