by Calculated Risk on 7/22/2011 04:27:00 PM

The current number of excess vacant housing units is a key piece of data for the housing market. Unfortunately available data is inconsistent.

Economist Tom Lawler has been arguing that many analysts are overestimating the vacant supply by using the HVS - and Lawler has been using the 2010 Census data to make his case. See: The “Excess Supply of Housing” War and Census 2010 Demographic Profile: Highlights, Excess Housing Supply Estimate, and Comparison to HVS

Lawler has also pointed out the most commonly used data for the homeownership rate appears incorrect. The Census Bureau agrees: Census Bureau on Homeownership Rate: We've got “Some 'Splainin' to Do”

Today Goldman Sachs lowered their estimate of the excess supply.

While the decennial census data are from the largest sample, we do not believe it is appropriate to ignore the other sources. ...

With the 2010 Census results in hand, we would now say that excess vacancies in the housing market are 1.5 to 3.5 million units—a wide range, reflecting discrepancies in the available data.

Clearly though, the census results suggest the risks to our previous estimate of 3.5 million units are to the downside. ... [A]t the current rate of housing production and with household growth of one million per year, it would take 5.1 years to clear 3.5 million units of excess inventory, but only 2.2 years to clear 1.5 million units of excess inventory.

A range of 2.2 years to 5.1 years to clear the excess inventory? We need better data!

Lawler thinks the excess supply is closer to the low end of that range.

Note: The Census Bureau is looking at the various data sources now, and is expected to provide analysis on the differences soon.