The World Bank Research Observer, vol. 16, no. 2
(
Fall 2001
), pp. 199–218
© 2001 The International Bank for Reconstruction and Development /
the world bank
199
Productivity Growth and Sustainability
in Post–Green Revolution Agriculture:
The Case of the Indian and
Pakistan Punjabs
Rinku Murgai, Mubarik Ali, and Derek Byerlee
This article attempts to determine the long-term productivity and sustainability of irrigated
agriculture in the Indian and Pakistan Punjabs by measuring trends in total factor produc-
tivity for production systems in both states since the advent of the Green Revolution. These
measurements over time and across systems have resulted in three major findings. First, there
were wide spatial and temporal variations between the two Punjabs. Although output growth
and crop yields were much higher in the Indian Punjab, productivity growth was higher by
only a small margin. Moreover, the lowest growth in productivity took place during the ini-
tial Green Revolution period (as opposed to the later intensification and post–Green Revolu-
tion periods) and in the wheat-rice system in both states. The time lag between adoption of
Green Revolution technologies and realization of productivity gains is related to learning-
induced efficiency gains, better utilization of capital investments over time, and problems with
the standard methods of productivity measurement that downwardly bias estimates, particu-
larly during the Green Revolution period. Second, input growth accounted for most of the
output growth in both Punjabs during the period under study. Third, intensification, espe-
cially in the wheat-rice system, resulted in resource degradation in both Punjabs. Data from
Pakistan show that resource degradation reduced overall productivity growth from techni-
cal change and from education and infrastructure investment by one-third. These findings
imply the need for policies that promote agricultural productivity and sustainability through
public investments in education, roads, and research and extension; and that reduce resource
degradation by decreasing or eliminating subsidies that encourage intensification of inputs.
The Indo-Gangetic Plain of northern India and Pakistan has one of the largest con-
centrations of poor people in the world. The agricultural sector, which employs more
than half the area’s 500 million inhabitants, has long been considered key to food
The World Bank Research Observer, vol. 16, no. 2
(
Fall 2001
)
200
security and poverty alleviation for this population. Beginning in the mid-1960s,
Green Revolution technologies were introduced in the area, including high-yielding
modern varieties of rice and wheat, the area’s two major crops. This change was
supported by investment in irrigation and market infrastructure. As a result, the area
experienced a dramatic increase in agricultural production, especially in India’s
Punjab State and Pakistan’s Punjab Province.
But despite this promising beginning, the further intensification of input use since
the adoption of Green Revolution technologies has provided lower marginal returns
(Byerlee 1992); and the continued intensification of cropping has sometimes caused
degradation of the resource base in the
form of salinization,
overexploitation of
ground-
water, physical and chemical deterioration of the soil, and pest and disease problems
(Fujisaka and others 1994; Siddiq 1994). Consequently, there is now great concern
about the potential for productivity growth in irrigated Green Revolution systems and
their sustainability over the longer term.
The Debate about Agricultural Performance
Despite the evidence of sharply lower growth rates for food grain yields, there is con-
siderable controversy about aggregate performance of the agricultural sector. In
particular, good performance in nonfood crops—such as cotton in Pakistan; oilseeds,
fruits, and vegetables in northwest India; and livestock in both countries—may have
offset the slowdown in food grains.
Moreover, crop yields are only a measure of partial factor productivity, whereas
overall agriculture sector performance is generally measured by total factor produc-
tivity (
tfp
). The
tfp
approach compares an index of output changes with an index of
input changes, thus making it possible to attribute residual output growth to tech-
nical progress, changes in input quality, and changes in the physical and economic
environment. Experience from industrialized countries suggests that, over the longer
run,
tfp
in the agriculture sector should grow at 1.5 to 2 percent a year, and that
one- to two-thirds of that growth will be due to investment in research and extension.
Recent estimates of
tfp
for agriculture in Pakistan and northwestern India pro-
vide conflicting conclusions. For Pakistan, two studies indicate negative
tfp
growth
in the post–Green Revolution period, especially in the Punjab (Azam and others 1991;
Ali and Velasco 1994). By contrast, another study (Khan 1994) concludes that
tfp
in Pakistan grew sharply in the period 1980–92, at an annual rate of 2.1 percent,
suggesting that the agricultural sector in that country performed well in recent years.
For northwest India, there is little recent evidence on aggregate
tfp
growth, although
the few available studies generally indicate that it was positive (Kaur 1991; Sidhu
and Byerlee 1991; Kumar and Rosegrant 1994; Evenson and others 1999). These
conflicting results are due in part to the studies’ widely varying coverage of inputs
Rinku Murgai, Mubarik Ali, and Derek Byerlee
201
and outputs, methods of valuing inputs, index procedures used to estimate
tfp
, and
levels of disaggregation.
In addition, there is little quantitative evidence of the impact of resource degrada-
tion on productivity growth. Thus, with the body of information that now exists, it
is difficult to accurately assess productivity growth and sustainability under intensi-
fication, and to reach definite policy conclusions about how best to ensure food
security and alleviate poverty in the Indo-Gangetic Plain.
