View original document

The full text on this page is automatically extracted from the file linked above and may contain errors and inconsistencies.

Documentation for Affordability and Availability of Rental
Housing in the Third Federal Reserve District: 2012
Community Development Studies & Education Department
Federal Reserve Bank of Philadelphia
Data and Methodology
Data: Through the American
Community Survey (ACS) program,
the U.S. Census Bureau has
surveyed roughly 3 million housing
units every calendar year since 2005.1
For each of these years, the Census
Bureau has released demographic
and housing estimates for the nation
as a whole and for jurisdictions with
a population of at least 65,000. Since
2008, estimates reflecting three years
of data collection have been released
annually for jurisdictions with a
population of at least 20,000. And
since 2010, estimates reflecting five
years of data collection have been
available for all standard census
geographies. Estimates for smaller
geographies require multiple years
of data – three or five, depending
on the size – because only about
2.5 percent of all households are
surveyed every year; thus, one-year
estimates for the least populous
geographies would be based on only
a handful of responses and would
not be nearly as reliable as three- or
five-year estimates.
In addition to releasing demographic
and housing estimates derived
from responses to the American
Community Survey, the U.S. Census

Bureau also makes an anonymized
subsample of survey responses
available to the public in the form
of Public Use Microdata Sample
(PUMS) files. Researchers use
PUMS files, which include survey
responses from roughly 1 percent of
the nation’s housing units each year,
to explore issues of interest to them
in more detail than the standard
ACS estimates allow. Three PUMS
housing files are released each year,
one for each of the one-, three-, and
five-year data collection periods.
This report on rental housing
affordability in the Third Federal
Reserve District is based on an
analysis of one-, three-, and five-year
ACS PUMS housing data sets.
Geography: Many precautions are
taken to ensure that individual
respondents to the ACS
questionnaire cannot be identified
in the PUMS files. One such
precaution is that there is little
geographic specificity associated
with the records in the PUMS data
set. Users can discern only the state
and the Public Use Microdata Area
(PUMA) for each record in the file.
Developed after every census by
state data centers in collaboration

December 2012
with regional, state, and local
partners, PUMAs are geographic
areas created for the sole purpose
of disseminating census and ACS
PUMS data.2
Because the minimum PUMA
population is 100,000, very populous
geographies (e.g., Philadelphia
and Montgomery counties in
Pennsylvania) include multiple
PUMAs that often fit together
within a larger geography like pieces
of a puzzle. Some counties, such as
Cape May County, New Jersey, have
a population of roughly 100,000
and form their own PUMA, while
smaller counties – or portions of
counties – must be grouped together
into a single PUMA in order to reach
the minimum population.
Using information provided by
the Missouri Census Data Center’s
MABLE/Geocorr2K application,3 we
were able to associate PUMAs with
the counties they overlap and thus
produce rental housing affordability
estimates for counties (or groups of
contiguous counties), MSAs,4 the
portions of states within the Third
District, and the Third District as
a whole.5 (See the appendix for a
list of PUMAs and the counties

Of the 3 million housing units selected into the sample each year, roughly 2 million interviews are routinely completed, in addition to 100,000-200,000
individuals living in group quarters. The ACS was under development and testing between 1996 and 2004, but until 2005, the sample size was much
smaller, and before 2006, those living in group quarters were excluded. For more information on the ACS, visit http://www.census.gov/acs/www/.

1

It is worth mentioning that this analysis uses PUMAs developed following the 2000 census. New PUMAs have been created based on 2010 census
data, but the ACS data sets used in this report do not incorporate the latest PUMA definitions. For general information on the historic development of
PUMAs, see U.S. Census Bureau, History of Public Use Microdata Areas (PUMAS): 1960-2000 (Washington, DC: U.S. Census Bureau).

2

3

http://mcdc.missouri.edu/websas/geocorr2k.html

Consistent with those used in reporting the 2010 ACS data, this analysis incorporates the MSA definitions released by the Office of Management and
Budget on December 1, 2009, in OMB Bulletin No. 10-02. In cases in which an MSA is primarily within the Third Federal Reserve District but includes
counties in another District, estimates reflect data from Third District counties only.

4

Visit http://www.census.gov/geo/www/maps/puma5pct.htm for maps that show the overlap of PUMAs and counties in each of the 50 states, the
District of Columbia, and Puerto Rico.

5

and MSAs to which they belong.)
Because of the way in which the
PUMAs were constructed, estimates
for the Third District deviate from
its standard geographic definition
in two instances. Jefferson County,
Pennsylvania, is not part of the
District but is included in this
analysis because it is part of a
PUMA that lies primarily within the
District. In addition, a small part of
Cambria County, Pennsylvania, is
not included in the analysis because
the PUMA that includes a portion of
the county lies primarily outside the
District.
One-, three-, or five-year estimates:
As mentioned above, three PUMS
housing files are produced annually
covering one, three, and five years
of survey data. One of the most
important decisions that a PUMS
user must make is which data set
to use for a given geography. The
decision should be based largely
on how populous the geography
is, which has implications for the
number of survey responses from
the geography and subsequently
for the reliability of the estimates
produced from an analysis of the
PUMS data set. In short, estimates
for more populous geographies –
many states and large MSAs, for
example – can be produced using
the one-year data sets. However, the
three- and five-year data sets are
more appropriate for smaller areas.6
After careful consideration of the
estimates and their respective

