Thursday, May 05, 2011

Lawler: Census Releases Demographic Profile of 12 States and DC: Confirms Bias of HVS

by Bill McBride on 5/05/2011 02:47:00 PM

CR Note: The Census Bureau is releasing the demographic profile data for every state this month. Many analysts use the quarterly Census HVS to track the homeownership rate - and to calculate the excess supply of housing units. Important: Estimates using the HVS appear to overstate the excess supply. When all the data is released (by the end of May), we will probably have a better estimate of the excess supply as of April 1, 2010.

From economist Tom Lawler: The Census Bureau released the “Demographic Profile” from Census 2010 for 12 states and DC today, and the release confirmed several of the biases I have noted in the Census’ quarterly Housing Vacancy Survey. As I’ve shown before, the HVS showed higher homeownership rates than either the 2000 decennial Census or the American Community Survey, and that overstatement continued for 2010. Below is a comparison of the homeownership rate from the decennial Census for April 1, 2010 with the HVS average homeownership for the first half of 2010. (The HVS quarterly data are monthly averages). Also shown are the homeownership rates for April 1, 2000 from Census 2000.

Homeownership Rate (%), Census 2010, Census 2000, and Housing Vacancy Survey
 Census 2010, April 1HVS, First Half 2010Census vs. HVSCensus 2000, April 1Census 2010 vs. Census 2000
DC4245.1-3.140.81.2
Florida67.470.3-2.970.1-2.7
Kentucky68.771.3-2.670.8-2.1
Maine71.373.6-2.371.6-0.3
Massachusetts62.364.4-2.161.70.6
Michigan72.174.5-2.473.8-1.7
Mississippi69.673.5-3.972.3-2.7
New Mexico68.568.8-0.370-1.5
North Dakota65.467.4-266.6-1.2
Rhode Island60.763.6-2.9600.7
South Carolina69.374.5-5.272.2-2.9
Tennessee68.271-2.868.20
West Virginia73.479.1-5.775.2-1.8
As the table shows, the HVS shows higher homeownership rates than Census 2010 in DC and all 12 states for which data have been released, with the HVS homeownership rates MATERIALLY higher in DC and 11 of the 12 states. As I’ve noted in past reports, the HVS appears to be a biased sample of the US housing stock, consistently over-estimating the homeownership rate, and overstating significantly both the gross vacancy rate and the rental vacancy rate.

This early look strongly suggests that there are material sampling problems with the HVS that need to be addressed, and suggests that the HVS data probably should not be used to analyze the US housing stock, or estimate the “excess supply” of housing.

CR Note: A key number for the U.S. economy is the excess supply of vacant housing units. The decennial Census probably provides the best estimate, and that is already over a year out of date.