long-term vacant housing in the united states
DESCRIPTION
Presented by Raven Molloy, Federal Reserve Board of Governors, on May 20, 2014, as part of the REALTOR® University Lecture SeriesTRANSCRIPT
Long-Term Vacant Housingin the United States
The analysis and conclusions set forth are those of the author and do not indicate concurrence by other members of the research staff or the Board of Governors.
Raven MolloyFederal Reserve Board of Governors
May 20, 2014
Unintended vacancy reflects an imbalance between supply and demand in the housing market.
Some amount of vacancy is normal as residents change. And some houses are intended to be vacant for parts of the year.
Unintended vacancy reflects underutilization of the housing stock.
When structures stand vacant for a long time, there could be adverse consequences for house prices and neighborhood quality.
In aggregate, the vacancy rate rose sharply during the housing market contraction and has yet to return to pre-crisis levels.
Introduction
Q1
89
1011
1213
1415
1965 1970 1975 1980 1985 1990 1995 2000 2005 2010 2015year
Pe
rce
nt o
f H
ousi
ng U
nits
Gross Vacancy Rate
Aggregate Vacancy Rate
Source. Housing Vacancy Survey.
Q1
34
56
78
910
1965 1970 1975 1980 1985 1990 1995 2000 2005 2010 2015year
Pe
rce
nt o
f H
ousi
ng U
nits
Vacancy Rate Excluding Seasonal and Occasionally Occupied Housing
Aggregate Vacancy Rate
Source. Housing Vacancy Survey.
Goal of this project is to learn more about the vacant stock: What fraction of the aggregate rise in vacancy reflects imbalances
between supply and demand? What types of locations have large imbalances between supply and
demand?
Measure imbalances between supply and demand using units that have been vacant for a long time. Short periods of vacancy could be due to normal turnover of residents. Long periods of vacancy are more likely to be associated with adverse
effects on the neighborhood.
Introduction
Aggregate statistics “Long-term vacancy” = units that have been vacant for an unusually long
period of time What fraction of the rise in aggregate vacancy is due to long-term
vacancy?
Neighborhood-level statistics Distribution of long-term vacancy across neighborhoods. Characteristics of neighborhoods with large shares of long-term vacant
housing.
Agenda
American Housing Survey Biennial panel dataset of housing units 1985 to 2011 Categorizes 6 reasons for vacancy:
For sale For rent Seasonal Usual Residence Elsewhere (URE) Rented/sold but awaiting occupancy Other
Length of vacancy Detailed property characteristics
Exclude seasonal, Usual-Residence-Elsewhere and rented/sold.
Aggregate statistics
Length of Vacancy, 1985 to 2003For Rent For Sale Other
Percent of housing units vacant:
Less than 1 month 35.7 13.3 8.8
1 to 2 months 13.3 6.8 4.4
2 to 6 months 26.7 27.9 16.5
6 to 12 months 8.8 16.9 10.9
1 to 2 years 5.6 12.1 13.4
2 years or more 7.4 14.1 42.7
Never occupied 2.6 8.9 3.2
Percent of all vacant housing units 22.8 8.7 16.3
Aggregate statistics
77% of units for sale vacant 12 months or less.
76% of units for rent vacant 6 months or less.
41% of “other” units vacant 12 months or less.
Reasons for “Other” Vacancy in the 2012 Housing Vacancy Survey
Percent of OtherVacant Stock
Personal/family reasons 19.7
Foreclosure 11.7
Legal proceedings 5.9
Possibly abandoned/to be demolished/condemned 6.7
Intended for year-round occupancy but vacant 6+ months 5.8
Being prepared to rent/sell 6.6
Needs repairs 13.9
Being repaired 9.3
Held for furniture storage 7.4
Held for Specific Use 1.0
Other/Don’t know 11.3
Presence of Negative Physical Characteristics byLength of Vacancy for “Other” Vacant Units, 2001-2011
At least 1 At least 3 At least 5Fraction of housing units vacant:
Less than 3 months 30.8 11.1 6.1
4 to 11 months 41.7 20.9 11.5
1 to 2 years 55.3 29.0 16.4
2 to 3 years 62.7 35.4 21.8
3 to 5 years 74.6 48.1 32.7
5 years or more 77.9 58.3 42.6
Aggregate statistics Owners that intend to leave a property vacant are more likely to maintain it, so units left
vacant unintentionally are more likely to be in a state of disrepair.