reliability, we have chosen to use
the one-year PUMS data sets –
covering the years 2005 through
2010 – to develop estimates for
geographies with at least 200,000
renter households. These include
the Third Federal Reserve District;
the portions of Pennsylvania,
New Jersey, and the PhiladelphiaCamden-Wilmington MSA that fall
within the District; and Philadelphia
County. For geographies with 50,000
to 200,000 renter households, we
have developed three-year estimates
using the most current ACS PUMS
files covering the years 2005-07 and
2008-10. For all others, we have used
the five-year data set that includes
information collected between 2006
and 2010.
Household income categories: In
addition to providing an analysis
of rental housing affordability in
the Third Federal Reserve District
for all renters, we have also
developed estimates for renters
in specific income categories. For
most, a renter household’s income
is compared to the median family
income (MFI) of the MSA in which
the renter lives. For renters who
do not live in an MSA, household
income is compared to the MFI of
the county or the PUMA in which
the household is located.7 (See the
appendix for a listing of the MFIs
used to categorize households in
each PUMA.)
In this analysis, we have developed
the following income categories

to classify renter households in
the District: 0-30 percent of the
MFI, referred to in this report as
extremely low income (ELI); 3150 percent of the MFI, referred to
here as very low income (VLI); and
51-80 percent of the MFI, referred
to here as low income (LI). The
income thresholds separating
these categories reflect straight
percentages of the MFI and thus are
not consistent with HUD’s official
income limits, which are subject
to a number of administrative
adjustments.8
In determining whether a
household falls within its area’s
ELI, VLI, or LI category, we adopt
the methodology used by HUD to
adjust for household size. In this
analysis, an area’s MFI is used to
classify a four-person household
into an income category; thresholds
are reduced by 10 to 30 percent for
smaller households (e.g., 70 percent
of the MFI is used to categorize
one-person households, 80 percent
for two-person households, etc.), and
the MFI is increased by 8 percent
for each additional person over four
that a household contains (e.g., 108
percent of the MFI for a five-person
household, 116 percent of the MFI
for a six-person household, etc.).
The rationale underpinning this
adjustment is that while a particular
income may be more than sufficient
to prevent financial hardship for
a renter living alone, the same
income may be inadequate for a
larger family. Rather than applying

To illustrate the increased reliability of multi-year estimates for counts of persons, households, or housing units, the standard errors for three- and
five-year estimates are roughly 58 percent and 45 percent, respectively, of the standard errors for one-year estimates (Michael Beaghen, et al., Research
Report Series (Statistics #2012-03) Interpretation and Use of American Community Survey Multiyear Estimates (U.S. Census Bureau, Center for Statistical
Research & Methodology, April 2012).

6

In the multi-year PUMS files, each household’s income is inflation-adjusted to the most current year covered by the data set and compared to the
multi-year median family income of the MSA, county, or PUMA (adjusted for inflation to the same year).

7

HUD’s income limits are subject to a number of adjustments. Official income limits can be adjusted upward where rental costs are particularly
high or downward where incomes are particularly high. There is also a 5 percent cap on year-over-year growth. It is important to note that the
30/50/80 percent thresholds to which the household size adjustment is applied in this report are the actual MFI estimates from the American
Community Survey and do not incorporate the other adjustments applied by HUD (see U.S. Department of Housing and Urban Development,
Office of Policy Development & Research, FY 2012 HUD Income Limits Briefing Material (Washington, DC: U.S. Department of Housing and Urban
Development, December 1, 2011).

8

2

the same 30/50/80 thresholds to
all households, we adjust the MFI
to account for household size to
recognize this difference.
Table Descriptions
This section provides table-specific
guidance on how estimates provided in Tables 1 through 6 should be
interpreted. It also includes information on the methodology used for
each calculation.
Table 1: Renter households by income
category: Provides estimates of the
number of renter households and
the percent that fall into the ELI,
VLI, and LI categories. Not shown
is the percent of renter households
that earn greater than 80 percent of
the MFI, which can be calculated
as 100 minus the share of renter
households classified as ELI, VLI, or
LI. For areas with one- and threeyear estimates, we calculate whether
the statistic is significantly higher or
lower than in the prior period at the
90 percent confidence level.
Table 2: Percent of renter households
spending more than 30 percent of
income on gross rent (including
utilities): Provides estimates of the
percent of renter households that
report a housing cost burden, which
is commonly defined as spending
in excess of 30 percent of income on
gross rent. For areas with one- and
three-year estimates, we calculate
whether the statistic is significantly
higher or lower than in the prior
period at the 90 percent confidence
level.
Table 3: Percent of renter households
spending more than 50 percent of
income on gross rent (including
utilities): Provides estimates of the
percent of renter households that

report a severe housing cost burden,
which is commonly defined as
spending in excess of 50 percent
of income on gross rent. For areas
with one- and three-year estimates,
we calculate whether the statistic is
significantly higher or lower than
in the prior period at the 90 percent
confidence level.
Table 4: Ratio of affordable rental
units for every 100 renter households
& surplus/deficit of affordable rental
units: Calculates the approximate
alignment between the number
of renter households in an income
category and the number of units
affordable to households within
the income category. A ratio below
100 suggests a deficit of affordable
units, and a ratio above 100 suggests
a surplus. For areas with one- and
three-year estimates, we calculate
whether the ratio is significantly
higher or lower than in the prior
period at the 90 percent confidence
level.
The methodology used to calculate
“affordable” (Table 4) and “affordable and available” (Table 5) ratios
and estimates is largely based on
HUD’s Worst Case Housing Needs series of reports to Congress.9 Methodologically, the first step in producing
these estimates is determining the
income at which each rental unit –
occupied and vacant – is affordable.
This calculation assumes that gross
rent should consume no more than
30 percent of income.10 Under this assumption, a unit with monthly gross
rent (rent plus utility costs) of $600
is affordable to a household with an
annual income of $24,000 ($600/30%
X 12 months = $24,000) or higher.
After we calculate the minimum
income at which a unit is affordable,

