The Spatial Distribution of Urban Poverty: Metropolitan Areas versus Small Towns Peter Lanjouw (World Bank) Urbanization and Poverty and Poverty Reduction: Bridging Urban and Rural Perspectives World Bank, May 13-14, 2013
Feb 22, 2016
The Spatial Distribution of Urban Poverty: Metropolitan Areas versus Small Towns
Peter Lanjouw (World Bank)
Urbanization and Poverty and Poverty Reduction: Bridging Urban and Rural
PerspectivesWorld Bank, May 13-14, 2013
Unpacking urban poverty in a selection of developing countries
• Why do we know so little about spatial distribution of urban poverty?
• Small area estimation of poverty• Urban poverty across city size in 9 countries• 4 Robustness checks:– Cost of living differences– Non-monetary indicators of poverty– Subjective welfare– Artifact of SAE methodology?
• Why do we observed a poverty-city size gradient?– Limited access to infrastructure services in small towns.
• Is Africa Different? Preliminary Evidence
Heterogeneity of Urban Poverty:Why do we know so little?
• Main source of information on distributional outcomes - household surveys - permit only limited disaggregation.
• Very large data sources (e.g. census) typically collect very limited information on welfare outcomes.
• Our solution: Small Area Estimation (ELL, 2002; 2003)– Impute a measure of welfare from household survey
into census, using statistical prediction methods.
International Experience• Poverty Maps are completed or underway in around
50+ countries– Including Brazil, China, India, Indonesia
• In some countries poverty maps available for multiple periods in time– Ecuador, Panama, Paraguay, Morocco– RIMISP project has produced panel poverty maps across
LAC• PovMap2 Software development by WB– Available at: http://iresearch.worldbank.org
Small Area Estimates of Urban Poverty:8 countries
Albania Brazil Kazakhstan Kenya Mexico Morocco Thailand Sri Lanka
Urban Population
1.3m 125m 8.2m 5.0m 54.5m 12.7m 18.5m 2.2m
Urban Population %
0.58
0.83
0.57
0.19
0.60
0.51
0.31
0.12
Census Year 2001 2000 1999 1999 2000 1994 2000 2001
Survey Year 2002 2002-3 2001 1997 2000 1998 2000 2002
Survey Name
LSMS POF HBS WMS III
ENIGH ENNVM SES HIES
Survey Sample Size
3,600 48,470 11,883 10,874 10,108 5,184 24,747 20,100
Poverty Line ALL 4,891
BRL 100
KZT 3,157
KES 2,648
PES 768
DHS 3,400
BAH 1370
LKR 1423
Poverty Line (2005 PPP$)
95 83 63 147 128 57 88 47
Equivalence Scale
No No No Yes No No No No
Box-Plot of Poverty Rate by City Size:Mexico and Thailand
Box-Plot of Poverty and City Size: Vietnam
Urban Poverty: SummaryCountry Reference Year Incidence of Poverty (%) Share of small town in
urban population (%)Share of small town poor in urban poor (%)
Smallest town size category
Largest city size category
Brazil 2000 30 9 33 55
Mexico 2000 31 18 10 18
India 2004/5 30 15 73 84
Sri Lanka 2001 12 8 33 40
Albania 2001 20 11 33 35
Kazakhstan 1999 19 3 26 35
Thailand 2000 14 2 45 76
Vietnam 2009 12 1.4 27 55
Morocco 1994 16 4 6 9
Robustness
• Are our findings of a poverty-city size gradient robust?– cost of living variation–non-income dimensions of wellbeing– Subjective welfare– Statistical artifact?
Spatial Price Variation
• Household surveys rarely permit construction of price indices that distinguish between city size
• Evidence from Brazil indicates that adjusting for price variation does attenuate the poverty city size gradient
• But overall gradient remains– Ferre et al (2012)
Non-Income Dimensions
• It is sometimes suggested that gradient disappears for non-income dimensions such as child-health
• Evidence from Mexico indicates presence of a similar gradient based on SAE estimates of child malnutrition (anthropometrics)– Lanjouw and Rascon (2011)
Subjective welfare
• Evidence from Vietnam indicates that after controlling for access to services, average welfare, as well as households’ real expenditure levels, households in large cities have lower subjective wellbeing than households in small town.– Lanjouw and Marra (2013)
• Consistent with concept of relative welfare, and also with presence of externalities such as pollution and congestion in large towns.
