Transcript *

Understanding the growth, poverty and inequality nexus:
insights from the Sri Lanka Poverty Assessment*
Ambar Narayan
SASPR
March 23, Washington DC
*Drawing on work by the Poverty Assessment team for Sri Lanka –
including staff of Department of Census & Statistics, Government of
Sri Lanka
Structure of presentation
• Recent poverty trends and patterns in Sri Lanka
• Relating poverty to trends in consumption growth
and distribution
• A focus on rising spatial and rural-urban inequality
• What characterizes poor areas?
• A consequence of spatial inequality: internal
migration
• What are the implications of the pattern of migration
for inequality and growth?
Poverty trends: Modest decline nationally
with widening rural-urban gap
Poverty headcount by sectors
Poverty Headcount: National
26.1
28.8
22.7
30
40
30
20
20
10
10
0
1990-91
0
1990-91
1995-96
1995-96
2002
2002
Urban
Rural
Estate
• Poverty headcount has fallen modestly over the decade for the country as a
whole – excluding conflict-affected areas
• Across sectors, a decidedly mixed picture…….
– Urban poverty has halved: from 16% in 90-91 to 8% in 2002; rural poverty declined
by less than 5 percentage pts
– Increase in estate poverty (although less accurately measured) is significant: from
21% in 90-91 to 30% in 2002
– The declines in national, urban and rural poverty headcount, and the increase in
estate poverty headcount between 90-91 and 2002 are all statistically significant
Large regional differences in poverty
incidence
% of
headcount
population
(%)
Western
33
11
North-Central
7
21
Central
15
25
Northwest
13
27
Southern
14
28
Sabaragamuwa
11
34
Uva
7
37
% of
poor
16
6
16
16
18
17
12
• Much of Sri Lanka’s economic activity is concentrated in Colombo and
surrounding areas (in Western Province). Western Province accounts for
almost half of the country’s GDP.
Rising consumption with widening inequality
Density and Cumulative Distribution of per capita (real) consumption expenditure
.6
.4
0
0
.2
.0001
.0002
.0003
cum_2002/cum_1990
.8
.0004
1
.0005
Pline=Rs. 1423
0
2000
4000
6000
8000
rpcexp_2002
den_2002
10000
0
2000
4000
6000
8000
10000
rpcexp_2002
den_1990
cum_2002
cum_1990
• Increase in inequality between 90-91 and 2002
• Poverty headcount is lower in 2002 for a wide range of poverty
lines
– An aside: note the concentration around the poverty line – indicating
high potential for frequent movements in and out of poverty
Who has benefited from growth?
F igu re 2: G ro w th I n cid en ce C u rv es for Pe r C ap ita C o ns u m ption Exp e nd itu re
10 11 12
Na tio na l (9 5-9 6 to 0 1-0 2 )
9
8
7
6
5
4
3
2
1
0
0
1
2
3
4
5
6
7
8
9
a nn u al gro wth in pc e xp 95 -96 t o 0 1-0 2
10 11 12
Na tio nal (9 0 -9 1 to 0 1 -0 2 )
1
10
20
30
40
50
60
70
% of p op ulat io n ra n ke d by pc ex p
80
90
1 00
10
20
30
40
50
60
70
% of po pu la tion ran ked b y p c ex p
80
90
1 00
80
90
10 0
80
90
10 0
10 11 12
R ura l (9 5 -9 6 to 0 1-0 2 )
9
8
7
6
5
4
3
2
1
0
0
1
2
3
4
5
6
7
8
9
a nn ua l g rowt h in pc ex p 9 5-9 6 to 01 -02
10 11 12
Rura l (90 -9 1 to 0 1 -02 )
1
1
10
20
30
40
50
60
70
% of p op ulat io n ra n ke d by pc ex p
80
90
1 00
10
20
30
40
50
60
70
% o f po pu la tion ran ked b y p ce xp
10 11 12
U rb a n (95 -9 6 to 0 1 -02 )
9
8
7
6
5
4
3
2
1
0
0
1
2
3
4
5
6
7
8
9
a nn ua l g rowt h in pc ex p 9 5-9 6 to 01 -02
10 11 12
Urb an (9 0 -9 1 to 0 1-0 2 )
1
1
10
20
30
40
50
60
70
% of p op ulat io n ra n ke d by pc ex p
80
90
1 00
1
10
20
30
40
50
60
70
% o f po pu la tion ran ked b y p ce xp
Consumption
increased
disproportionately among the
better-off, in both
urban and rural
areas
The impact of growth and inequality on poverty
• Rapid increase in average consumption with rising inequality
– Mean per capita real consn. increased by 29% from 90-91 to 2002, while
gini increased by almost 24% (by 19% and 30% for urban and rural)
– Mean consumption increased by 50% for the top quintile, and by 2% for the
bottom quintile
– GICs show that the consumption benefits from 90-91 to 2002 accrued
largely among the better-off – in both urban and rural areas
• How has rising inequality affected poverty reduction?