the next step is to determine the
30, 50, and 80 percent of MFI
thresholds for the area in which a
unit is located, taking into account
the number of people likely to live
in a unit based on the number of
bedrooms it has. Following HUD’s
methodology, we assume that one
person lives in an efficiency, a onebedroom unit can accommodate 1.5
persons comfortably, and that each
additional bedroom can comfortably
accommodate an additional 1.5
persons (e.g., three persons can live
in a two-bedroom unit, etc.). Thus,
implicit in the calculation of the 30,
50, and 80 percent of MFI thresholds
for each unit is an assumption
about the household size that would
occupy it. Using the same household
size adjustments described above
(i.e., 10 percent less than the MFI for
each person fewer than four, and
8 percent more than the MFI for
each person over four), we calculate
income category thresholds for each
unit and compare these thresholds
with the income at which the unit is
affordable. Assuming the $600 unit
described above has two bedrooms
and thus can accommodate three
people, the calculation of income
thresholds in an area with an MFI of
$50,000 would be:
• 30% threshold = 30% X $50,000
MFI X 90% household size
adjustment for a 3-person
household = $13,500
• 50% threshold = 50% X $50,000 X
90% household size adjustment
for a 3-person household =
$22,500
• 80% threshold = 80% X $50,000 X
90% household size adjustment
for a 3-person household =
$36,000

For the latest in the series, see Barry L. Steffen, et al., Worst Case Housing Needs 2009: Report to Congress (Washington, DC: U.S. Department of Housing
and Urban Development, Office of Policy Development and Research, February 2011).

9

This is the standard applied to many subsidized housing programs, which require tenants to pay 30 percent of their income – or of a given income
threshold – in monthly rent. For information on the evolution of this standard, see Danilo Pelletiere, Getting to the Heart of Housing’s Fundamental
Question: How Much Can a Family Afford? (Washington, DC: National Low Income Housing Coalition, February 2008).

10

3

Based on these calculations, a twobedroom unit with a monthly gross
rent of $600 is affordable within the
51-80 percent of MFI range because
the income at which it’s affordable
($24,000) is higher than the 50
percent threshold ($22,500) but
lower than the 80 percent threshold
($36,000).
After calculating the affordability
category to which each rental unit is
assigned, the next step is to calculate
the ratio of affordable units for every
100 renter households. For example,
if there are 100,000 rental units
affordable to households earning
no more than 30 percent of MFI and
200,000 such renter households, then
the ratio is 50 (100,000 units/200,000
households * 100 = 50). In other
words, there are only 50 rental units
affordable for every 100 ELI renter
households.
It is important to note that this
analysis likely overestimates the
number of affordable units in the
following two ways. First, the ACS
does not include information on utility costs for vacant rental units, so
the gross rent – and thus the income
at which a unit is affordable – is underestimated for these vacant units
(roughly 127,000 units, or 8.2 percent of the rental stock in the Third
District, in 2010). Compared to their
current classification, some number
of these vacant units would likely be
considered affordable to households
in the next-highest income category
if their utility costs were known.
Second, during the study period,
between 5.3 and 6.3 percent of all
renter households in the Third District reported paying no cash rent,
including anywhere from 0.6 to 1.0
percent who reported no gross housing costs at all. In this analysis, we
assume that units with no reported
gross housing costs are affordable
for households earning no more

11

4

than 30 percent of MFI. However, if
the arrangements that make units
rent free or entirely costless for
the current occupant(s) would not
be available to other renters in the
broader housing market, then these
units may overstate the supply of
affordable housing for ELI renters.
Together, these two issues likely
make the affordability estimates in
this report somewhat conservative.
It is also worth noting that the
ratios provided in Tables 4 and 5 are
based on the number of households
and rental units that fall within
income and affordability ranges
relative to the median family income
and are a good approximation of
affordability when households
and units are similarly distributed
within these ranges. In instances in
which household incomes and rents
are unevenly distributed within
these ranges, these ratios are less
instructive. For example, when
renter households are clustered
at the bottom of the 0-30 percent
income range and units are largely
affordable only to households at the
top end of the range, the stated ratio
of affordable rental units for every
100 ELI households overestimates
the degree of alignment between the
supply and demand of affordable
rental housing. Likewise, when
units are affordable to households
with incomes at the bottom of the
0-30 percent income range and
the majority of ELI households
have incomes at the upper end
of the category, the ratio does
not adequately capture the deep
affordability of the rental stock.
In addition to the ratios described
above, Table 4 also includes an
estimate of the surplus (positive)
or deficit (negative) of affordable
rental units by income category. It
is simply the difference between the
number of rental units affordable

to households in a given income
category and the number of renter
households in the same category.
In the example above, the deficit is
100,000 rental units (100,000 units –
200,000 households = -100,000).
Note that we have provided surplus/
deficit estimates for only the most
recent period (2010, 2008-10, or
2006-10) rather than comparing
them to earlier years. We do this
because the most recent ACS data
are “controlled” to the 2010 census,
whereas earlier ACS data are
predicated on population estimates
calculated by the Census Bureau
and based on the 2000 census.
Differences between earlier totals
and totals calculated from the most
recent data may be attributable to
this methodological shift rather than
to any real change in rental housing
affordability.11
Table 5: Ratio of affordable and
available rental units for every 100
renter households & surplus/deficit of
affordable and available rental units:
The methodology of assigning rental
housing units to income categories
described for Table 4, above, can be
applied to Table 5 as well. The sole
difference between Tables 4 and 5
is that while the former includes
all units that are affordable to a
particular income category, the
latter identifies only units that are
both affordable to households in
a given income category and are
either occupied by a household
in the same income category or
vacant. These units are considered
both “affordable and available” to
households in that income category.
The basic premise underpinning
the calculations in Table 5 is that
Table 4 overestimates the alignment
of affordable units and renter
households because it includes lowcost rentals that are occupied by