Objectives of the Study
The main objectives of this study are, first, to provide comparable estimates of
tfp
growth in the Indian and Pakistan Punjabs since the advent of the Green Revolu-
tion and, second, to relate productivity trends to changes in resource quality. The
article is organized as follows. The next section outlines the methodology. We then
describe major trends in the agriculture sector, especially those related to production
performance, input use, and resource degradation. We then present estimates of
tfp
at the state level and by cropping system. In each case, we decompose output growth
into the contributions from growth in input use and growth in
tfp
. In the subsequent
section, we use detailed data on resource degradation in Pakistan to further decom-
pose productivity trends into the effects of technology, resource degradation, human
resources, and infrastructure. The final two sections discuss policy implications and
summarize our main findings.
Methodology
Our approach is to estimate growth in
tfp
for three periods corresponding to differ-
ent phases of technical change: (1) The Green Revolution period itself (1966–74),
when input-responsive modern varieties of wheat and rice were widely adopted, lead-
ing to an immediate and dramatic increase in production; (2) the input intensifica-
tion period (1975–85 in India, 1975–84 in Pakistan), when the use of fertilizers and
capital inputs increased rapidly; and (3) the post–Green Revolution period (1986–
94 in India, 1985–94 in Pakistan), when input use leveled off (Byerlee 1992).
We base our calculations (box 1) on district-level data on all inputs, outputs, and
prices, collected from statistical agencies and secondary sources in both states. The
data cover the period 1961–94 in India and 1966–94 in Pakistan. Input categories
include land, labor, water, machinery, draught animals, fertilizer, and pesticide costs.
To minimize aggregation bias in
tfp
, inputs of different qualities are valued by the
price of each quality type. Land is divided into irrigated and unirrigated, labor into
skilled and unskilled (based on the rural literacy rate in each district), water into canal
The World Bank Research Observer, vol. 16, no. 2
(
Fall 2001
)
202
and tubewell, and fertilizer into individual nutrient sources (nitrogen, phosphorous,
and potassium).
Outputs are aggregated into an output index using district-specific farm harvest
prices for crops and market center–specific prices for livestock products.
These data are used to estimate
tfp
separately for different agro-ecological zones,
defined in terms of cropping systems, to avoid the problem of aggregation across
heterogeneous regions. This approach enables direct comparison of productivity
trends and helps determine whether productivity slowdown and environmental deg-
radation are associated with particular cropping systems and ecologies. In India, the
districts are divided into three cropping systems: wheat-rice, wheat-cotton, and
wheat-maize. In Pakistan, they are divided into wheat-rice, wheat-cotton, wheat–
mixed summer crops (often maize or sugarcane), and wheat-mungbean (or wheat-
fallow). The district-level data are then aggregated to quantify
tfp
growth in terms
of the dominant cropping pattern. For Pakistan, where we collected test results from
1971–94 on the quality of groundwater and soil (organic matter, phosphorous con-
tent, and soluble salts), productivity growth is also econometrically decomposed into
the effects of technology, resource degradation, human resources, and infrastructure.
Disaggregated data on resource quality for the Indian Punjab and for the pre-1971
years in Pakistan were not available.
Trends in Production, Input Use, and Resource Degradation
Table 1 describes the production record in the two Punjabs for the three major crops—
wheat, rice, and cotton—during the Green Revolution, intensification, and post–
Green Revolution periods. During the first period, modern wheat and rice varieties
were widely and rapidly adopted in both states. Wheat production increased by more
than 7 percent annually, with yield increases accounting for slightly more than half
Box 1.
Calculating Change in Total Factor Productivity as a Result of the Green Revolution
Of the several ways to measure
tfp
using different rules for aggregating outputs and inputs (Alston and others
1995), we use the chain-linked Tornqvist-Theil index, because it provides an exact measure of technical
change for the linear homogenous translog production function with Hicks-neutral technical change (Diewert
1976).
tfp
is obtained by taking the difference between the growth rates of the aggregate output and input
indices:
TFP
≈
ln(
TFP
t
/
TFP
t
–1
) = [ln(
QI
t
/
QI
t–
1
) – ln(
XI
t
/
XI
t–
1
)] =
Σ
t
1/2 (
R
it
+
R
it
–1
)ln(
Q
it
/
Q
it
–1
)
–
Σ
j
1/2 (
S
jt
+
S
jt
–1
) ln(
X
jt
/
X
jt
–1
)]
where
QI
t
is the aggregate output index,
XI
t
is the aggregate input index, and
R
it
and
S
jt
are the revenue share
of output
i
and cost share of input
j
at time
t
, respectively.
Rinku Murgai, Mubarik Ali, and Derek Byerlee
203
that growth; rice production also grew rapidly, especially in the Indian Punjab. In
the post–Green Revolution period, however, yield growth rates in both states de-
creased to an average of 2 percent a year for wheat and became stagnant or negative
for rice, creating concerns that the Green Revolution may not be sustainable.
