Upload
others
View
3
Download
0
Embed Size (px)
Citation preview
Varying the Economies of Scale in Housing:
Impact on Supplemental Poverty Measure
Statistics
Thesia I. Garner, Bureau of Labor StatisticsTrudi Renwick, U.S. Census Bureau
Seventh Meeting of the Society for the Study of Economic InequalityThe Graduate Center, City University of New York
New York City, NYJuly 17-19, 2017
Disclaimer
This presentation reports the results of researchand analysis undertaken by researchers within theBureau of the Census (Census) and Bureau ofLabor Statistics (BLS).
Any views expressed are those of the authors andnot necessarily those of the Census or BLS.
Results are preliminary and not to be quotedwithout authors’ permission.
2
First, thresholds to be based only on food, clothing, shelter, utilities (FCSU), and multiplier for other basic goods and services like for personal care and nonwork related transportation
Second, since NAS Panel report was issued in 1995, became clear that significant number of low-income families own a home without a mortgage and therefore have quite low shelter expense requirements (see ITWG document, March 2010)
Not taking this into account may overstate their poverty rates
Suggests the need to adjust SPM thresholds for housing status, distinguishing renters, owners with a mortgage, and owners without a mortgage.
There would be THREE thresholds, NOT a single threshold for all.
3 — U.S. BUREAU OF LABOR STATISTICS • bls.gov
Why Three Thresholds?
35%
10%
9%
11%
12%
15%
22%
29%
39%
60%
27%
17%
22%
29%
36%
38%
32%
30%
25%
22%
38%
73%
69%
60%
52%
47%
46%
42%
35%
19%
Allconsumer
units
Lessthan
$5,000
$5,000to
$9,999
$10,000to
$14,999
$15,000to
$19,999
$20,000to
$29,999
$30,000to
$39,999
$40,000to
$49,999
$50,000to
$69,999
$70,000and
more
Percent Distribution of Consumer Units by Before Tax Income and Housing Type: 2015
With mortgage
Without mortgage
Renter
Source: Bureau of Labor Statistics, Prepublication Table 1202. Income before taxes: Annual means, standard errors, coefficients of
variation, and quarterly percents reporting, Consumer Expenditure Interview Survey, 2015.
Motivation Supplemental Poverty Measure (SPM) assumes that all SPM resource units
within each housing type (renter, owner with a mortgage, owner without a mortgage) devote the same share of their threshold to housing costs. This is important because … Geographical adjustments applied to only the housing share of threshold
Value of housing assistance benefits capped at no greater than housing share of the threshold
Concerns about underestimation of the value of housing benefits led to an investigation of this assumption… Should the housing share vary by household size?
Should economies of scale in housing differ from those of other commodities in SPM thresholds?
Bishop (2010) critique regarding holding economies of scale parameter constant and concern with implications for housing
5
Prior Research by Renwick and Mitchell
on Alternative Housing Shares
34.7
17.6
50.5
21.6
31.8
13.6
29.926.0
0
10
20
30
40
50
60
Poverty ratefor units with
housingassistance
Percent withzero subsidies
Percentcapped
Aggregateamount of
housingsubsidies
SPM New Housing Shares
Source: Renwick and Mitchell, 2015, using 2013 CPS ASEC.
Changed housing shares to 70 percent for one person units; 60 percent for two person units
Increased effect of housing subsidies on SPM rates.