U.S. Census Bureau, American Community Survey Research Note: Change in Population Controls (Washington, DC: U.S. Census Bureau, September 2011).

higher-income households and thus
are not truly addressing the demand
for affordable housing for those with
lower incomes. By excluding rental
units that are occupied by higherincome households, the ratios and
surplus/deficit estimates are lower in
Table 5 than in Table 4.
In the Table 4 example above, there
were 100,000 rental units affordable
to ELI renters and 200,000 ELI
renter households. If 25,000 of these
affordable units were occupied
by renters with higher incomes,
then only 75,000 would be both
affordable and available for renter
households in the income category.
Subsequently, the ratio of affordable
and available rental units for every
100 ELI renter households would be
37.5 (75,000 units/200,000 households
* 100 = 37.5), and the deficit of
affordable and available rental units
would be 125,000 (75,000 units –
200,000 households = -125,000).12
For areas with one- and three-year
estimates, we calculate whether the
ratio is significantly higher or lower
than in the prior period at the 90
percent confidence level.
Table 6: Renter households with
incomplete kitchen/plumbing facilities
or crowded: Provides estimates of

the percent of renter households
that have inadequate kitchen or
plumbing facilities and/or are
considered crowded. A unit has an
incomplete kitchen if it is missing
any one of the following: a sink
with a faucet, a stove/range, or a
refrigerator. Plumbing facilities are
considered incomplete if the unit
lacks any one of the following: hot
and cold running water, a flush
toilet, or a bathtub/shower. A unit
is considered crowded if it includes
more than one person per room.13
One- and three-year estimates
are limited to the 2008-10 time
period because according to the
Census Bureau, changes to the ACS
questionnaire in 2008 directly led
to increases in the reported level of
incomplete kitchen and plumbing
facilities when compared with
pre-2008 estimates.14 In addition
to different instructions, response
options for the number of rooms
were also changed substantially
in 2008. Before 2008, survey
respondents had to check a box on
the questionnaire to indicate the
number of rooms in the unit, with
the largest option corresponding to
“9 or more rooms.” After 2008, the
number of units had a fill-in-theblank option. As a result, before 2008
it is impossible to know whether

very large households were crowded
because the number of rooms was
top-coded at nine.
For areas with one-year estimates,
we calculate whether the statistic is
significantly higher or lower than
in the prior period at the 90 percent
confidence level.
Comparability with Prior Reports
In 2010, we published a special
report titled Affordability and
Availability of Rental Housing in
Pennsylvania15 that provided an
in-depth look at rental housing
in Pennsylvania. The 2010 report
used both ACS data from 2005 and
2006 and Comprehensive Housing
Affordability Strategy (CHAS) data
sets from 1990 and 2000 produced by
HUD in order to better understand
housing affordability trends over
the 16-year period.16 In 2011, we
provided updated estimates for
Pennsylvania and its counties using
CHAS data from 2005-07, which
were based on ACS data collected
during those three years.17
The estimates in these prior
reports should not be compared
to the estimates presented here
for a number of reasons. First
and foremost, regarding the 1990
and 2000 CHAS data sets, there

Although the results in this report are not directly comparable to those in the department’s earliest rental housing report (described below),
the latter publication describes this calculation in greater detail. See Appendix C in Erin Mierzwa, Kathryn P. Nelson, and Harriet Newburger,
Affordability and Availability of Rental Housing in Pennsylvania (Philadelphia: Federal Reserve Bank of Philadelphia, Community Development Studies
and Education Department, March 2010).

12

13
Crowded units can be measured in multiple ways, but calculating the number of persons per room is the most common measurement. The
delineation between crowded and not crowded varies in the literature and can range from 0.75 to 2.0. See Kevin S. Blake, Rebecca L. Kellerson, and
Aleksandra Simic, Measuring Overcrowding in Housing (Washington, DC: Prepared by Econometrica for the U.S. Department of Housing and Urban
Development, Office of Policy and Research, 2007).
14

See “Comparing 2010 American Community Survey Data” at http://www.census.gov/acs/www/guidance_for_data_users/comparing_2010/.

15

Mierzwa, Nelson, and Newburger (March 2010)

16
HUD produced CHAS data sets following the 1990 and 2000 censuses in order to provide good local-level housing data to practitioners and
researchers. HUD has also begun a program to create more current CHAS data sets based on the American Community Survey. For more information
on CHAS data, visit http://www.huduser.org/portal/datasets/cp.html.
17
Federal Reserve Bank of Philadelphia, Community Development Studies and Education Department, New Rental Housing Data Based on the 2005-07
American Community Survey (ACS) (Philadelphia, PA: Federal Reserve Bank of Philadelphia, 2011).

5

are material differences between
estimates based on the long-form
survey administered in conjunction
with the decennial censuses, on
which these data sets are based, and
the ACS. Of greatest relevance to
this report, differences have been
observed in reported income levels
and rents,18 as well as in vacancy
rates.19
Even the 2011 report update, which
uses CHAS data based on the ACS,
cannot be considered comparable to
this ACS-based report. Methodological differences between this report
and the construction of the CHAS
data sets include the following,
which can have noticeable impacts
on the estimates produced:
• Current CHAS data sets do
not classify renter households
that report zero or negative
household income with regard
to their level of housing cost
burden.20 This report assumes
that these households are in the
ELI (0-30 percent MFI) category
and, if they report rent or utility
costs, that they spend more than
50 percent of their income on
housing.

