The impact of climate change on major agricultural crops: evidence from Punjab, Pakistan.
Siddiqui, Rehana ; Samad, Ghulam ; Nasir, Muhammad 等
1. INTRODUCTION
It is necessary for a country to make its agriculture sector
efficient to enhance food security, quality of life and to promote rapid
economic growth. The evidence from least developed countries (LDCs)
indicates that agriculture sector accounts for a large share in their
gross domestic product (GDP). Thus the development of the economy cannot
be achieved without improving the agriculture sector. According to the
Economic Survey of Pakistan (2011-12) its main natural resource is
arable land and agriculture sector's contribution to the GDP is 21
percent. The agricultural sector absorbs 45 percent of labour force and
its share in exports is 18 percent. Given the role of agricultural
sector in economic growth and its sensitivity to change in temperature
and precipitation it is important to study the impact of climate change
on major crops in Pakistan.
There are two crops seasons in Pakistan namely, Rabi and Kharif.
Rabi crops are grown normally in the months of November to April and
Kharif crops are grown from May to October. These two seasons make
Pakistan an agricultural economy and its performance depends on the
climate during the whole year. Climate change generally affects
agriculture through changes in temperature, precipitation.
Schlenker (2006) estimated the impact of climate change on crop
yield for the agriculture sector of United States. This study found
threshold levels of temperatures to be 29[degrees]C for corn and
soybeans and 33[degrees]C for cotton. It concluded that the temperature
above threshold would harm the crops. The hypothesis was tested by
incorporating 3000 counties of US in the analysis. Though temperatures
in all seasons, except in autumn, reduced the farm value but high
precipitation increased the agriculture production of the US [Mendelsohn
(1994)]. Therefore, for the United States global warming has very little
impact on the agriculture sector. However, at the beginning climate
change may have small effects for developed countries but in future
negative effects will be very large and stronger. Countries with longer
latitude, climate change may lead to net benefits but countries with low
latitude are more vulnerable [Stern (2006)].
In recent decades, high temperatures have been observed in Asia and
the Pacific regions. In these regions agriculture sector is more
vulnerable as 37 percent of the total world emissions from agriculture
production are accumulating from Asia and the Pacific. Countries most
vulnerable to climate change include Bhutan, Indonesia, Pakistan, Papua
New Guinea, PRC, Sri Lanka, Thailand, Timor-Leste, Uzbekistan, and
Vietnam [Asian Development Bank (2009)]. On the other hand, there is
also a possibility that agriculture sector may harm the climate. This
problem is identified by Paul, et al. (2009). It is observed that 14
percent of nitric oxide and methane is coming from the agriculture
sector and 18 percent is due to deforestation for agriculture use.
Season and location really matters for the production in
agriculture sector. African crops are more sensitive to marginal change
in temperature as compare to change in precipitation. For African crops
temperature rise has positive effects, while reduction in rainfall
negatively affects net revenues. These observations were based on seven
African field crops (maize, wheat,' sorghum, sugarcane, groundnut,
sunflower and soybean) of 300 districts in South Africa [Gbetibouo
(2005)]. Study also suggested that one can shift the growing season of a
crop according to temperature but there is a possibility that, this type
of action may lead to complete elimination of some crops of some
regions.
The agriculture sector in Pakistan plays a pivotal role as the
income of more than 47 percent of the population is dependent on this
sector. This sector is under threat from climate change. It is projected
that temperatures will increase by 3[degrees]C by 2040 and 5[degrees]C
to 6[degrees]C by the end of this century. Due to this scenario, Asia
can lose 50 percent of its wheat production [MOE (2009)]. Moreover,
agriculture sector of Pakistan is more vulnerable to climate change
because of its geographical location [Janjua, et al. (2011)]. This study
explains that due to anthropogenic activities, temperature of earth is
rising and it may have negative effect on the production of wheat. Using
Vector Auto Regressive (VAR) model on the annual data from 1960 to 2009,
the study did not find significant negative impact of climate change on
wheat production in Pakistan. However, on the other hand, Shakoor (2011)
found significant negative impact of temperature-rise on agriculture
production and also found the positive impact of rain fall on
agriculture production. Analyses were based on the wheat crop and study
concluded that the negative impact of temperature is greater than the
positive impact of rainfall for Pakistan. The authors also estimated
cost of arid regions due to 1 percent increase in temperature, which
came to Rs 4180/- to the net revenue per annum.