– With no change in distribution, growth in average per capita consn. would
have reduced poverty by 15 percentage pts. between 90-91 and 2002 (by 12
and 18 in urban and rural areas)
– Contrast with the observed reduction of only 3 percentage pts. in poverty
headcount (by 8 and 5 in urban and rural areas)
• IF consumption Gini grows at the average rate of increase seen
from 90-91 to 2002, growth elasticity of poverty is only 0.9
– Sri Lanka will need to grow at an annual average rate of around 10% to
achieve the MDG target of halving poverty by 2015 (current rate is 5.7%)
– With constant Gini, poverty headcount will be more than halved to around
8% by 2015
Decomposition of inequality index
Decomposition of Theil Inequality index
% increase
90-91 2002 from 90/91
to 2002
National
16.8
26.4
57
Within-district
15.4
23.3
52
Inter-district
1.5
3.1
112
Within-sector
15.5
24.4
57
Inter-sector
1.3
2.0
52
•A substantial increase in inequality – by 50 percent or more – within
and between districts as well as within and between sectors.
•While inequality within districts or sectors contributes much more to
aggregate inequality than that between sectors or districts, the
percentage increase in inter-district inequality was by far the highest
Inter-district inequality has deepened over the last decade
•
Over the past decade, a tendency towards wider disparity across districts
– Between 90-91 and 2002, poverty reduction has been substantial for the three districts in Western
Province (Colombo, Kalutara, Gampaha).
– More remote and rural districts experienced no reduction or even increase in poverty headcount –
four registered increases of 10 % or above
– 5 poorest districts in 1990-91 experienced increase or no reduction in poverty by 2002
•
Wider regional disparity also apparent from the provincial shares in national GDP.
– Western Province’s share in national GDP increased from 40 % in 1990 to 48 % in 2002, while that
of all the other provinces (except Southern province) declined by between 1 and 4 % points; the
share of Uva, the poorest province, halved
Rising urban-rural gap: stagnation in agriculture
Fig. 4: Contribution to GDP (at constant 1996 prices) by
Sector
Fig. 5: Real per capita GDP and sectoral outputs
800
4.75
4.25
Agriculture
Industry
Services
GDP
Agriculture
Industry
2002
2001
2000
1999
1998
1997
1996
1991
2002
2001
2000
1999
1998
1997
1996
1995
1994
1993
3.5
1992
0
1991
3.75
1995
4
200
1994
400
4.5
1993
600
1992
log of Rs.
5
Rs. (billion)
1000
Services
Source: Central Bank of Sri Lanka Annual Report (multiple issues)
% growth in incomes among rural and agricultural households from 1995/96 to 2001/02
•
•
Agricultural output per capita did not
grow over the decade while still
employing a third of population
(higher in rural areas)
Poorest rural and agricultural
households suffered a decline in real
incomes – consistent with increase in
rural gini
40%
Percent Growth in Houshold
Income
Stagnation in agriculture
30%
Rural
Agric
20%
10%
0%
0
10
20
30
40
50
60
-10%
-20%
-30%
Income Percentile
70
80
90
100
What characterizes poor areas
• Perhaps not surprising for those who know Sri
Lanka, but instructive…….
– A household is more likely to be poor if it belongs to a
district with (controlling for a wide range of householdspecific attributes)
• lower average access to markets
• lower proportion of households using electricity
• higher proportion of household heads with education below
primary level or employment as agricultural workers (when the
accessibility index is omitted)
– Spatial factors also important at a more disaggregated level
• Replacing district level indicators with DS division level averages
yields similar results (a unit improvement in average accessibility
index of a DS division reduces the probability of a household
located there to be poor by 12 percent)
Poverty Map at DS division level
Accessibility Index
Regional characteristics matter for
poverty in Sri Lanka
• Poverty associated with geographic isolation - areas with
limited access to markets, infra and lower endowment of
human capital
– But these are more useful in explaining differences in poverty
between Western Province and outside, and less so for differences
between areas outside of Western Province
• Suggests a stylized story of a “dual” economy circumscribed
by geography
– Western Province (incl. Colombo) with better access to markets and
infrastructure, higher concentration of educated people, and predominance of non-agricultural sectors
– The rest of the country, including 6 conflict-affected districts for
which little data is available, where largely the converse of these
conditions hold
Dynamic consequence of spatial inequality:
migration
• High and increasing rates of internal migration in Sri Lanka
– Household data shows doubling of internal migration from 96-97 to 03-04
– Out-migration from all provinces except WP and all sectors but urban
increased: WP and the urban sector are thus the main recipients
– Consistent with expanding gap between WP and the rest – rising gap in
monthly wages even for elementary occupations
• Migration can be a viable option for the poor in remote rural areas
to lift themselves out of poverty
– Internal migration to urban growth centers can also be a growth vehicle
through fostering urban agglomeration (WDR 2000)
• Potential impact of migration on inequality and growth
– Loss of human capital in hinterlands, further widening spatial inequality
– Over-concentration in urban areas can eventually place a limit on economic
growth
Analysis of internal migration in Sri Lanka
• Lack of panel data; or even a household survey with a detailed
migration module
• Comprehensive use of available data…….