• The CHAS data sets do not
calculate housing costs as a
percent of income for households with no cash rent and
positive household income, even
if the households report utility
costs. The concern is that these
households may live “rent free”
as a condition of their employment (e.g., a property manager
or minister) but not include
the value of their accommodations when reporting household
income. In the CHAS data sets,
such households are automatically classified as not burdened
by their housing costs.21 This
report calculates housing costs
as a percent of income for households with no cash rent, using
the sum of all utility costs as the
estimated gross rent.
• The CHAS data sets use 30, 50,
and 80 percent income category
thresholds that are subject to
administrative adjustments (e.g.,
higher in areas with high housing costs, etc.), calculated for fair
market rent (FMR) areas, and set
at the state’s nonmetropolitan
median if it would otherwise
be lower. This report calculates

these thresholds as straight
percentages, for standard MSAs,
and imposes no state minimum.
Finally, there are even differences
between this report and the original
2005-06 estimates included in the
2010 report that relied on tabulations
of the ACS PUMS housing files.
The original report used official
HUD FMR areas and the associated
income limits for counties in Pennsylvania as the threshold for the
income categories, a methodology
very similar to the one employed
in the CHAS data sets described in
the third bullet above. The original
report also took a straight average
of 2005 and 2006 estimates from the
one-year PUMS files and applied
the results to all counties or PUMA
groups of counties. In this report,
we use the standard ACS data sets
produced and weighted by the Census Bureau and choose the most current – but still reliable – one-, three-,
or five-year estimates to report for
each geography. Last, none of the
ACS estimates in this report cover
the same time period as the original
report (2005 and 2006).

For a discussion of the differences between the 2000 census and ACS data collected in the same year, see Gregg J. Diffendal, Rita Jo Petroni, and
Andre L. Williams, Meeting 21st Century Demographic Data Needs – Implementing the American Community Survey, Report 8: Comparison of the American
Community Survey Three-Year Averages and the Census Sample for a Sample of Counties and Tracts (Washington, DC: U.S. Census Bureau, 2004). See also
Kirby G. Posey, Edward Welniak, and Charles Nelson, Income in the American Community Survey, Comparisons to Census 2000 (Washington, DC: U.S.
Census Bureau), presented at the American Statistical Association Meetings (August 2003).
18

19
See Deborah H. Griffin, Comparing 2010 American Community Survey 1-Year Estimates of Occupancy Status, Vacancy Status, and Household Size with the
2010 Census – Preliminary Results (Washington, DC: U.S. Census Bureau, December 2011).
20

Based on correspondence with HUD research staff

21

Based on correspondence with HUD research staff

FEDERAL RESERVE BANK OF PHILADELPHIA
Community Development Studies & Education Department
Ten Independence Mall • Philadelphia, PA 19106-1533
6

Appendix. Relationship of PUMAs to Standard Geographies & Median Family Incomes Used in Analysis

Median Family Income Used to Classify Households

State
DE
DE
DE
DE
DE
DE
NJ
NJ
NJ
NJ
NJ
NJ
NJ
NJ
NJ
NJ
NJ
NJ
NJ
NJ
NJ
NJ
NJ
NJ
PA
PA
PA
PA
PA
PA
PA
PA
PA
PA
PA
PA
PA
PA
PA
PA
PA
PA
PA

County/Counties
Kent
New Castle
New Castle
New Castle
New Castle
Sussex
Atlantic
Atlantic
Burlington
Burlington
Burlington
Camden
Camden
Camden
Camden
Cape May
Cumberland
Gloucester
Gloucester/Salem
Mercer
Mercer
Ocean
Ocean
Ocean
Adams/Franklin
Bedford/Fulton/Huntingdon
Berks
Berks
Blair
Bradford/Sullivan/Tioga
Bucks
Bucks
Bucks
Bucks
Cambria (part)
Cameron/Elk/McKean/Potter
Carbon/Lehigh
Centre
Chester
Chester
Chester
Clearfield/Jefferson
Clinton/Juniata/Mifflin/Snyder/Union

PUMA Code
00200
00101
00102
00103
00104
00300
00101
00102
02001
02002
02003
02101
02102
02103
02104
00200
02400
02201
02202
02301
02302
01201
01202
01203
02801
02700
03401
03402
02600
00500
03901
03902
03903
03904
02501
00400
03702
01300
04301
04302
04303
01400
01200

MSA (if applicable)
Dover, DE MSA
Philadelphia-Camden-Wilmington, PA-NJ-DE-MD MSA
Philadelphia-Camden-Wilmington, PA-NJ-DE-MD MSA
Philadelphia-Camden-Wilmington, PA-NJ-DE-MD MSA
Philadelphia-Camden-Wilmington, PA-NJ-DE-MD MSA

PA
PA
PA
PA
PA
PA
PA
PA
PA
PA
PA
PA

Columbia/Luzerne
Cumberland
Cumberland/Perry
Dauphin
Dauphin
Delaware
Delaware
Delaware
Delaware
Franklin
Lackawanna
Lackawanna/Wyoming

00903
03101
03102
03001
03002
04201
04202
04203
04204
02802
00801
00802

Scranton--Wilkes-Barre, PA MSA (Luzerne)
Harrisburg-Carlisle, PA MSA
Harrisburg-Carlisle, PA MSA
Harrisburg-Carlisle, PA MSA
Harrisburg-Carlisle, PA MSA
Philadelphia-Camden-Wilmington, PA-NJ-DE-MD MSA
Philadelphia-Camden-Wilmington, PA-NJ-DE-MD MSA
Philadelphia-Camden-Wilmington, PA-NJ-DE-MD MSA
Philadelphia-Camden-Wilmington, PA-NJ-DE-MD MSA