1.1. Objectives of the Study
The objective of present study is to investigate the impact of
climate (through changes in temperature and precipitation) on four major
crops namely; Wheat, Rice, Cotton and Sugarcane in the Punjab Province
of Pakistan. Estimations based on the time series data from 1980-2008.
The study also makes projections regarding the effects of changes in
temperature and precipitation on the crops production. This is the first
study incorporating scientific information on the stages of development
of each crop in order to assess the impact of climate change on each
stage of the crops.
1.2. Organisation of the Study
Section 1 of this study includes definition of key terms, problem
and objectives. Section 2 describes data description and methodology.
Section 3 covers empirical estimations and results. Section 4 concludes
the study with recommendations and finally Section 5 describes the
limitation of the study.
2. DATA AND METHODOLOGY
2.1. Data Description
The analysis is carried out using the data of four major crops
namely Wheat, Rice, Cotton and Sugarcane form the province of Punjab.
The scientific information of production stages of these crops and its
optimal temperature and precipitations were taken from the Pakistan
Agricultural Research Council (PARC), Rice Research Institute, Kala Shah
Kaku, Cotton Research Institute, Faisalabad, and Sugarcane Research
Institute, Faisalabad respectively. For each of the crops analysis the
station wise selection of the districts were made according to their
productivity e.g. the districts were varied from crops to crops
depending on their productivity size.
The wheat and rice production has been consists of three different
stages of production and of three different optimal temperature and
precipitations. The optimal temperature of the cotton production remain
the same therefore, scientifically it has not been divided into
different production stages. Similarly, the sugarcane production has
been divided into four different production stages that of their optimal
temperature and precipitations. The data on districts wise productivity
of each crop were taken from statistical year book of Ministry of
Agriculture, the data on temperature and precipitation were taken from
the department of Metrology. We faced many problems in unbalance panel;
therefore we use the balance panel design for the year 1980-2009.
2.2. Specification of the Model
Fixed Effect Model (FEM) is used on the base of the balanced data
design, the dependent variable is Crops (Wheat, Rice, Cotton, Sugarcane)
productivity and explanatory variables are first stage temperature (FT),
second stage temperature (SST), third stage temperature (TST), fourth
stage temperature (FST), first stage precipitations (FP), second stage
precipitation (SSP), third stage precipitation (TSP), fourth stage
precipitation (FSP). In order to capture the nonlinearity impact, we
have included squared term for these variables
The general equation of this study is
[Crops.sub.w,r,c,s,] = f (FT, [FT.sup.2], SST, [SST.sup.2], TST,
[TS.sup.2], FST, [FS.sup.2], FP, [FP.sup.2], SSP, [SSP.sup.2], TSP,
[TSP.sup.2], FSP, [FSP.sup.2])
[(Crops).sub.it], = [[alpha].sub.i], + [[beta].sub.1],
[(FT).sub.it] + [[beta].sub.2] [([FT.sup.2]).sub.it] -- + [[beta].sub.n]
[(Tem, Pre).sub.it] + [V.sub.it]
(i = 1,2 ... N; t = 1,2 ... T)
[V.sub.it] = [[mu].sub.i] + [summation][W.sub.it]
[V.sub.it], is composite error term, and [[mu].sub.i] is
unobservable individual country specific effects and
[summation][w.sub.it] is other disturbances.
3. ESTIMATION RESULTS
In this section, we put forward the estimation results of the four
crops and discuss the results in detail. Section 3.1 discusses the
results of wheat crop in the Punjab province. The results for rice crop
are presented in Section 3.2. The impact of climate change on cotton
crop is inspected in Section 3.3. Section 3.4 discusses the impact of
climate change on sugarcane. The last section discusses the simulation
results for various scenarios changes in temperature.
3.1. Wheat Production
This section discusses the estimation results of wheat crop in
Punjab province. The cropping period for wheat is from December to
April. Consequently, we have divided the cropping period in three stages
due to different requirement of temperature and precipitation for each
stage. The first stage covers the month of December whereas the second
stage consists of the period from January to March. The third stage
again consists of only one month, namely, April. The estimation results
are presented in Table 1.