– Household surveys: Information on out-migrants and their households at
the “origin”– doesn’t cover those who migrated with their household
– Population Census 2001 as a valuable source of information: on migrants
and recent migrants at the “destination”
• ….. And some experimentation
– Small area estimation method for estimating poverty incidence of migrants
in Colombo MC
– Quantifying the cost of over-concentration in Colombo MC – drawing on
cross-country results
• Given data limitations, a static and thus partial analysis,
but one that still offers useful insights
Characteristics of migrants into Colombo
Figure: Share of household heads with tertiary
education – migrants and residents by origin district
80
•Recent migrants in Colombo
city more likely to belong to
poorer and conflict-affected
districts
60
40
migrants by origin district
Kegalle
Ratnapura
Moneragala
Badulla
Polonnaruwa
Anuradhapura
Puttalam
Kurunegala
Hambantota
Matara
Galle
Matale
Kandy
Kalutara
Gampaha
0
Nuwara Eliya
20
•But also likely to be more
educated - loss of human
capital from lagging areas?
residents by district
Figure: Educational attainments and occupation for migrants, recent migrants and non-migrants in
Colombo city
% of elementary occupation among workers
40
% of individuals with tertiary education
60
35
51
32
27
30
Presence of both “push”
and “pull” factors
50
40
26
20
20
10
0
•Migrants have better jobs and
lower poverty incidence than
long-term Colombo residents
0
all
RECENT
migrants
all
migrants
Non
migrants
all
RECENT
migrants
all
migrants
Non
migrants
•The role of “networks” in
influencing migration patterns
– an open question
Likely impact of internal migration
• At the place of “origin”
– Better-educated people are more likely to migrate due to
lack of employment opportunities
– No indication that migration is a positive force for education
– But remittances are somewhat effective in reducing poverty
among households with migrants
• While migration is an avenue out of poverty for those
in lagging areas
– More of an option for those with better endowments, and
sometimes legal status (those with NICs in plantations)
• A force perpetuating regional disparities?
– Loss of human capital likely to reduce growth prospects
(and political power) in lagging regions
Impact of internal migration on Colombo
urban area
• On the one hand, benefits due to agglomeration – increasing returns to
scale
• But growing concern about over-concentration
– Increasing population density in Colombo MC (17,000/sq. km. in 2001); population
during daytime expands 3-fold: around 1.5 million people commute every day
– Residential land prices increased more than 100 % in even poor areas between
2000 and 2005
– Net out-migration from Colombo District to the neighboring districts, especially
among poorer workers
• Aggregate impact on economic growth
– Estimated growth losses due to over-concentration (based on cross-country model
in Henderson 2000): 1.5 % annual growth rate – just indicative, but instructive
– Urban primacy of Colombo is 35-50%, compared with an “optimal” primacy of
25% for a country of Sri Lanka’s per capita GDP
• A cyclical relationship between inequality, migration and growth?
– Increasing regional disparities lead to higher incentives to migrate – can perpetuate
regional inequalities and lead to over-concentration in the main urban area, which
imposes limits on economic growth
Concluding thoughts and policy questions
• Enhancing mobility among households in rural/remote areas
– Improve access to quality education
– Address land policy distortions that tie people to their land
• Better links to markets and infrastructure in rural/remote areas
– The conundrum: Sri Lanka in fact has high road density for a developing nation
– But poor maintenance; and lack of planning – e.g. focus on rural roads is much
more on links between remote villages rather than to markets or growth centers
• How to plan infrastructure development to promote alternative growth
centers to reduce primacy of Colombo?
– Recognize agglomeration benefits of Colombo Metropolitan area – may not be
efficient to attempt to replicate the infrastructure of Colombo; is it better to
identify specific comparative advantages of potential growth centers – more
likely in areas neighboring Colombo – and invest on appropriate infrastructure?
• Coordinating urban planning and rural development is key to optimal
allocation of resources
– To reduce costs of over-concentration– allow urban land markets to operate;
strengthen local municipalities; develop integrated plan for Greater Colombo
area
– The classic question of equity vs. efficiency: the tradeoff between investing
scarce resources in projects to uplift remote areas, or investing in and promoting
links to urban growth centers?