See American Housing Survey (AHS) 2005, pdf published Aug 2006, link
Total number of HHs (homes) estimated was 108,872,000, of which 33,940,000 renter occupied and 74,931,000 owner occupied. Any estimate of the number of mortgages requires examination of the total population of HHs -- not merely "owner occupied" housing. An notice of census.gov collection methods follows.

In prior years, the American Housing Survey-National Sample was conducted biennially in housing units selected from the 1980 Census and new construction universe. Data collection for AHS-N was conducted in odd-numbered years. The last AHS-N was conducted in 2005 between May and September with a sample size of approximately 60,000 housing units.

Also in prior years the American Housing Survey-Metropolitan Sample was conducted biennially in even-numbered years in 41 metropolitan areas on a rotating basis. The last AHS-MS was completed in early October 2004. The 2004 AHS-MS included 13 metropolitan sampling areas (MSAs) each with approximately 4,700 housing units [approx. 61K total sample]. The 2004 sample consisted of returning housing units selected from the 1990 Census and new construction universe.

In early 2006, the Department of Housing and Urban Development (HUD), the sponsoring agency, made the decision to conduct the AHS in odd-numbered years only beginning with the 2007 data collection. We will no longer survey the national survey in odd-numbered years an the metro survey in even-numbered; however, we will collect data for bot the national and metropolitan sample at the same time. The 2007 AHS national sample will consist of approximately 55,000 housing units and the metropolitan sample will include seven MSAs each with approximately 3,000 housing units for a total of approximately 76,000 in sample.
(Census.gov Demographic Survey Abstracts, Jan 2007 , pdf)

See also most recent 2007 AHS National Data, dataset list. The downloads are *.exe, MS-DOS executable only. That excludes my machine.

See also AHS Table 8, "Housing Vacancies and Homeownership"

Doubtless periodic reports by independent researchers such as CalculatedRisk and Case-Shiller who maintain series data on inventory (existing, new) provide a valuable public service.

In "Fed: Household Percent Equity Declines" CR cautions, "31% of HHs do not have a mortgage. So the 50+ million households with mortgages have far less equity than 45.2%."

Oddly enough 33.1% was the percentage of 2005 owner occupied HHs not mortgaged:  24.776M HHs. (See Table 2, "The American Housing Survey and Non-Traditional Mortgage Products," HUD, Sep 2007; HUDUser.org/publications). CR implies the number of mortgages has increased 2.1% at least since 2005. "Occupied means", too, primary residence as distinct from "investment property."

Conversely, 66.9% should be the percentage of 2005 owner-occupied HHs mortgaged, or 50.1M HHs. However the detail of mortgage types for this class evaluates to 64.6% or 48.6M "owner-occupied" HHs, "regular mortgage and/or home-equity mortgage." 90% of these were conventional, fixed loans only. Additionally, AHS did not collect data for 4.5M "owner occupied" properties.

Additionally, 33.94M HHs were renter occupied in 2005. How many of these "investment properties" were mortgaged? The data do not say.

Additionally, 3.7M seasonal, or second, homes are excluded from the total of occupied housing in 2005. How many of these were mortgaged? The data do not say.

Is it reasonable to assume that the universe of mortgaged housing was closer to 90M in 2005 and likely peaked Q4 2007 around 100M? Yes, the lower bound of a conservative estimate of RE secured by "primary mortgages" is 85.9M - 87.9M.

The AHS considers a home to be "mortgaged" if it has one or more of the following: a regular mortgage, a lump-sum home equity loan, or a home equity line of credit. ...The AHS considers a home to have a "primary mortgage" if it has one or both of the following: a regular mortgage or a lump-sum home equity loan. ...The AHS considers a home whose only lien is a home equity line of credit to be "mortgaged" but not to have a primary mortgage. ["The American Housing Survey and Non-Traditional Mortgage Products," p9]

The FRB's Q1 2008 report of US Flow of Funds (Z.1 Release, Sep 2008) does not explain the number of residential properties mortgaged. Two Calculated Risk charts describe declines in mortgage holders' aggregated equity since 1952 and HH real estate assets and mortgage debt  and housing value as percentages of GDP. The relevant FRB tables are as follows (download each table category to view).

Over the period Q4 2005 - Q4 2007 the value of all "home mortgages" increased from $9,383.8T to $11,239.0T, or 19.6%. Although AHS data tell us "occupied" housing increased from about 108M to 110M, or 1.85%, it doesn't tell us how much of the mortgage value appreciation is attributable to "primary mortgage" growth or additional inventory sales such as second homes and vacant "investment" property. Clearly the number of home mortgages retired is insignificant.

OK. That's enough of a critical read to whack NPR "reporting" -- understatement of the ages.


Diversity is the key to economic and political evolution.