Rural Poverty, Employment and Agricultural Growth: A

Recent Evidence on Poverty and
Inequality in India
Abhijit Sen and Himanshu
Inequality in India had decreased in the 1950s, but per capita
consumption was relatively stagnant till the mid 1970s, when India
broke out of the “Hindu rate of growth”. The 1980s not only saw
over 5% growth but also some return to declining inequality.
36.00
35.00
34.00
33.00
32.00
31.00
30.00
87
88
89
90
91
86-Jun
87-Jun
88-Jun
89-Jun
90-Jun
Jul
Jul
Jul
Jul
Jul
Jan 83-Dec 83
Jul 77-Jun 78
Oct 72-Sep 73
Oct 73-Jun 74
Dec
Aug
Mar
Sep
Jul
Jul
Jul
Sep
Feb
Jul
Jul
Jul
Jul
Jul
Jul
Jul
55-May
56-Feb
57-Aug
57-May
58-Jun
59-Jun
60-Aug
61-Jul
63-Jan
64-Jun
65-Jun
66-Jun
67-Jun
68-Jun
69-Jun
70-Jun
56
57
57
58
59
60
61
62
64
65
66
67
68
69
70
71
29.00
55-May
56-Feb
57-Aug
57-May
58-Jun
59-Jun
60-Aug
61-Jul
63-Jan
64-Jun
65-Jun
66-Jun
67-Jun
68-Jun
69-Jun
70-Jun
56
57
57
58
59
60
61
62
64
65
66
67
68
69
70
71
Jul
Jul
Jul
Jul
Jul
86-Jun
87-Jun
88-Jun
89-Jun
90-Jun
87
88
89
90
91
Jan 83-Dec 83
Jul 77-Jun 78
Oct 72-Sep 73
Oct 73-Jun 74
Dec
Aug
Mar
Sep
Jul
Jul
Jul
Sep
Feb
Jul
Jul
Jul
Jul
Jul
Jul
Jul
Consequently, Indian poverty which had no trend up to 1973-74
declined quite significantly from then to the end of the 1980s.
The rate of poverty decline during this period was at well over
1% point per annum.
65.00
60.00
55.00
50.00
45.00
40.00
35.00
30.00
In 1991 India began her process of liberalisation. Since then there has been
some confusion regarding what happened to poverty. Most estimates showed
no poverty decline till 1999-00 but then the next round showed a very sharp
drop in poverty. This led to a lively debate on what had exactly happened.
Poverty Headcounts (Urban+Rural)
45
Datt-Ravallion
40
35
Official-Gupta
30
Sundaram-Tendulkar
25
20
Deaton-Tarozzi
1999-00
1997
1995-96
1994-95
1993-94
1992
1990-91
1989-90
1988-89
1987-88
1986-87
15
The main aspect was a change in survey design to grapple with the
fact that India’s NSS surveys which measure poverty and
inequality were increasingly diverging from the national accounts.
Ratio of nominal consumption expenditure
estimates from NSS to that from NAS (1980-81)
0.85
0.80
0.75
0.70
1997
1995-96
1994-95
1993-94
1992
1990-91
1989-90
1988-89
1987-88
1986-87
1983
1977-78
0.65
However, the NSS-NAS difference was not unique to
India:



NA-survey differences exist elsewhere also.
The NAS-NSS gap in India is similar to
that in the USA, where in fact the
corresponding divergence is increasing
even faster.
Moreover, the NSS shortfall from NAS is
concentrated very largely in urban areas
and follows a pattern that is consistent
with an increasing share of incomes
accruing to the relatively rich, whose
consumption the NSS is unable to capture
fully.
Item-wise comparison of NSS and NAS
consumption estimates show NSS falling
short mainly on items which are either not
consumed much by the poor or where NAS
overestimation is more likely than NSS
underestimation (Kulshresta-Kar 2003).
Nonetheless, the NSS which had experimented with
alternative reference period during 1994-98 decided
to make a change
In rounds 51 to 54, two schedules were canvassed to independent
samples from the same population:
 Schedule 1 used the conventional uniform 30 day recall for all items
 Schedule 2 used 7 day recall for food and intoxicants and 365 day
recall for five items with low frequency of purchase
 The Results:
 The 7 day recall for food and intoxicants returned about 30% more
consumption of these items than the 30 day recall
 The 365 day recall for low frequency items returned slightly lower
mean expenditure on these than the 30 day recall, but returned a
much more equal distribution – leading to Gini ratios for overall
consumption about 5 points lower
Both effects reduced measured poverty. Schedule 2 returned only
about half the poverty headcount estimated from Schedule 1

The massive fall in poverty in 1999-00 was due to this change. This was
accepted by most analysts who accepted that poverty decline was
therefore exaggerated but nonetheless disagreed on its extent. Our
analysis showed that this is how schedule change could have affected the
distributions obtained.