Presence of 19 physical characteristics associated with deterioration: Cracks in floors, walls or ceilings Broken windows Holes in roof Incomplete plumbing No heating equipment
Long-term Vacant Units Over Time
Aggregate statistics Define long-term vacancy as:
Units for rent for more than 6 months Units for sale for more than 1 year Other vacant units that have been vacant for more than 2 years
0.0
05.0
1.0
15.0
2
19851987198919911993199519971999200120032005200720092011
Fra
ctio
n o
f A
ll H
ousi
ng U
nits
Other Vacant > 24 mo. For Sale > 12 mo. For Rent > 6 mo.
Aggregate statistics Vacant stock rose from 12.9 million units in 2001 to 17.5 million units
in 2011 Increase = 4.6 million units, of which:
Seasonal/occasionally occupied = 2.0 million units Long-term vacant = 1.1 million units For rent < 6 months = 0.7 million units “Other” vacant < 2 years = 0.6 million units For sale < 2 years = 0.2 million units
Geographic Distribution of Long-Term Vacancy Long-term vacancy is not a large fraction of the housing stock and
accounted for only ¼ of the rise in vacancy since 2001.
But implications for some neighborhoods could be substantial if the stock is concentrated in relatively few locations.
Neighborhood-level data from US Postal Service Postal carriers report an address as vacant if mail has not been collected for at
least 90 days. For each Census tract, HUD reports total number of vacant residential addresses
by length of vacancy Tract = geographic area with population size between 1,200 and 8,000; “optimum” size of 4,000.
Define long-term vacancy as vacant 2+ years Sample restrictions:
Exclude tracts with a large share of seasonal housing (2010 Census) Exclude tracts with a number of housing units or a vacancy rate in 2010 that are not close to
estimates in the 2010 Census.
Geographic Distribution of Long-Term Vacancy Final USPS sample is 52,000 tracts = roughly ¾ of tracts and housing
units in the US In the USPS data, 4 percent of addresses in 2013 had been vacant for
at least 2 years. This fraction is higher than the measure of long-term vacancy in the American Housing Survey (AHS) because: Some vacant addresses may be seasonal AHS excludes housing units that have fallen into a state of severe disrepair.
Geographic Distribution of Long-Term Vacancy Most metropolitan areas have a stock of long-term vacant units that
is roughly proportional to the stock of occupied housing units.Lo
ng-T
erm
Vac
ant
Add
ress
es
(Log
Sca
le)
Occupied Addresses (Log Scale)
10000 100000 1.0e+061.0e+06 5.0e+06
500
1000
5000
10000
25000
50000
100000
50000
100000
300000
Abilene,
Akron, O
Albany, Albany-S
Albuquer
Alexandr
Allentow
Altoona,
Amarillo
Ames, IA
AndersonAnderson
Ann ArboAnniston
Appleton
Athens-C
Atlanta-
Auburn-O
Augusta- Austin-R
Bakersfi
Baltimor
Baton Ro
Battle C
Bay City
Beaumont
Billings
Binghamt
Birmingh
Bismarck
Blacksbu
BloomingBlooming
Boise Ci
Boston-C
Boulder,
Bowling
Bremerto Bridgepo
Buffalo-
BurlingtCanton-M
Cape Gir
Carson C
Casper,
Cedar Ra
ChampaigCharlest
Charlest
Charlott
Chattano
Cheyenne
Chicago-
Chico, C
Cincinna
Clarksvi
Clevelan
Clevelan
College
Colorado
Columbia
Columbia
Columbus
Columbus
Columbus
Corpus CCorvalli