Does SAE drive finding?
• Evidence in India, based on NSS sample survey data, is consistent with SAE results.– World Bank (2010)
• Difficult to test elsewhere due to lack of appropriate survey data.
Why a poverty gradient?
• Internal dynamism of large cities– Agglomeration externalities• “new” economic geography
– Scarcity of competent city planners and administrators
• Neglect of small towns?– Skewed distribution of infrastructure availability• Ferre et al (2012) document lower per-capita service
availability in small towns
Service delivery and town size in IndiaPublic spending in India is skewed against small towns:
Andhra Pradesh and Orissa Town class (in
‘000s)
No. of latrines (water borne)
Domestic electricity
connections
Hospital beds
Piped Water
primary drinking
water source
Total Revenues
Gov’t Grants
Total Expenditures
Public Works
(per 1000 population)
(% of towns)
(Rs per capita)
<50 99.5 125.2 0.88 48.5 461.6 285.2 430.2 80.0 50-100
96.0 135.9 2.40 76.5 577.5 272.0 509.0 113.6
100-500
90.5 126.4 1.78 76.5 638.1 248.4 552.4 119.4
>500 145.0 186.0 2.04 100.0 677.2 164.5 522.2 166.6 Total 98.1 127.6 1.25 57.7 509.7 275.8 462.6 92.3
Urban Poverty & City Size in Africa: Some Preliminary results
Table 1: Data Sources Mali Swaziland Togo Malawi Survey ELIM 2009/10 SHIES 2009/10 QUIBB 2011 IHS (2010/11) # households 9,035 3,191 5,491 12,268 # individuals 86,492 14,140 29,676 55,985 Poverty Headcount (national povline)
43.6% 63.4% 58.7% 50.7%
Census 2009 2007 2010 2010 # households 2,359,356 212,100 1,206,957 2,857,704 # individuals 14,701,684 1,017,526 5,847,453 12,922,799 % urban (indiv.) 22.7% 21.8% 37.3% 85.1% Administrative Structure
Region (9/9/8)
Region (4/4/4)
Region (6/6/5)
Region (3/3/3)
(# Total/Urban/Rural) Cercle (50/25/49)
Inkhundla (55/23/52)
Prefecture (41/41/36)
District (32/29/28)
Commune (703/37/666)
Major Area (195/27/181)
Canton (386/48/372)
TA/Ward (353/137/216)
CCQ (513/141/372)
Senegal Gabon Niger Guinea Survey ESPS 2005 EGEP 2005 QUIBB 2005 EIBEP (2002) # households 13,568 7,865 6,690 7,095 # individuals 122,000 37,400 43,000 53,547 Poverty Headcount (national povline)
52.6% 32.8% 61.8% 48.8%
Census 2002 2003 2001 1996 # households 1,075,918 267,085 1,632,582 1,029,000 # individuals 9,961,678 1,183,199 11,179,668 7,200,000 % urban (indiv.) 40.7% % 16.3% 29.6% Administrative Structure
Region (11/11/11)
Province (9/9/9)
Region (8/8/7)
Region (8/8/7)
(# Total/Urban/Rural) Department (35/35/32)