6
Housing Status Thresholds and Adjustments
7 — U.S. BUREAU OF LABOR STATISTICS • bls.gov
FCSU expenditures for CUs with 2 children convert to FCSU for reference unit 2 adults with 2 children based on means within 30th to 36th
percentile range of FCSU for reference CUs
Thresholds produced for each housing tenure group, j
Owners with mortgages, Owners without mortgages. Renters
For reference 2A+2C, shelter+utility shares of thresholds produced, αj
Apply 3-parameter equivalence scale to derive Tj,s
Finally geographically adjust housing part of threshold for cost of living, geo-adjusted thresholdj = p(αjTj,s )+ (1- αj )Tj,s
Research Experimental Supplemental Poverty
Measure (SPM) Thresholds for Two Adults and
Two Children, 2015
Threshold amount Housing Share
SPM Owners with mortgages
$25,930 0.505
SPM Owners without mortgages
$21,806 0.411
SPM Renters $25,583 0.498
http://www.bls.gov/pir/spm/spm_chart1_2015data.htm
Housing Shares of
SPM Thresholds for 2A+2C: 2015
50.5
41.1
49.8
0
10
20
30
40
50
60
Homeownerswith a Mortgage
Homeownerswithout aMortgage
Renters
Source: Bureau of Labor Statistics, Division of Price and Index Number Research, September 2016.
Housing shares for the reference SPM unit (2 adults+2 children) used and impact two aspects of the SPM calculation:
Geographic adjustment on only the housing share of the threshold
Value of housing assistance capped using the housing portion of threshold minus renter payment (e.g., $12,663)
9
Restated: Concern with Holding Housing Shares Constant
10 — U.S. BUREAU OF LABOR STATISTICS • bls.gov
Housing is expected to exhibit greater economies of scale than food and clothing, so seems reasonable to examine using smaller or larger portions of thresholds for account for economies of scale differences
Portion of SPM thresholds subject to geographic adjustment would be larger for smaller families, thereby increasing thresholds for those living in areas with housing costs greater than national median and decreasing for those in lower housing costs areas
Because value of housing subsidy capped, increase value of housing subsidies for some smaller families that could reduce their poverty rates
Options to Deal with Greater
Economies of Scale in HousingOPTION 1
Apply 3-parameter equivalence scale for other household sizes and types But now, housing share varies also by consumer unit size
Tg,β,j,s= pg(βj,sTj,s)+ (1- βj,s )Tj,s
New Βj,s from survey data
OPTION 2 Apply housing equivalence scale to housing part of 2A+2C threshold Apply F+C equivalence scale to F+C part of 2A+2C threshold For example, for 1-person thresholds, with N=4:
Estimate new scales parameters, h and f, using survey data
11 — U.S. BUREAU OF
LABOR STATISTICS • bls.gov
𝑇𝑔, 𝑗, 1 = pg
(α𝑗𝑇𝑗,2+2)
𝑁ℎ + ((𝟏−𝛼𝒋)𝑇𝑗,2+2)
𝑁𝑓
Alternative Estimates of Housing
Shares
0%
10%
20%
30%
40%
50%
60%
70%
80%
1 2 3 4 5 6 7+
Household Size
Figure 1. Total Population Housing Expenditures as a Percent of Income
Gross Rent as aPercentage ofHousehold Income
Selected MonthlyOwners Costs as aPercentage ofHousehold Income -Mortgage
Selected MonthlyOwners Costs as aPercentage ofHousehold Income -No Mortgage
Source: 2014 American Community Survey Five Year Data.For more information, see census.gov/acs.
0%
10%
20%
30%
40%
50%
60%
70%
80%
1 2 3 4 5 6 7+
Household Size
Figure 2. Housing Expenditures as a Percent of Income - Households with Income Less than 200
percent of Official Poverty
Gross Rent as aPercentage ofHousehold Income
Selected MonthlyOwners Costs as aPercentage ofHousehold Income -Mortgage
Selected MonthlyOwners Costs as aPercentage ofHousehold Income -No Mortgage
Source: 2014 American Community Survey Five Year Data. For more information, see census.gov/acs.
12
Comparing ACS to CE Estimates
0%
10%
20%
30%
40%
50%
60%
70%
80%
1 2 3 4 5 6 7+
Household Size
Figure 2. Housing Expenditures as a Percent of Income - Households with Income Less than 200
percent of Official Poverty
Gross Rent as aPercentage ofHousehold Income
Selected MonthlyOwners Costs as aPercentage ofHousehold Income -Mortgage
Selected MonthlyOwners Costs as aPercentage ofHousehold Income -No Mortgage
Source: 2014 American Community Survey Five Year Data. For more information, see census.gov/acs.