Atlantic City-Hammonton, NJ MSA
Atlantic City-Hammonton, NJ MSA
Philadelphia-Camden-Wilmington, PA-NJ-DE-MD MSA
Philadelphia-Camden-Wilmington, PA-NJ-DE-MD MSA
Philadelphia-Camden-Wilmington, PA-NJ-DE-MD MSA
Philadelphia-Camden-Wilmington, PA-NJ-DE-MD MSA
Philadelphia-Camden-Wilmington, PA-NJ-DE-MD MSA
Philadelphia-Camden-Wilmington, PA-NJ-DE-MD MSA
Philadelphia-Camden-Wilmington, PA-NJ-DE-MD MSA
Ocean City, NJ MSA
Vineland-Millville-Bridgeton, NJ MSA
Philadelphia-Camden-Wilmington, PA-NJ-DE-MD MSA
Philadelphia-Camden-Wilmington, PA-NJ-DE-MD MSA
Trenton-Ewing, NJ MSA
Trenton-Ewing, NJ MSA
New York-Northern New Jersey-Long Island, NY-NJ-PA MSA
New York-Northern New Jersey-Long Island, NY-NJ-PA MSA
New York-Northern New Jersey-Long Island, NY-NJ-PA MSA

Reading, PA MSA
Reading, PA MSA
Altoona, PA MSA
Philadelphia-Camden-Wilmington, PA-NJ-DE-MD MSA
Philadelphia-Camden-Wilmington, PA-NJ-DE-MD MSA
Philadelphia-Camden-Wilmington, PA-NJ-DE-MD MSA
Philadelphia-Camden-Wilmington, PA-NJ-DE-MD MSA
Johnstown, PA MSA
Allentown-Bethlehem-Easton, PA-NJ MSA
State College, PA MSA
Philadelphia-Camden-Wilmington, PA-NJ-DE-MD MSA
Philadelphia-Camden-Wilmington, PA-NJ-DE-MD MSA
Philadelphia-Camden-Wilmington, PA-NJ-DE-MD MSA

Scranton--Wilkes-Barre, PA MSA
Scranton--Wilkes-Barre, PA MSA

2005
$56,778
$67,830
$67,830
$67,830
$67,830
$50,608
$61,240
$61,240
$67,830
$67,830
$67,830
$67,830
$67,830
$67,830
$67,830
$59,679
$53,074
$67,830
$67,830
$80,637
$80,637
$67,419
$67,419
$67,419
$54,458
$47,184
$60,206
$60,206
$45,987
$43,516
$67,830
$67,830
$67,830
$67,830
$43,126
$46,814
$62,379
$55,240
$67,830
$67,830
$67,830
$42,368
$46,237

2010
$63,962
$74,506
$74,506
$74,506
$74,506
$54,069
$61,541
$61,541
$74,506
$74,506
$74,506
$74,506
$74,506
$74,506
$74,506
$72,075
$64,583
$74,506
$74,506
$85,547
$85,547
$74,756
$74,756
$74,756
$63,528
$50,592
$62,493
$62,493
$53,448
$52,044
$74,506
$74,506
$74,506
$74,506
$52,914
$48,435
$67,207
$62,828
$74,506
$74,506
$74,506
$46,035
$49,096

2005-07
$56,130
$73,536
$73,536
$73,536
$73,536
$55,700
$64,812
$64,812
$73,536
$73,536
$73,536
$73,536
$73,536
$73,536
$73,536
$63,359
$56,926
$73,536
$73,536
$85,169
$85,169
$73,088
$73,088
$73,088
$60,339
$47,960
$61,446
$61,446
$50,444
$45,931
$73,536
$73,536
$73,536
$73,536
$46,249
$49,460
$66,012
$62,306
$73,536
$73,536
$73,536
$44,074
$48,591

2008-10
$62,831
$76,710
$76,710
$76,710
$76,710
$58,265
$64,381
$64,381
$76,710
$76,710
$76,710
$76,710
$76,710
$76,710
$76,710
$69,632
$61,557
$76,710
$76,710
$89,891
$89,891
$76,228
$76,228
$76,228
$64,610
$51,474
$63,785
$63,785
$52,637
$50,097
$76,710
$76,710
$76,710
$76,710
$52,189
$49,004
$68,667
$64,445
$76,710
$76,710
$76,710
$47,216
$49,892

2006-10
$60,949
$77,000
$77,000
$77,000
$77,000
$59,053
$66,920
$66,920
$77,000
$77,000
$77,000
$77,000
$77,000
$77,000
$77,000
$69,978
$60,642
$77,000
$77,000
$88,694
$88,694
$76,709
$76,709
$76,709
$64,790
$50,901
$63,724
$63,724
$53,166
$49,414
$77,000
$77,000
$77,000
$77,000
$50,900
$50,086
$68,935
$65,121
$77,000
$77,000
$77,000
$46,864
$50,293

$47,844
$62,658
$62,658
$62,658
$62,658
$67,830
$67,830
$67,830
$67,830
$52,286
$49,623
$49,623

$55,185
$66,619
$66,619
$66,619
$66,619
$74,506
$74,506
$74,506
$74,506
$59,617
$55,682
$55,682

$51,551
$66,778
$66,778
$66,778
$66,778
$73,536
$73,536
$73,536
$73,536
$58,969
$53,337
$53,337

$54,808
$68,537
$68,537
$68,537
$68,537
$76,710
$76,710
$76,710
$76,710
$59,803
$54,983
$54,983

$54,835
$69,389
$69,389
$69,389
$69,389
$77,000
$77,000
$77,000
$77,000
$61,410
$56,045
$56,045