Department (48/48/47)
Department (37/37/36)
Prefecture (34/34/33)
Collectivite locale
(426/106/320)
Canton (219/68/151)
Canton (173/42/131)
Commune (340/40/331)
Côte d’Ivoire CAR Mauritania Sierra Leone Survey ENV 2002 ECASEB 2008 EPCV 2004 SLIHS (2003) # households 6,897 9,385 3,720 # individuals 32,000 52,600 22,970 Poverty Headcount (national povline)
38.5% 61.9% 46.3% 67.0%
Census 1998 2003 2000 2004 # households 2,700,000 642,839 352,349 788,210 # individuals 15,400,000 3,102,440 2,049,594 4,769,701 % urban (indiv.) 42.2% 37.8% 42.2% 37.5% Administrative Structure
Region (19/19/19)
Region (7/7/6)
Wilaya (13/13/12)
Province (4/4/4)
(# Total/Urban/Rural) Department (58/58/58)
Prefecture (17/17/16)
Moughata (53/34/43)
District (14/14/14)
Sous-Prefecture (254/127/253)
Sous-Prefeture (79/44/71)
Commune (216/38/204)
Chiefdom (166/91/164)
Secteur (444/136/426)
Commune (177/46/158)
Table 2: City/Town Size Distribution (in unit) Mali Swaziland Togo Malawi XXS (<5,000) 2 18 6 6 XS (5,000 to 10,000) 6 2 9 4 S (10,000 to 25,000) 8 11 14 M (25,000 to 50,000) 5 1 3 4 L (50,000 to 100,000) 5 2 6 1 XL (100,000 to 1,000,000) 5 1 3 XXL (>1,000,000) 1 1 Total 32 23 37 32
Senegal Gabon Niger Guinea XXS (<5,000) 4 30 1 4 XS (5,000 to 10,000) 16 6 6 7 S (10,000 to 25,000) 25 6 22 10 M (25,000 to 50,000) 5 5 5 8 L (50,000 to 100,000) 5 3 5 XL (100,000 to 1,000,000) 5 2 3 1 XXL (>1,000,000) 1 1 Total 61 49 40 36
Côte d’Ivoire CAR Mauritania Sierra Leone XXS (<5,000) 5 6 37 XS (5,000 to 10,000) 5 10 12 25 S (10,000 to 25,000) 16 16 7 12 M (25,000 to 50,000) 15 5 3 5 L (50,000 to 100,000) 13 1 1 2 XL (100,000 to 1,000,000) 8 2 1 3 XXL (>1,000,000) 1 Total 58 29 30 84
Urban Poverty in AfricaCountry Reference Year Incidence of Poverty (%) Share of small town in
urban population (%)Share of small town poor in urban poor (%)
Smallest town category Largest city size category
Kenya 1999 49 44 21 25
Mali 2009 29 9 0.2 0.3
Swaziland 2007 37 31 22 24
Togo 2010 43 29 3.3 5.1
Malawi 2010 28 17 2 3.8
Senegal 2002 51 34 3.4 4.6
Gabon 2003 43 26 6.4 9.0
Niger 2001 56 37 2.4 3.4
Guinea 1996 17 24 0.6 0.6
Cote d’Ivoire 1998 34 15 0.7 1.0
Central African Republic
2003 58 48 1.3 1.2
Mauritania 2000 39 29 2.6 3.2
Sierra Leone 2009 71 36 5.3 7.5
Table 4: Non-monetary Indicators for different town/city sizes for four African Countries
Em
ploy
men
t R
ate
Self-
Empl
oy.
Rat
e
Net
Sch
ool
enro
l. Pr
imar
y
Net
Sch
ool
enro
l. Se
cond
.