Adding to these concerns, the gap between yields in the two states widened over time,
even though the two states had similar cropping patterns and ecologies. Wheat yields
in India during the post–Green Revolution period were nearly double those in Paki-
stan, although they share similar agro-climatic conditions. Rice yields in India were
also much higher during the post–Green Revolution period, although this was known
to be partly due to Pakistan’s specialization in low-yielding, highly valued Basmati rice.
However, cotton yields in Pakistan were higher in both level and growth rate.
Table 1.
Yield Performance of Major Crops, Indian and Pakistan Punjabs
Indian Punjab Pakistan Punjab
Growth rate in yields (%)
Wheat
[3.6] [2.2]
Green Revolution 4.7 5.1
Intensification 2.6 1.1
Post–Green Revolution 2.5 2.1
Rice
[4.1] [-0.3
ns
]
Green Revolution 9.4 4.2
Intensification 2.3 –1.6
Post–Green Revolution 0.7
ns
–1.4
Cotton
[1.6] [3.6]
Green Revolution 0.4
ns
-0.6
ns
Intensification 0.1
ns
2.8
ns
Post–Green Revolution 7.3 8.0
Average yields (kg/ha)
Wheat
Green Revolution 2,004 1,246
Intensification 2,750
1,605
Post–Green Revolution 3,643 1,902
Rice
Green Revolution 1,609 1,320
Intensification 2,777
1,366
Post–Green Revolution 3,246 1,215
Cotton
Green Revolution 347 288
Intensification 316 267
Post–Green Revolution 504 601
Note:
The figures in brackets [ ] indicate growth rate in the parameter value during
overall study period.
ns = not significantly different from zero at 10 percent.
Sources:
Ali and Byerlee (forthcoming) and Murgai (forthcoming).
The World Bank Research Observer, vol. 16, no. 2
(
Fall 2001
)
204
The data indicate that the differences between the states in production performance
are associated with differences in input use and cropping intensity (table 2). In the
first period, the adoption of modern varieties stimulated rapid input intensification
in both Punjabs. In India, fertilizer use jumped from 33 to 156 kg of nutrients per
hectare of cropped area between the first and third periods; labor use gradually de-
clined; and the use of mechanical power (tractors, harvesters, and threshers) in-
creased from 4.3 to 41 hours per hectare. Pakistan followed the same patterns, al-
though cropping intensities and the use of fertilizer and machinery were considerably
lower in all periods.
The data also show considerable degradation of the water and soil resource base
in both states. There are indications that the wheat-rice cropping system in the Indian
Table 2.
Changes in Cropping Intensity, Irrigated Area, and Input
Use, Indian and Pakistan Punjabs
Indian Punjab Pakistan Punjab
Cropping intensity (%) [1.0] [0.8]
Green Revolution 138.0 117.0
Intensification 158.0
126.0
Post–Green Revolution 174.0 136.0
Irrigated area (%) [1.7] [0.3]
Green Revolution 68.6 81.9
Intensification 81.9 85.1
Post–Green Revolution 90.8 86.3
Fertilizer (kg/ha) [13.1] [12.6]
Green Revolution 33.0 14.1
Intensification 99.2 48.3
Post–Green Revolution 155.9 86.1
Machines (hrs/ha) [12.8] [12.0]
Green Revolution 4.3 1.5
Intensification 15.3 5.7
Post–Green Revolution 41.0 14.8
Tubewells (#/1,000ha)
a
[11.5] [3.9]
Green Revolution 27.3 8.2
Intensification 80.6 16.2
Post–Green Revolution 104.4 26.0
Labor (days/ha) [–1.4] [–0.9]
Green Revolution 84.2 85.0
Intensification 75.9 98.7
Post–Green Revolution 64.6 71.1
Note:
The figures in brackets [ ] indicate growth rate in the parameter value during
overall study period.
ns = not significantly different from zero at 10 percent.
a. Tubewell numbers are not directly comparable. Tubewells in Pakistan are much
larger.
Sources:
Ali and Byerlee (forthcoming) and Murgai (forthcoming).
Rinku Murgai, Mubarik Ali, and Derek Byerlee
205
Punjab was hurt by a steep decline in the water table, while rising water levels in the
wheat-cotton zone led to severe waterlogging in the wheat-cotton zone. Data from
the Pakistan Punjab also confirm a serious problem of waterlogging and salinity, due
in part to deterioration in the quality of tubewell water (reflected in a significant in-
crease in residual carbonate and electroconductivity of groundwater). Soil quality
in Pakistan (in terms of available soil organic matter and phosphorus) also deterio-
rated, particularly in the wheat-rice zone (figures 1a and 1b).
Source:
Ali and Byerlee (forthcoming).
a.
0
50
100
150
200
250
300
Year
Available phosphorous
Organic matter
Soluble salt
b.
0
50
100
150
200
250
Year
Residual carbonate
Electroconductivity
Figure 1. a.
Indices of Trends in Soil Quality;
b.
Indices of Trends in Tubewell Water Quality