Source of Median Family Income
Dover, DE MSA
Philadelphia-Camden-Wilmington, PA-NJ-DE-MD MSA
Philadelphia-Camden-Wilmington, PA-NJ-DE-MD MSA
Philadelphia-Camden-Wilmington, PA-NJ-DE-MD MSA
Philadelphia-Camden-Wilmington, PA-NJ-DE-MD MSA
Sussex Co
Atlantic City-Hammonton, NJ MSA
Atlantic City-Hammonton, NJ MSA
Philadelphia-Camden-Wilmington, PA-NJ-DE-MD MSA
Philadelphia-Camden-Wilmington, PA-NJ-DE-MD MSA
Philadelphia-Camden-Wilmington, PA-NJ-DE-MD MSA
Philadelphia-Camden-Wilmington, PA-NJ-DE-MD MSA
Philadelphia-Camden-Wilmington, PA-NJ-DE-MD MSA
Philadelphia-Camden-Wilmington, PA-NJ-DE-MD MSA
Philadelphia-Camden-Wilmington, PA-NJ-DE-MD MSA
Ocean City, NJ MSA
Vineland-Millville-Bridgeton, NJ MSA
Philadelphia-Camden-Wilmington, PA-NJ-DE-MD MSA
Philadelphia-Camden-Wilmington, PA-NJ-DE-MD MSA
Trenton-Ewing, NJ MSA
Trenton-Ewing, NJ MSA
New York-Northern New Jersey-Long Island, NY-NJ-PA MSA
New York-Northern New Jersey-Long Island, NY-NJ-PA MSA
New York-Northern New Jersey-Long Island, NY-NJ-PA MSA
PUMA 02801
PUMA 02700
Reading, PA MSA
Reading, PA MSA
Altoona, PA MSA
PUMA 00500
Philadelphia-Camden-Wilmington, PA-NJ-DE-MD MSA
Philadelphia-Camden-Wilmington, PA-NJ-DE-MD MSA
Philadelphia-Camden-Wilmington, PA-NJ-DE-MD MSA
Philadelphia-Camden-Wilmington, PA-NJ-DE-MD MSA
Johnstown, PA MSA
PUMA 00400
Allentown-Bethlehem-Easton, PA-NJ MSA
State College, PA MSA
Philadelphia-Camden-Wilmington, PA-NJ-DE-MD MSA
Philadelphia-Camden-Wilmington, PA-NJ-DE-MD MSA
Philadelphia-Camden-Wilmington, PA-NJ-DE-MD MSA
PUMA 01400
PUMA 01200
Weighted average of Scranton--Wilkes-Barre, PA MSA (Luzerne
Co) and Columbia Co
Harrisburg-Carlisle, PA MSA
Harrisburg-Carlisle, PA MSA
Harrisburg-Carlisle, PA MSA
Harrisburg-Carlisle, PA MSA
Philadelphia-Camden-Wilmington, PA-NJ-DE-MD MSA
Philadelphia-Camden-Wilmington, PA-NJ-DE-MD MSA
Philadelphia-Camden-Wilmington, PA-NJ-DE-MD MSA
Philadelphia-Camden-Wilmington, PA-NJ-DE-MD MSA
Franklin Co
Scranton--Wilkes-Barre, PA MSA
Scranton--Wilkes-Barre, PA MSA

Appendix. Relationship of PUMAs to Standard Geographies & Median Family Incomes Used in Analysis

Median Family Income Used to Classify Households

State
PA
PA
PA
PA
PA
PA
PA
PA
PA
PA
PA
PA
PA
PA
PA
PA
PA
PA
PA
PA
PA
PA
PA
PA
PA
PA
PA
PA
PA
PA

County/Counties
Lancaster
Lancaster
Lancaster
Lebanon
Lehigh
Lehigh
Luzerne
Luzerne
Lycoming
Monroe
Montgomery
Montgomery
Montgomery
Montgomery
Montgomery
Montgomery
Montour/Northumberland
Northampton
Northampton
Philadelphia
Philadelphia
Philadelphia
Philadelphia
Philadelphia
Philadelphia
Philadelphia
Philadelphia
Philadelphia
Philadelphia
Philadelphia

PUMA Code
03301
03302
03303
02900
03600
03701
00901
00902
01000
00700
04001
04002
04003
04004
04005
04006
01100
03801
03802
04101
04102
04103
04104
04105
04106
04107
04108
04109
04110
04111

MSA (if applicable)
Lancaster, PA MSA
Lancaster, PA MSA
Lancaster, PA MSA
Lebanon, PA MSA
Allentown-Bethlehem-Easton, PA-NJ MSA
Allentown-Bethlehem-Easton, PA-NJ MSA
Scranton--Wilkes-Barre, PA MSA
Scranton--Wilkes-Barre, PA MSA
Williamsport, PA MSA

PA
PA
PA
PA
PA

Pike/Susquehanna/Wayne
Schuylkill
York
York
York

00600
03500
03201
03202
03203

New York-Northern New Jersey-Long Island, NY-NJ-PA MSA (Pike)

Philadelphia-Camden-Wilmington, PA-NJ-DE-MD MSA
Philadelphia-Camden-Wilmington, PA-NJ-DE-MD MSA
Philadelphia-Camden-Wilmington, PA-NJ-DE-MD MSA
Philadelphia-Camden-Wilmington, PA-NJ-DE-MD MSA
Philadelphia-Camden-Wilmington, PA-NJ-DE-MD MSA
Philadelphia-Camden-Wilmington, PA-NJ-DE-MD MSA
Allentown-Bethlehem-Easton, PA-NJ MSA
Allentown-Bethlehem-Easton, PA-NJ MSA
Philadelphia-Camden-Wilmington, PA-NJ-DE-MD MSA
Philadelphia-Camden-Wilmington, PA-NJ-DE-MD MSA
Philadelphia-Camden-Wilmington, PA-NJ-DE-MD MSA
Philadelphia-Camden-Wilmington, PA-NJ-DE-MD MSA
Philadelphia-Camden-Wilmington, PA-NJ-DE-MD MSA
Philadelphia-Camden-Wilmington, PA-NJ-DE-MD MSA
Philadelphia-Camden-Wilmington, PA-NJ-DE-MD MSA
Philadelphia-Camden-Wilmington, PA-NJ-DE-MD MSA
Philadelphia-Camden-Wilmington, PA-NJ-DE-MD MSA
Philadelphia-Camden-Wilmington, PA-NJ-DE-MD MSA
Philadelphia-Camden-Wilmington, PA-NJ-DE-MD MSA