URBAN
10
10
5
5
% difference
% difference
RURAL
0
-5
-10
0
-5
-10
-15
-15
-20
-20
Poor 40%
Mid 40%
Rich 20%
Poor 40%
Mid 40%
Rich 20%
Round 38
Round 43
Round 50
Round 38
Round 43
Round 50
Round 51
Round 52
Round 53
Round 51
Round 52
Round 53
The NSS has only gone back to its original schedule in
its most recent survey which gives the following picture:
Poverty has reduced but the pace of poverty reduction has slowed down.
Rural+Urban Poverty Incidence
10
uniform 30 day
1980s trend
mixed 30/365 day
MRP
15
2004-05
15
2003
20
2001-02
20
2000-01
25
1999-00
25
1997
30
1995-96
30
1994-95
35
1993-94
35
1992
40
1990-91
40
1989-90
45
1988-89
45
1987-88
50
1986-87
50
1983
55
1977-78
URP
(% population below official poverty line)
Till the most recent round there was some confusion on inequality. At first sight, the
data suggests a declining trend in the rural Gini (left chart). But, allowing for schedule
change (right chart), these had implied that an earlier inequality decline was reversed
very sharply after mid-1990s. This was not however, not generally accepted.
32
34
28.5
33
27.5
32
26.5
31
25.5
30
24.5
29
23.5
28
22.5
27
21.5
31
30
29
URP
27
26
25
24
MRP 365 & 30
198 3
1986-87
1987-88
1988-89
1989-90
1990-91
1992
1993-94
1994-95
1995-96
1997
1999-00
2000-01
2001-02
2003
MRP 365 only
uniform 30 day
198 6-87
198 7-88
198 8-89
198 9-90
199 0-91
199 2
199 3-94
199 4-95
199 5-96
199 7
199 9-00
200 0-01
200 1-02
200 3
URP
1983
1977-78
22
197 7-78
23
trend
mixed 30/365 days
MRP
28
41
36.2
37
40
35.2
36
39
34.2
35
38
33.2
37
32.2
36
31.2
35
30.2
34
29.2
33
28.2
URP
38
34
33
32
MRP 365 & 30
1983
2003
2001-02
1997
1999-00
2000-01
1994-95
1995-96
1993-94
1990-91
1992
1989-90
1988-89
1986-87
1987-88
MRP 365 only
uniform 30 day
1986-87
1987-88
1988-89
1989-90
1990-91
1992
1993-94
1994-95
1995-96
1997
1999-00
2000-01
2001-02
2003
URP
1983
1977-78
30
1977-78
31
trend
mixed 30/365 days
MRP
Similarly, although urban Ginis appear flat at first sight, these could be
interpreted as showing very large increase on allowing for schedule change.
Moreover, these alternative Gini movements were almost synchronous with what
Banerjee-Piketty (2003) had reported with independent income tax data. But
nonetheless, again were not universally accepted.
The most recent data which now allows
comparability does in fact show that inequality has
increased quite sharply.
Rural Gini
32
Urban Gini
29.5
31
28.5
30
40
37.5
39
36.5
38
35.5
37
34.5
36
33.5
35
32.5
34
31.5
33
30.5
27.5
29
26.5
28
25.5
1977-78
1983
1987-88 1993-94 1999-00 2004-05
URP
MRP
1977-78
1983
1987-88 1993-94 1999-00 2004-05
URP
MRP
Given the NAS-NSS difference it is useful to go back to NAS data.