0%
10%
20%
30%
40%
50%
60%
70%
80%
1 2 3 4 5 6 7+
Consumer Unit Size
Figure 3. Housing Expenditures as a Percent of the Sum of Food, Clothing, Shelter, Utilities
(FCSU), and Other Basic Needs (20%) Expenditures
Renters
Owners withMortgage
Ownerswithout aMortgage
Source: Quarterly Consumer Expenditure Survey, all consumer units participating In 2008Q2-2013Q1
13
Regression-Based Equivalence Scales
Basic model:
𝑙𝑛𝐸𝑥𝑝 = 𝛽0 + 𝛽1𝑙𝑛𝑌 − 𝛽2𝑙𝑛𝑁 +⋯+ 𝜀
Rearranging predicted values yields an expression for log income share devoted to, for example, housing.
𝑙𝑛𝑆𝐻𝐴𝑅𝐸 = 𝛽0 + 𝛽1 − 1 𝑙𝑛 𝑌 − 𝛽2𝑙𝑛𝑁 +⋯
All else constant, a consumer unit with 𝑌𝑁 will be equally well-off as a single person with 𝑌1if:
𝛽0 + 𝛽1 − 1 𝑙𝑛 𝑌1 +⋯ = 𝛽0 + 𝛽1 − 1 𝑙𝑛 𝑌𝑁 − 𝛽2𝑙𝑛𝑁 +⋯
Cancelling and rearranging terms yields the single-parameter approximation:
𝑌𝑁
𝑌1= 𝑁
𝛽2𝛽1−1
• Run separately for housing and for food+clothing• 5 years of CE Interview data, expenditures in constant dollars• Control variables: urban/rural, region (NE,S,MW,W), time (2008, 2009, 2010, 2011, 2012)
Consumer Unit Size Equivalence Scale Parameters
0.569
0.316
0.589
0.118
0.571
0.120
0.665
0.286
0.582
0.000
0.100
0.200
0.300
0.400
0.500
0.600
0.700
0.800
0.900
1.000
3-P
aram
eter
Sca
le
Ho
usi
ng
Foo
d+C
loth
ing
Ho
usi
ng
Foo
d+C
loth
ing
Ho
usi
ng
Foo
d+C
loth
ing
Ho
usi
ng
Foo
d+C
loth
ing
All Housing Types Together Owners with Mortgages Owners withoutMortgages
Renters
Source: Bureau of Labor Statistics, Garner’s analysis of internal CE Interview data from 2008Q2 through 2013Q1, four quarters of data for each
consumer unit. Three parameter scale based on 1 person=1adult, 2 people=1 adult+1 child, 3 people=2 adults+1 child, 4 people=2 adults+2 children, 5
people=2 adults+3 children, and 6 people=3 adults+3 children.
2015 SPM Thresholds: Varying Equivalence Scales
$10,000
$15,000
$20,000
$25,000
$30,000
$35,000
Three-Parameter Scale
Owner With Mortgage
Owner Without Mortgage
Renter
$10,000
$15,000
$20,000
$25,000
$30,000
$35,000
1 2 3 4 5 6
Single Parameter for All CUs
Owner With Mortgage
Owner Without Mortgage
Renter
$10,000
$15,000
$20,000
$25,000
$30,000
$35,000
1 2 3 4 5 6
Single Parameter for Housing Type Separately
Owner With Mortgage
Owner Without Mortgage
Renter
Source: Bureau of Labor Statistics, Garner’s analysis of internal CE Interview data from 2008Q2 through 2013Q1, four quarters of
data for each consumer unit.
Housing Share of SPM Thresholds by
Methodology before Geo Adjustment
17
1 2 3 4 5 6
SPM 49.8% 49.8% 49.8% 49.8% 49.8% 49.8%
ACS 59.3% 52.8% 52.8% 49.8% 47.3% 45.8%
CE 53.3% 51.8% 50.3% 49.8% 48.3% 47.8%
SP for All Units 59.2% 54.5% 51.8% 49.8% 48.3% 47.0%
SP by Housing Type 59.9% 54.9% 51.9% 49.8% 48.1% 46.8%
45.0%
50.0%
55.0%
60.0%
Current SPM
CE Shares
Single parameter by housing type
Single parameter -- all units
ACS Shares
Source: Authors’ calculations based on ACS 2014 5 year data, shares and alternative equivalence scales derived by Garner from CE data.