Table 1 shows the results of two models estimated for identifying
the impact of temperature and precipitation on wheat crop. In the first
model, both temperature and precipitation have been used along with
their square terms, assuming a non-linear relationship between the
variables. The results of this model show that temperature affect wheat
crop non-linearly only in first stage of production. Surprisingly, this
non-linear relationship is of U-shaped type. This means that after the
temperature of 14.76[degrees]C, further increase in temperature will
positively affect wheat crop. In the second and third stages of
production, however, variations in temperature have insignificant effect
on wheat production. On the other hand, the precipitation has
significant non-linear relationship with wheat crop in the first two
stages of production. The optimal precipitations for the first two
stages are 111 mm and 84.50 mm respectively. That is, beyond these
optimal limits, further precipitation will adversely affect growth of
plant and it's fruiting. As was the case with temperature, in the
third stage precipitation does not affect wheat crop.
The constant term (intercept) shows the average production of the
seven districts included in the model due to district specific
characteristics whereas the coefficients of district dummies show
deviations from this mean production. The significance of coefficients
of these dummies variables indicates that district specific
characteristics do have significance in the production of wheat crop.
These results shows that, Jhelum, Lahore and Mianwali respectively
produce 325.69, 324.13 and 108.92 thousand tonnes less, whereas,
Bahawalpur, Faisalabad, Multan and Sialkot respectively produce 306.21,
338.69, 41.65 and 72.18 thousand tonnes more than the average production
(which is 749.56). The model performed well on represented by F-Stats,
significance of the model.
In the second model, the insignificant terms of temperature for the
second and third stage were dropped from estimation. The results confirm
the robustness of coefficients in terms of both sign and significance.
It is also evident from the table that values of coefficients are not
volatile either. This model also confirms that the positive effect of
temperature in the first stage starts from 14.14[degrees]C. Likewise,
the optimal precipitations for the first two stages are 112 mm and 97 mm
respectively. Similarly, the deviation of district dummies variables
from the mean is not significant and the sign and significance of the
coefficient of these dummies have not changed. The DW statistics
confirms the absence of serial correlation problem and F-stats shows the
overall significance of the model.
3.2. Rice Production
This section explores the impact of climate change on rice
production in the seven districts of Punjab province. The crop period
for rice in Punjab consists of four months, from August to November.
There are three main stages of production for rice crop, namely,
Germination, Flowering and Ripening. Accordingly, we have classified
time period of rice crop production in three stages. The first stage
consists of the month of August, while the September and October jointly
constitute the second stage. Third stage reaches in the month of
November. The estimation results for rice crop are presented in Table 2.
Two models have been estimated to investigate the impact of climate
change on rice production as shown in Table 2. In the first model, both
temperature and precipitation have been used with their square terms to
inspect the non-linear impact of these variables. The results of this
model confirm the notion that temperature affect rice crop non-linearly
in first two stages of production. Accordingly, a rise in temperature is
beneficial for rice production initially, in the first stage. However,
beyond a certain optimal temperature 27[degrees]C for the first stage,
further increase in temperature becomes harmful for production. In the
second stage, however, the non-linear relationship is of U-shaped.
Initially, a rise in temperature is harmful for production, but beyond a
certain temperature limit (which is 26.75[degrees]C) the effect becomes
positive. This outcome may be a result of overlapping of different
stages of growth of the plant due to our classification of these stages
using monthly data, as both low and high temperatures are harmful for
production [Chaudhary, et al. (2002)]. The third stage of production is
not affect by increase in temperature. It means that, for Punjab, the
temperature for the third stage remains in the optimal limits for the
entire period of this stage. The average temperature for included
districts of Punjab is 22 degree centigrade, whereas the optimal
required temperature for this stage ranges from
20[degrees]C-25[degrees]C [Chaudhary, et al. (2002)].
An interesting result is the insignificance of precipitation for
rice production in all the stages. This result is, however, justifiable
on the grounds that the annual precipitation in Pakistan is less [only
20 mm] than the optimal required precipitation [which is 40mm on the
lower bound] for rice production. This deficiency has been met by the
artificial arrangements of irrigation water through canals and tube
wells, thereby reducing the dependency on rainfall. For 75 days [which
is almost the first two stages], the rice fields should have 6 mm of
slow moving water. However, the water requirement gradually decreases
during the maturity period of crop. This maturity period is the third
stage of production, which is in the month of November in our case. The
data shows that the average rainfall during this month is only 5 mm and,
hence, may not be harmful for the crop. In a nutshell, we may say this
climate variable is irrelevant for rice crop in the sense that both
neither the lower nor the upper levels of precipitation are harmful. The
lower precipitation is covered by irrigation methods and the upper level
does not reach at all.
