China's (Uneven) Progress Against Poverty

advertisement
Monitoring and Evaluating the Poverty Reduction
and Social Protection:
A Case Study of China
Shaohua Chen
Development Research Group, World Bank
Further reading: Martin Ravallion and Shaohua Chen,
"China's (Uneven) Progress against Poverty: An Update,"
World Bank Policy Research Working Paper, forthcoming
Outline
• Monitoring and evaluating poverty and
inequality in China
• China’s poverty line
• Evaluating social protection program
using household survey data
– Urban Dibao
– Southwest poverty reduction project
1. Monitoring and evaluating
poverty and inequality in China
Trend increase in income inequality but
more so in some periods
60.00
National without adjustment
for cost-of-living difference
50.00
Gini index in %
National, with adjustment
40.00
Rural
Urban
30.00
Signs of that inequality
might be stabilizing at
around Gini=47%
(42% with COL adjustment)
20.00
10.00
0.00
1975
1980
1985
1990
1995
2000
2005
2010
4
Rising inequality between urban and rural
areas since mid 1990s
Ratio of urban to rural mean income
3.50
Ratio of urban mean income to rural mean income
Without any
adjustment for cost-
3.00
2.50
2.00
With adjustment for urban-rural costof-living differential and urban/rural
1.50
1.00
1975
1980
1985
1990
1995
2000
2005
5
2010
Absolute income inequality, 1978-2009
7.00
(divided by 1990 national mean)
Mean income divided by 1990 national mean
6.00
5.00
4.00
Urban
3.00
2.00
National
1.00
Rural
0.00
1975
1980
1985
1990
1995
2000
2005
2010
6
The recent signs of a stabilization of
inequality are due to rural areas
• Rising ratio of
urban mean to
rural mean +
rising urban
inequality
=>The stabilization
is mainly due to
rural areas.
• Pro-poor policies
in rural areas.
40.00
Rural
Gini index (%)
35.00
Urban
30.00
25.00
20.00
15.00
10.00
5.00
0.00
1975
1980
1985
1990
1995
2000
2005
2010
But is this sustainable? There were also signs that inequality
was stabilizing in mid-1990s, but not sustained.
7
Nonetheless, China has made huge
overall progress against absolute poverty
Percentage of population consuming less than
various poverty lines
100.00
Headcount index in %
90.00
80.00
$1.25 a day
70.00
60.00
50.00
40.00
New poverty line
30.00
20.00
Old poverty line
10.00
0.00
1980
1985
1990
1995
2000
2005
8
Headcount index in% under $1.25/day
120
Urban and rural poverty incidence
100
Rural
80
60
National
40
Urban
20
0
1980
1985
1990
1995
2000
2005
2010
9
2. China’s poverty line
-- Overtime comparison
-- International comparison
China’s poverty lines
Yuan/year
$/day
$/day
(2005 prices)
(1993 PPP)
(2005 PPP)
Old official rural poverty line
683
$0.77
$0.63
Rural low income line (2008)
1196
$1.13
$0.97
Comparable rural/urban poverty line (2002)
Rural (average)
925
$1.05
$0.85
Urban (average)
1271
$1.44
$0.85
951
$1.08
International poverty line
Rural
1361
$1.25
Urban
1865
$1.25
Poverty line – Asia
$/person/month in 2005 PPP
year
Cambodia
2004
China
2002
China
2008
India
2004/05
India
2010
Laos
1997.5
Mongolia
2002.5
Philippines
1988
Thailand
1992
Vietnam
2002
pc_05
75.06
120.78
148.18
84.24
131.48
pl_05 Fitted value
42.8
46.33
25.89
62.96
29.32
31.25
50.34
37.01
32.1
.
80.55 57.88
48.61
134.17 46.02
67.91
243.52 57.58
108.41
81.18 32.52
49.45
Poverty line – Country with Similar PC
$/person/month in 2005 PPP
year
Kyrgyz Republic 2003
Djibouti
2002
Kenya
1997
Cameroon
2001
Cote d'Ivoire
1998
China u
2002
Moldova
2001
Philippines
1988
Lesotho
1994.5
Morocco
1998.5
Armenia
1998.5
pc_05
109.85
111.70
112.80
112.96
117.07
120.78
124.89
134.17
135.84
167.73
174.84
pl_05 Fitted value
60.81
58.51
95.61
59.16
84.71
59.50
69.62
59.61
50.36
60.90
25.89
62.96
60.81
64.65
46.02
67.91
49.37
68.73
55.33
82.66
73.36
85.93
Poverty line – Country with Similar PL
$/person/month in 2005 PPP
country
Senegal
Tanzania
Benin
China u
Malawi
Burkina Faso
Nepal
Chad
Mozambique
Rwanda
India
Nigeria
Bangladesh
year
pc_05
1991
78.92
2000.5 45.26
1999.5 72.82
2002
120.78
2004.5 31.34
2003
68.54
2003.5 54.55
1995.5 47.04
2002.5 45.52
2000.1 41.33
2004.5 84.24
1985
61.49
2000
64.34
pl_05 Fitted value
19.05
48.12
19.20
37.82
23.57
45.72
25.89
62.96
26.11
37.16
26.27
44.48
26.43
40.55
26.60
38.23
29.54
37.77
30.17
37.23
31.25
50.34
31.38
42.53
31.46
43.33
China has made huge overall progress
against poverty cross various PL
China: Percentage of population consuming less than various
poverty lines, 1981-2008
100.00
90.00
$2.0 a day
80.00
Headcount index in %
70.00
$1.25 a day
60.00
50.00
40.00
New poverty line
30.00
20.00
Old poverty line
10.00
0.00
1980
1985
1990
1995
2000
2005
15
Policy implications
• China has made huge overall progress against
poverty cross wide range of poverty lines (PL);
• Compare with the countries at same development
level, China’s PL is still low;
• As China’s rapid growth continue, new PL will be
needed every other years;
• It is better to set up several PL for different
programs/time frame.