York-Hanover, PA MSA
York-Hanover, PA MSA
York-Hanover, PA MSA

2005
$60,767
$60,767
$60,767
$55,588
$62,379
$62,379
$49,623
$49,623
$47,164
$57,661
$67,830
$67,830
$67,830
$67,830
$67,830
$67,830
$47,417
$62,379
$62,379
$67,830
$67,830
$67,830
$67,830
$67,830
$67,830
$67,830
$67,830
$67,830
$67,830
$67,830

2010
$61,760
$61,760
$61,760
$60,842
$67,207
$67,207
$55,682
$55,682
$49,997
$62,944
$74,506
$74,506
$74,506
$74,506
$74,506
$74,506
$49,652
$67,207
$67,207
$74,506
$74,506
$74,506
$74,506
$74,506
$74,506
$74,506
$74,506
$74,506
$74,506
$74,506

2005-07
$63,499
$63,499
$63,499
$60,588
$66,012
$66,012
$53,337
$53,337
$48,929
$62,833
$73,536
$73,536
$73,536
$73,536
$73,536
$73,536
$49,744
$66,012
$66,012
$73,536
$73,536
$73,536
$73,536
$73,536
$73,536
$73,536
$73,536
$73,536
$73,536
$73,536

2008-10
$63,807
$63,807
$63,807
$61,334
$68,667
$68,667
$54,983
$54,983
$51,629
$64,463
$76,710
$76,710
$76,710
$76,710
$76,710
$76,710
$48,949
$68,667
$68,667
$76,710
$76,710
$76,710
$76,710
$76,710
$76,710
$76,710
$76,710
$76,710
$76,710
$76,710

2006-10
$64,672
$64,672
$64,672
$62,174
$68,935
$68,935
$56,045
$56,045
$52,124
$64,763
$77,000
$77,000
$77,000
$77,000
$77,000
$77,000
$50,066
$68,935
$68,935
$77,000
$77,000
$77,000
$77,000
$77,000
$77,000
$77,000
$77,000
$77,000
$77,000
$77,000

$53,796
$45,782
$56,459
$56,459
$56,459

$62,052
$49,664
$66,964
$66,964
$66,964

$58,375
$49,145
$63,291
$63,291
$63,291

$63,115
$53,264
$67,820
$67,820
$67,820

$62,205
$53,083
$67,624
$67,624
$67,624

Source of Median Family Income
Lancaster, PA MSA
Lancaster, PA MSA
Lancaster, PA MSA
Lebanon, PA MSA
Allentown-Bethlehem-Easton, PA-NJ MSA
Allentown-Bethlehem-Easton, PA-NJ MSA
Scranton--Wilkes-Barre, PA MSA
Scranton--Wilkes-Barre, PA MSA
Williamsport, PA MSA
Monroe Co
Philadelphia-Camden-Wilmington, PA-NJ-DE-MD MSA
Philadelphia-Camden-Wilmington, PA-NJ-DE-MD MSA
Philadelphia-Camden-Wilmington, PA-NJ-DE-MD MSA
Philadelphia-Camden-Wilmington, PA-NJ-DE-MD MSA
Philadelphia-Camden-Wilmington, PA-NJ-DE-MD MSA
Philadelphia-Camden-Wilmington, PA-NJ-DE-MD MSA
PUMA 01100
Allentown-Bethlehem-Easton, PA-NJ MSA
Allentown-Bethlehem-Easton, PA-NJ MSA
Philadelphia-Camden-Wilmington, PA-NJ-DE-MD MSA
Philadelphia-Camden-Wilmington, PA-NJ-DE-MD MSA
Philadelphia-Camden-Wilmington, PA-NJ-DE-MD MSA
Philadelphia-Camden-Wilmington, PA-NJ-DE-MD MSA
Philadelphia-Camden-Wilmington, PA-NJ-DE-MD MSA
Philadelphia-Camden-Wilmington, PA-NJ-DE-MD MSA
Philadelphia-Camden-Wilmington, PA-NJ-DE-MD MSA
Philadelphia-Camden-Wilmington, PA-NJ-DE-MD MSA
Philadelphia-Camden-Wilmington, PA-NJ-DE-MD MSA
Philadelphia-Camden-Wilmington, PA-NJ-DE-MD MSA
Philadelphia-Camden-Wilmington, PA-NJ-DE-MD MSA
Weighted average of New York-Northern New Jersey-Long Island,
NY-NJ-PA MSA (Pike Co), Susquehanna Co, and Wayne Co
Schuylkill Co
York-Hanover, PA MSA
York-Hanover, PA MSA
York-Hanover, PA MSA

Note: Some counties are listed both individually and in conjunction with another county (see Franklin, PA, and Adams/Franklin, PA). In these cases, at least one PUMA is wholly contained within the individual county (Franklin), and another PUMA overlaps both
counties. Note that in this particular example, different incomes are used to categorize households into income categories: Franklin County for the former and the two-county PUMA for the latter. In other instances, where both PUMAs are contained within the
same MSA, the same income is used (see Lackawanna, PA, and Lackawanna/Wyoming, PA). Within the report, statistics are reported for the combined counties.
Source: 2005-10, American Community Survey, Table B19113 for median family incomes; Missouri Census Data Center's MABLE/Geocorr2K application for information on the relationship between Public Use Microdata Areas and standard census geographies
(http://mcdc.missouri.edu/websas/geocorr2k.html).