Although the NAS is silent on personal distribution, it does give:
Factorial Distribution of National Income: 1993-94 (inner circle) and
2002-03 (outer circle)
4%
11%
8%
5%
8%
7%
6%
7% 5%6%
21%
28%
16%
18%
25%
25%
Ag w ages
Oth inf ormal w ages
Self -emp Ag
Self -emp Non-Ag
Govt salaries
Pvt Org Salaries
Govt Surplus
Pvt Org Surplus
Corresponding distributions of workforce can be obtained from the population
Census and DGET, using NSS Employment-Unemployment Surveys only to
break down self-employed and employees in unorganised non-agriculture:
Distribution of Work Force
5 2
18
62
19
27
24
14
35
17
32
Ag wages
Oth informal wages
Self-emp Ag
Self-emp Non-Ag
Govt Salaries
Pvt Org salaries
Together, these throw up the following pictures on distributional
changes implicit in the NAS, completely independent of NSS
Consumer Expenditure Surveys:
Per Worker Participation Incomes
Indices of real per worker incomes
(1993-94=100)
(monthly, at constant 1993-94 prices)
Govt salaries
220
Pvt Org
Salaries
200
Self-emp NonAg
180
Government
Employees
Pvt Org.Sector
Employees
Self-emp Nonagri
160
Self-emp Ag
140
Oth informal
wages
Self-emp
Agriculture
120
Ag wages
Other Informal
Labour
2002-03
2001-02
2000-01
1999-00
1998-99
1997-98
1996-97
1995-96
1994-95
9000
8000
80
1993-94
1993-94 2001-02
7000
6000
5000
4000
3000
2000
1000
0
100
Agriculture
Labour
In the light of this, it is useful to again look at the
long series NSS and examine different aspects of
inequality.
Between States Inequality: Rural
uniform
30 days
(population weighted std deviation of logs of nominal mpce)
0.2
0.16
mixed
30/365
days
0.12
polynomial
fit
0.08
kinked
trend
15
20
25
30
35
40
NSS Rounds
45
50
55
Between States Inequality: Urban
(population weighted std deviation of logs of nominal mpce)
uniform
30 days
0.2
mixed
30/365
days
0.16
polynomial
0.12
fit
0.08
kinked
trend
15
20
25
30
35
40
NSS Rounds
45
50
55
Within-State Urban-Rural Disparity
population weighted averages of state level ratios of urban to rural nominal mpce
1.8
1.7
uniform 30 days
1.6
mixed 30/365
days
1.5
polynomial fit
1.4
kinked trend
1.3
1.2
15 18 21 24 27 30 33 36 39 42 45 48 51 54 57
NSS Rounds
1986-87
1987-88
1988-89
1989-90
1990-91
1992
1993-94
1994-95
1995-96
1997
1999-00
2000-01
2001-02
1983
1977-78
1959-60
1960-61
1961-62
1962-63
1964-65
1965-65
1966-67
1967-68
1968-69
1969-70
1970-71
1972-73
URP
25
29
24
28
23
27
22
26
21
25
MRP
Within-State Rural Inequality
population weighted averages of state level ginis
33
28
32
27
31
uniform 30
day
26
30
mixed
30/365 days
polynomial fit
kinked trend
1993-94
1994-95
1995-96
1997
1999-00
2000-01
2001-02
1986-87
1987-88
1988-89
1989-90
1990-91
1992
1983
1977-78
1959-60
1960-61
1961-62
1962-63
1964-65
1965-65
1966-67
1967-68
1968-69
1969-70
1970-71
1972-73
URP
40
35
39
34
38
33
uniform 30
day
37
32
mixed
30/365 days
36
31
35
30
34
29
33
28
32
27
MRP
Within-State Urban Inequality
population weighted averages of state level ginis
polynomial fit
kinked trend
The most recent data also shows that wage rates
have stopped growing except for the most
educated.
Real wages of Regular Male Workers
1993-94
1999-00
2004-05
350.0
300.0
250.0
200.0
150.0
100.0
50.0
Urban
Graduates
Rural
Graduates
Urban
Secondary
Rural
Secondary
Urban
Primary
Rural
Primary
Urban
Illiterates
Rural
Illiterate
0.0
And also greater returns to skill within
organised industry.
Incomes of Operatives within Industry
13000
11000
9000
7000
5000
3000
wages per worker
2003-04
2002-03
2001-02
2000-01
1999-00
1998-99
1997-98
1996-97
1995-96
1994-95
1993-94
1992-93
1991-92
1990-91
1989-90
1000
emoluments per non-worker
But as wages have stopped growing, employment growth has
picked up as also GDP growth which has now been over 8% in
the last four years running.
NON-AGRICULTURE EMPLOYMENT
220.0
200.0
160.0
140.0
120.0
100.0
UPS
US (PS+SS)
CWS
Jan-04
Jan-01
Jan-98
Jan-95
Jan-92
Jan-89
Jan-86
80.0
Jan-83
millions
180.0
CDS