by Calculated Risk on 4/30/2011 06:10:00 PM

Some key numbers for the U.S. economy are: 1) the current number of excess housing units, 2) how many new households are being formed each year, and 3) how many housing units are being added to the housing stock each year (at a record low this year).

Unfortunately reliable data for the first two numbers is unavailable except with a significant lag.

I've used the quarterly Housing Vacancy Survey (HVS), but that is not really designed for this purpose.

Goldman Sachs put out an estimate yesterday of 3.5 million units based on the HVS: "Based on data from the Census Bureau, we estimate that about 3.5 million housing units currently sit vacant, above and beyond normal seasonal and frictional vacancies." They calculated a range of 2.5 to 4.5 million units based on different assumptions.

Recently economist Tom Lawler took a long look at the 2010 Census data, and estimated there were about 2 to 3 million excess vacant housing units as of April 1, 2010. With the record low number of housing units delivered last year, Lawler estimated that as of April 1, 2011 the excess “would be somewhere in the range of 1.45 to 2.45 million units – with the latter almost certainly too high”. With another record low number of units added to the housing stock this year, the excess will be in the 750 thousand to 1.7 million range next April (with the latter “certainly too high"). This suggests the excess supply will be gone sometime between early 2014 and 2016.

Goldman has a higher estimate of excess vacant units, but they also have a higher estimate for household formation (partially because of pent up formation from all those people who doubled up during the recession). The conclude "[W]e think the recovery in single family housing starts will remain very slow. A plausible central scenario would be an increase in starts from about 475,000 units in 2010 to 600,000 by 2012. We do not expect single-family housing starts to return to their historical average rate of about 1 million units until 2015 or later."

In Thoughts on Residential Investment Recovery, I noted: 'My guess is housing starts will return to "normal" in 2015 or 2016.' It is hard to pinpoint an exact date because the data is uncertain. Frustrating!

But here is a little good news on data via Tom Lawler:

The Census Bureau announced that beginning next week it will release “demographic” profiles of 13 states each Thursday in May, with the first round of “states” will be the District of Columbia, Florida, Kentucky, Maine, Massachusetts, Michigan, Mississippi, New Mexico, North Dakota, Rhode Island, South Carolina, Tennessee and West Virginia. For folks interested in the housing markets, the data will include (among other things) data on housing tenure (owners/renters), occupancy/vacancy (include vacancy status – for rent, for sale, seasonal, etc.), age and sex distributions, and household types. These releases will help analysts get a much better “feel” for the overall housing market as of April 1, 2010, and will highlight just how “messed up” the data from the Census Housing Vacancy Survey really is – and why analysts should not use it to try to estimate the “excess supply” of housing.

I think the Goldman estimate is too high, but no one really knows. Hopefully by the end of May we will have a much better estimate for the excess supply as of April 1, 2010.

Earlier:
• Summary for Week ending April 29th