Cumberla
Dallas-F
Dalton,
Danville
Danville
Davenpor
Dayton,
Decatur,
Decatur,
Denver-A
Des Moin
Detroit-
Dothan, Dover, D
Dubuque,
Durham-C
Eau Clai
Elizabet
Elkhart-
Elmira,
El Paso,Erie, PA
Eugene-S
Evansvil
Fairbank
Fargo, NFarmingt
Fayettev
FayettevFlint, M
Florence
Florence
Fond du
Fort Smi
Fort Way
Fresno, Gadsden,
Gainesvi
GainesviGoldsbor
Grand Fo
Grand Ju
Grand Ra
Great Fa
Greeley,
Greensbo
Greenvil
Greenvil
Gulfport
Hagersto
Hanford-
HarrisbuHartford
Hattiesb
Hickory-
Hinesvil
Holland-Honolulu
Houma-Ba
Houston-
Huntingt Huntsvil
Idaho Fa
Indianap
Iowa Cit
Ithaca,
Jackson,
Jackson,
Jackson,
Jacksonv
Janesvil
Jefferso
Johnson
JohnstowJonesbor
Joplin,
KalamazoKankakee
Kansas C
Kennewic
Killeen-Kingspor
Knoxvill
Kokomo,
La Cross
Lafayett
LafayettLake Cha
LancasteLansing-
Laredo, Las Cruc
Las Vega
Lawrence
Lawton, Lebanon,
Lewiston
Lewiston
Lexingto
Lima, OH
Lincoln,
Little R
Logan, U
LongviewLongview
Los Ange
Louisvil
Lubbock,
Lynchbur
Macon, G
Madison,
ManchestManhatta
Mankato-
Mansfiel
Medford,
Memphis,
Merced,
Midland,
Milwauke
Minneapo
Missoula
Mobile,
Modesto,
Monroe,
Monroe,
Montgome
MorgantoMorristo
Muncie,
Muskegon
Nashvill
New HaveNew Orle
New York
Ocala, F
Odessa, Ogden-Cl
Oklahoma
Olympia,
Omaha-Co
Oshkosh-Owensbor
Oxnard-T
Parkersb
Pascagou
Pensacol
Peoria,
Philadel
Pine Blu
Pittsbur
Pocatell
Portland
Poughkee Providen
Provo-Or
Pueblo, Racine,
Raleigh-
Rapid Ci
Reading,
Redding,
Reno-Spa
Richmond
Roanoke,
Rocheste
RochesteRockford
Rocky MoRome, GA
Saginaw-
St. Clou
St. Jose
St. Loui
Salem, O
SalisburSan Ange
San Anto
San Dieg
San Fran
San Jose
Santa Ba
Savannah
Scranton
Seattle-
SheboygaSherman-
Shrevepo
Sioux Ci
Sioux FaSouth Be
SpartanbSpokane,
Springfi
Springfi
Springfi
SpringfiSteubenv
Stockton
Sumter,
SyracuseTallahas
Terre Ha
Texarkan
Toledo,
Topeka, Trenton-
Tucson,
Tulsa, O
Tuscaloo
Tyler, T
ValdostaVallejo-
Vineland
Virginia
Visalia-
Waco, TX
Warner R
Washingt
WaterlooWausau, Wheeling
Wichita,
Wichita
Winston-
Worceste
Yakima,
York-Han
Youngsto
Yuba Cit
Geographic Distribution of Long-Term Vacancy Long-term vacancy varies much more by neighborhood.
Long
-Ter
m V
acan
t A
ddre
sse
s (L
og S
cale
)
Occupied Addresses (Log Scale)
500 1000 2500 5000 10000
1
10
100
500
1000
2000
Geographic Distribution of Long-Term Vacancy Long-term vacancy varies much more by neighborhood.
Mean Mean + 1 Std. Dev.
0.0
2.0
4.0
6.0
8F
ract
ion
of C
ens
us
Tra
cts
-.02 0 .02 .04 .06 .08 .1 .12 .14 .16 .18 .2Long-Term Vacancy Rate in 2013
Geographic Distribution of Long-Term Vacancy Distribution by metropolitan area is much more symmetric.
Mean Mean + 1 Std. Dev.0
.01
.02
.03
.04
.05
Fra
ctio
n o
f Met
rop
olita
n A
reas
0 .01 .02 .03 .04 .05 .06 .07 .08 .09 .1 .11 .12 .13 .14Long-Term Vacancy Rate in 2013
Characteristics of Tracts with High LT Vacancy Characteristics of tracts can shed light on possible causes of long-term
vacancy.
Merge USPS data with tract characteristics from 2008-2012 American Community Survey and the Census of 1990, 2000, and 2010.
Use principal component analysis to categorize tracts with a high long-term vacancy rate into groups with similar characteristics.
Analysis generates 4 groups.