Monthly Average Housing Expenditures Implicit in SPM
Thresholds for Washington DC Metro Area Renters:
1 Person vs 6 People
$2,235
$2,056
$2,168
$1,946$1,922
Six person
18
$793
$944$841
$1,104 $1,152
One Person
SPM
ACS Shares
CE Shares
Garner1
Garner2
Overall SPM Poverty Rates:2015
14.32 14.2 14.26 14.56 15.14
0
5
10
15
20
25
30
35
Overall Poverty Rates
Current Methodology
Shares based on ACS data
Shares based on CE data
Single parameter for allhousing units
Single parameter byhousing type
19
Source: 2015 Current Population Survey Annual Social and Economic Supplement.
Poverty Rates by Consumer Unit
Size: 2015
20
0
5
10
15
20
25
30
35
1 person 2 person 4 person 6 person 7+ person
SPM
ACS Shares
CE Shares
Garner1
Garner2
Source: 2015 Current Population Survey Annual Social and Economic Supplement.
Poverty Rates by Consumer Unit
Age: 2015
21
0
5
10
15
20
25
30
35
Under 18 18 to 64 65+
SPM
ACS Shares
CE Shares
Garner1
Garner2
Source: 2015 Current Population Survey Annual Social and Economic Supplement.
Poverty Rates by Place of
Residence:2015
22
0
5
10
15
20
25
30
35
Principal City Suburbs Outside metro
SPM
ACS Shares
CE Shares
Garner1
Garner2
Source: 2015 Current Population Survey Annual Social and Economic Supplement.
Consumer Units with Housing
Subsidy: 2015
0
5
10
15
20
25
30
35
40
45
50
SPM Rate for Unitswith Housing
Assistance
Percent with $0Subsidy
CurrentMethodology
Shares based on ACSdata
Shares based on CEdata
Single parameter forall housing units
Single parameter byhousing type
23
0
5
10
15
20
25
30
35
40
45
50
Percent with cappedsubsidy
Aggregate Amount ofSubsidies(billions)
Source: 2015 Current Population Survey Annual Social and Economic Supplement.
Marginal Impact of Housing Subsidies on
Poverty Rates: 2015
0.79
0.85 0.830.79
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
Impact of Housing Subsides for AllUnits
Current Methodology
Shares based on ACSdata
Shares based on CEdata
Single parameter forall housin units
Single parameters byhousing type
21.1
22.621.9
21.0 20.9
0.0
5.0
10.0
15.0
20.0
25.0
Impact of Housing Assistance on SPMRate for Units with Assistance
24
Source: 2015 Current Population Survey Annual Social and Economic Supplement.
Negative impact Negative impact
Conclusions and Future Research
Impact on poverty rates Little overall Adjustments through equivalence scales impact rates
Consumer unit size—higher for smaller sizes and lower for larger Elderly rates higher
Impact on housing subsidies and result Adjustments based on equivalence scales
Reduce percentage with $0 subsidy Reduce percentage with capped subsidy
Little marginal impact of housing subsidies on poverty rates
Future Develop a methodology to validate the existing and/or proposed
equivalence scales, consider adults and children Making adjustments for differences in prices for FCSU at local level
and interarea before producing reference unit thresholds
25
Contact Information
Thesia I. GarnerSupervisory Research Economist Divisionof Price and Index Number Research/
Office of Prices and Living Conditions
Bureau of Labor Statistics
http://stats.bls.gov/pir/spmhome.htm
202-691-6576
Trudi J.RenwickAssistant Division Chief
Economic Characteristics
Social, Economic and Housing Statistics Division
U.S. Census Bureau
census.gov
301-763-5133
26