16
3. Evaluating the impacts of poverty
reduction and social protection program
using household survey data
Impact evaluation: key issues
• the counterfactual – what will happen
without the program;
• Selection bias – the participants of poverty
reduction or social protection program
usually poorer than non participants.
18
Evaluation strategy:
Matching/weighting + diff-in-diff
• Propensity score matching/weighting:
– Matching/weighting on observed initial characteristics
likely to jointly influence poor-area targeting and how
incomes evolve over time.
• Difference-in-difference:
– Difference in gains over time between participants and
non-participants.
– This eliminates any time-invariant bias due to missmatching, selection bias, omitted variables etc
19
Some results from SW poverty
reduction project
20
Diff-in-diff estimate of the impact of SWP on household
income and consumption
1996 mean in
SWP villages
2000 income
971.4
consumption
841.7
saving
129.7
2004 income
971.4
consumption
841.7
saving
129.7
Note: Trimmed sample
Gain in
SWP
project
206.8
67.1
139.7
442.8
345.9
96.8
Gain in
non-SWP
villages
66.0
70.5
-4.5
387.4
287.7
99.7
Diff-inDiff
140.8
-3.4
144.2
55.4
58.3
-2.8
t-ratio
2.02
-0.07
2.33
0.63
0.87
-0.03
Impacts on income and consumption (alternative estimators)
Trimmed sample
2000 Income
Consumption
Saving
2004 Income
Consumption
Saving
Total sample
2000 Income
Consumption
Saving
2004 Income
Consumption
Saving
Simple
DD
t-ratio
PS
weighted
DD
t-ratio
Kernel
matched
DD
t-ratio
140.817
-3.388
144.218
55.433
58.26
-2.815
2.021
-0.067
2.332
0.626
0.87
-0.034
193.162
-17.662
210.84
53.482
58.535
-5.037
2.751
-0.313
2.921
0.574
0.786
-0.059
179.657
-45.762
225.437
52.741
18.312
34.448
2.512
-0.751
2.757
0.694
0.223
0.415
208.583
21.84
186.755
40.673
20.258
20.427
3.346
0.510
3.141
0.537
0.371
0.303
213.605
-151.054
364.696
-47.159
36.752
-83.874
3.287
-1.180
3.371
-0.423
0.633
-0.705
199.342
-189.569
388.953
-38.636
25.893
-64.486
3.426
-1.427
3.494
-0.304
0.439
-0.515
Income poverty
-2
0
2
DD
DD impact
impact on
on headcount
headcount index
index
2004
-8
-6
-4
2000
2004
-14
-12
-10
2000
350
450
550
650
750
808
950
Poverty lines (Yuan per person per year)
Year 2000
Year 2004
1050
1150
Consumption poverty
-2
0
2
DD impact on headcount index
-8
-6
-4
2000
-14
-12
-10
2004
350
450
550
650
750
808
950
Poverty lines (Yuan per person per year)
Year 2000
Year 2004
1050
1150
Some results from China urban
Dibao study
25
1
.8
.6
.4
.2
0
0
5
10
15
lnpinc
icdf11
inccdf
icdf11_d
inccdf_d
26
0
.2
.4
.6
.8
1
Comparison between Beijing and Gansu
0
5
10
15
lnpinc
icdf11
icdf62
icdf11_d
icdf62_d
27
Lessons we learnt from these studies
• Start the monitoring and evaluation as soon
as the project start;
• Choose as many non participants
(comparison group) as possible;
• Using existing data whenever if possible
and link different data (household surveys,
census etc.) together:
– reduce the cost
– increase the coverage
Yunnan:
Prefecture poverty incidence
<5%
5%-10%
10%-15%
15%-20%
20%-30%
>=30%
N=126
Yunnan:
County poverty incidence
<5%
5%-10%
10%-15%
15%-20%
20%-30%
>=30%
N=126
Township poverty incidence
<5%
5%-10%
10%-15%
15%-20%
20%-30%
>=30%
N=1571
Download