Lite
racy
Rat
e
Girl
/Boy
ratio
- Pr
imar
y
Girl
/Boy
- Se
cond
ary
Impr
oved
sa
nita
tion
Impr
oved
Wat
er
Supp
ly
Elec
trici
ty
Gas
/ele
c/C
oal a
s co
okin
g fu
el
MALI Rural 60.9 33.5 41.0 15.4 28.0 0.77 0.38 75.2 24.5 8.8 11.9
Urban 50.1 44.4 75.0 42.1 63.0 0.95 0.62 97.0 63.9 65.8 43.7 XXS 50.1 28.6 58.4 12.9 31.0 0.47 0.09 96.3 52.8 26.9 4.8 XS 56.7 34.1 50.8 14.6 27.2 0.59 0.29 94.2 51.9 13.9 3.9 S 52.8 41.8 52.1 28.0 45.4 0.87 0.46 86.9 55.3 37.6 20.7 M 47.5 45.5 68.5 42.1 56.1 0.93 0.54 89.7 69.4 49.5 21.7 L 44.8 41.5 72.8 43.8 66.6 0.93 0.63 92.8 61.6 50.9 24.5 XL 50.6 44.3 74.7 42.6 62.6 0.93 0.56 97.7 50.5 52.8 15.9 XXL 50.9 45.3 78.4 43.3 65.4 0.98 0.68 99.0 70.3 79.1 63.8 SWAZILAND
Rural 22.5 27.4 77.9 32.9 94.9 0.93 1.04 74.3 49.6 21.2 15.3 Urban 51.3 20.2 66.9 40.8 97.7 1.03 1.14 88.5 86.9 64.9 87.1 XXS 51.6 13.6 67.7 41.6 96.8 1.03 1.11 79.9 85.5 71.9 74.8 XS 60.1 11.2 64.7 41.5 95.8 1.05 1.03 86.1 93.1 58.6 88.6 M 50.7 22.7 58.3 33.6 98.2 1.03 1.19 91.3 86.2 62.2 93.1 L 50.3 23.2 70.1 43.3 98.1 1.02 1.15 91.3 87.1 63.9 89.6
Table 4: Non-monetary Indicators for different town/city sizes for four African Countries
Em
ploy
men
t R
ate
Self-
Empl
oy.
Rat
e
Net
Sch
ool
enro
l. Pr
imar
y
Net
Sch
ool
enro
l. Se
cond
.
Lite
racy
Rat
e
Girl
/Boy
ratio
- Pr
imar
y
Girl
/Boy
- Se
cond
ary
Impr
oved
sa
nita
tion
Impr
oved
Wat
er
Supp
ly
Elec
trici
ty
Gas
/ele
c/C
oal a
s co
okin
g fu
el
TOGO Rural 72.7 91.1 76.9 32.6 64.7 0.86 0.45 20.9 28.3 8.5 10.6
Urban 62.5 69.5 80.6 57.5 86.3 1.07 0.81 83.6 78.2 75.6 86.9 XXS 59.7 74.8 83.8 60.5 87.4 1.01 0.55 58.6 39.6 45.6 34.7 XS 59.5 77.2 82.6 59.3 82.1 0.96 0.52 40.4 33.7 46.5 36.0 S 59.4 75.9 83.5 56.8 87.5 0.99 0.57 54.1 66.2 67.1 54.5 M 63.0 76.5 80.3 46.3 71.7 0.98 0.57 46.6 51.0 58.1 71.7 L 57.3 70.6 83.1 62.9 89.6 1.06 0.73 72.4 65.6 72.1 78.8 XL 66.1 68.3 80.0 54.9 84.9 1.10 0.90 96.7 90.4 85.4 97.3 XXL 62.7 67.6 79.4 58.1 87.4 1.07 0.87 91.0 82.7 75.6 94.7 MALAWI
Rural 58.9 84.6 72.1 7.3 82.2 1.02 0.87 2.1 70.0 1.9 4.4 Urban 54.4 47.3 80.1 30.7 96.1 1.03 1.01 20.5 94.0 37.8 58.2 XXS 51.6 42.9 85.1 33.4 96.4 1.02 0.98 19.7 95.1 38.4 24.7 XS 52.8 51.2 82.3 30.0 94.1 1.01 1.09 19.1 97.2 34.7 29.2 S 53.7 60.6 77.9 27.7 94.5 1.03 0.99 16.6 96.0 29.0 34.0 M 53.3 63.2 77.8 24.1 93.8 1.03 1.02 18.3 95.7 27.0 32.5 L 52.7 39.2 80.6 34.8 96.8 1.05 1.00 37.4 97.5 49.5 50.9 XL 54.7 43.9 80.6 31.6 96.6 1.03 1.02 20.5 93.2 39.7 66.1
Should policymakers focus on small towns?
• How best to intervene?– Target delivery of improved services?– But • possibly higher per capita cost of providing access• no guarantee that improved service delivery will result
in economic growth of towns.
• One additional argument:– Small town development may stimulate rural non-
farm development