Characteristics of Tracts with High LT Vacancy
“Poor Inner-City” tracts: Close to metropolitan area center
High density
High % multifamily housing
Low % owner-occupied housing
Low median income; high unemployment and poverty rate; high fraction of residents without a high school degree
Net population decline 1990-2010
High vacancy rate in 1990
Characteristics of Tracts with High LT Vacancy
“Housing Boom” tracts: Large % housing units built 2000-2009
High population growth 1990 to 2010
Far from metropolitan area center
Low density
High % owner-occupied housing
Characteristics of Tracts with High LT Vacancy
“Outlying” tracts: Far from metropolitan area center
Large fraction of tracts in rural or micropolitan areas
Low density
% housing units built 2000-2009 not higher than average
Unemployment and poverty rates not that different from the typical tract
Characteristics of Tracts with High LT Vacancy
“Dense Suburban” tracts: Average distance to metropolitan area center
Denser than the typical tract with less single-family and owner-occupied housing
Average population growth 1990-2010
Median income, poverty, and unemployment same as the typical tract, but a larger-than-normal fraction of college-educated residents
Vacancy rate in 1990 and 2000 not much higher than in the typical tract
Median house values and rents slightly higher than average.
Characteristics of Tracts with High Long-Term Vacancy
Four groups have very different characteristics, so there is probably not a single cause of long-term vacancy.
“Poor Inner-City”: characteristics of the population suggest that long-term vacancy is related to poor economic conditions.
“Housing boom”: vacancy may be partly due to over-building during the past 10 years.
“Dense suburban”: population characteristics not consistent with poverty or urban decline, and not much evidence of overbuilding—vacancy may be related to the severe economic recession.
Characteristics of Tracts with High Long-Term Vacancy
Exact groupings of tracts depend on tract characteristics included, method of analysis, cutoff used to define a high long-term vacancy rate.
In most cases, analysis identifies a “housing boom” group, an “inner-city” group, and at least one “suburban” group.
Where are Tracts with High Long-Term Vacancy?
Note. Map excludes metropolitan areas with fewer than 10 Census tracts.
High Long-Term Vacancy Tracts by Metropolitan Area
Where are Tracts with High Long-Term Vacancy?
Note. Map excludes metropolitan areas fewer than 10 Census tracts.
“Poor Inner-City” Tracts by Metropolitan Area
Where are Tracts with High Long-Term Vacancy?
Note. Map excludes metropolitan areas with fewer than 10 Census tracts.
“Housing Boom” Tracts by Metropolitan Area
Where are Tracts with High Long-Term Vacancy?
Note. Map excludes metropolitan areas with fewer than 10 Census tracts.
“Outlying” Tracts by Metropolitan Area
Where are Tracts with High Long-Term Vacancy?
Note. Map excludes metropolitan areas with fewer than 10 Census tracts.
“Dense Suburban” Tracts by Metropolitan Area
Long-term vacancy is not a large fraction of the aggregate housing stock. Nor is it responsible for much of the increase in the aggregate vacancy rate since 2001.
But long-term vacancy is concentrated in a small number of neighborhoods with very high vacancy rates.
These neighborhoods are not concentrated in only a few types of metropolitan areas.
The characteristics of these neighborhoods not all the same. Some are locations on the edges of MSAs with a lot of construction
during the housing boom. Others are inner-city locations with poor economic fundamentals—i.e.
low income and high poverty/unemployment. And others are in dense suburban areas.
Summary
In most neighborhoods, it is doubtful that a large stock of long-term vacant housing is holding back the recovery in prices or new construction.
But most metropolitan areas have at least a few neighborhoods with a high long-term vacancy rate.
Because the neighborhoods that have high long-term vacancy rates have diverse characteristics, the effectiveness of policies aimed at reducing the long-term vacant stock will depend on the type of neighborhood and metropolitan area being considered.
Implications for today’s housing market
Thank You
Gross Vacancy Rates Over Time
68
1012
14
1950 1960 1970 1980 1990 2000 2010year
HVS AHSACS Census
Pe
rce
nt o
f A
ll H
ous
ing
Un
its
Year-Round Vacancy Rates Over Time
46
810
1950 1960 1970 1980 1990 2000 2010year
HVS AHSACS Census
Pe
rce
nt o
f N
on-S
easo
nal
Ho
usin
g U
nits
Characteristics of Tracts with High LT Vacancy
Fraction ofHigh Long-Term Vacancy Tracts
Poor Inner City 35
Housing Boom 22
Outlying 20
Dense Suburban 18
Other 5