Lastly, the significance of district dummies confirms the fact the
production of rice crop does respond to district specific
characteristics. The intercept term in the model represents the mean
rice production of these seven districts, whereas coefficients of
district dummies show the deviation from this average production. It is
evident from the results that, except for Sialkot, all other districts
produce less rice than the average production. The [R.sup.2] and F-Stats
validate the significance of the overall model.
In the second estimation, the insignificant variable precipitation
has been dropped from the model from all stages of production. The
results are robust as only first two stages of production are affected
by change in temperature. In addition, all the district dummies are also
significant. Hence, one may easily conclude that these results are
robust in terms of values, signs and significance for all the
parameters. The optimal temperature for the first stage is
28.33[degrees]C in the respective case. Whereas, the positive effect of
temperature in the second stage starts beyond 28.11[degrees]C. The
differences between these temperatures between the two models are
1.33[degrees]C and 1.36[degrees]C respectively for the two stages.
However, these optimal temperatures in both the models for both stages
are consistent with optimal required temperature determined
scientifically in literature [see for example, Chaudhary, et al. (2002)
for details] (1). Again, the [R.sup.2] and F-stats confirm the
significance of the overall model.
3.3. Cotton Production
The underlying section deals empirically with the impact of climate
change on cotton production. The period for cotton crop in Punjab is
from May to September. Since the optimal temperature and precipitation
requirement is same for the whole period of crop production. We have not
made different stages of production for cotton. The maximum temperature
and precipitation required for cotton crop during the production period
is 32[degrees]C and 40mm respectively. (2) Since the data shows obvious
deviation from the maximum limits for both variables, we take the
deviation from maximum limits for purpose of estimation. This is in
contrast to what we have done for wheat and rice crops where the
historical data appeared to lie in the optimal limits and no clear
deviations from maximum limits of either variable were observable. In
the following lines we discuss the estimation results for cotton
production.
Table 3 represents the results of impact of climate change on
cotton production in five districts of Punjab province. Two models have
been estimated for this purpose. Model 1 is estimated for investigating
the non-linear relationships between the cotton production and climate
variables namely changes in temperature and precipitation. The results
of model 1 show that square terms of both the variables are
statistically insignificant, suggesting that the relationship is linear.
For this purpose, the square terms of these variables are dropped in the
second model and a linear relationship is estimated. It is evident from
the table that the coefficients all the variables (including districts
dummies) are robust both in terms of sign and significance. Moreover,
the values of the coefficients are not volatile either. It is important
to mention that these results are presented after correcting for the
problems of autocorrelation and heteroscedasticity. The overall models,
represented by F-tests, are statistically significant at the
conventional level of significance.
As is mentioned in the above lines, the climate variables are taken
in the form of deviation from standard maximum required levels.
Therefore, one should be careful in interpreting these results. Since
the second model is the best one in terms of explaining the true
relationship, we interpret the results of this model. The results
indicate that a one degree centigrade deviation of temperature from the
maximum required level (which is 32C) during the whole period reduces
the production of cotton by 42.33 thousands bales. Similarly, a one
millimetre deviation of precipitation from the maximum required level
(which 40 mm) reduces the production of cotton by 0.50 thousands bales.
This is a significant loss in the production of cotton due to change in
the climate variables. The reduction in production due to both the
variables indicates the climate change has been harmful for cotton
production in this region.
Before explain the district dummies, it is worthwhile to recall
that constant term in the model shows the mean production of the five
districts. Consequently, the coefficients of the district dummies should
be interpreted as deviation from this mean. The results show the mean
production of cotton (after controlling for districts specific
characteristics) is 403.52. Thus, the Bahawalpur and Multan districts
produce more cotton (735.10 and 316.60 thousands bales respectively)
than the mean production. On the other hand, in Faisalabad, Jhelum and
Mianwali districts cotton production is lower than the average
production. These results should not be surprising as cotton production
in these three districts is significantly lower than production in
Bahawalpur and Multan districts. For example, the average production of
cotton during period 1987-2008 in Bahawalpur and Multan was 992 and 800
thousand bales respectively. Whereas, for the same period, the average
production for Faisalabad, Jhelum and Mianwali was 105.5, 0.35, and
13.76 thousand bales only. The significance of district dummies,
however, indicates that the district specific characteristics do have
important impact on cotton production.
3.4. Sugarcane Production
Finally, in this section we are computing the impact of climate and
precipitation change on the sugarcane production in seven districts
namely Bahawalpur, Faisalabad, Jhelum, Mianwali, Sialkot, Lahore and
Multan which are the prone cultivated areas of sugarcane in Pakistan. In
Pakistan the sugarcane harvesting consists of two seasons. The
cultivation of sugarcane crop starts in Feb-December. The production
time is about nine month. However, 30 percent harvesting of crop is in
Sept-December with its total duration of 14 months. The mill owners
prefer this crop due to the high quality of sugarcane production as
compare to the 9 months crop but the farmers enduring 9 month crop so
that the land can be ready for wheat crops otherwise they have to forgo
the wheat production. Similarly, globally two methods are pertinent for
its harvesting e.g. firstly, by germination and secondly, by sowing
seeds. Our farmers are using the first method as the second method
normally takes two years to germinate.
With the consultation of the Sugarcane Research Institute,
Faisalabad we divided the sugarcane production into four stages of
production. These are: Germination of duration 45 days, tillering of
duration of 90 days, vegetative of duration 90 days and maturing
normally 60-75 days.
First stage: Optimum temperature for sowing: 20-32[degrees]C
Optimum temperature for germination: 32-28[degrees]C
Second stage: Maximum temperature decreasing tillering:
30[degrees]C
Third stage: Optimum temperature for sugarcane: 28-38[degrees]C
Fourth stage: Temperature for good sugar production: 10[degrees]C
For the 9 months duration 22 times irrigation are required for good
sugarcane production. The optimum rainfall for sugarcane is: 1250-2500
mm.
The results of Table 4 show that the increase in temperature in the
first three stages of production are highly insignificant. If
temperature rises in the first stage up to 28[degrees]C the temperature
has positive impact on sugarcane production but beyond 28[degrees]C up
to 32[degrees]C it becomes negative. In the second stage the temperature
beyond 30[degrees]C would cause decreasing the telliring the square of
the temperature becomes positive but its magnitude is minimal. The most
important and vulnerable stage is third or vegetative stage of sugarcane
production, the coefficients of the estimation shows that initially the
increase in temperature causes increase in productivity which may be
possibly the optimal temperature ranged from 28-38[degrees]C in this
stage but the square of temperature results in negative productivity.
Finally, the maturity is the fourth and last productivity stage of
production. The sweetness starts in this stage of production, which
requires minimum temperature.
The increase in temperature in these months would reduce the
sweetness and ultimately the yields. The optimal temperature required in
this stage is 10[degrees]C, in the first stage the increase in
temperature has negative impact on sugarcane productivity/yield. The
further increase e.g. the square of the temperature again has positive
but minimal effect on productivity/yields. It is important to mention
that these results are presented after correcting for the problems of
autocorrelation and heteroscedasticity. The overall models, represented
by F-tests, are statistically significant at the conventional level of
significance.
3.5. Simulation Analysis
The results of the simulations analysis for these four major crops
are annexed. The simulations analysis carried out from 2008 to 2030. It
covers almost one-generation period. The simulations results for wheat
production in (000) tonnes shows that the when the temperature increases
by 1C the cumulative loss up to 2030 would be 0.02 percent and if the
temperature increases by 2C the cumulative loss up to 2030 would be 0.75
percent that of 2008. Moreover, the results for simulation analysis of
rice production in (000) tonnes shows that when temperature increases by
1C the respective gain to rice productivity up to 2030 would be 1.85
percent and if the temperature increases by 2C the rice productivity
gain would by 3.95 percent.
The simulation results for cotton production (000) bales with
increase of 1C and 2C shows that the loss to cumulative cotton
production up to 2030 is 13.29 percent and 27.98 percent respectively.
Finally, for the same increase of 1C and 2C the sugarcane (000) bales,
cumulative loss up to 2030 are 13.56 percent and 40.09 percent
respectively.
4. CONCLUSION
The study focuses on the impact of on changes in climate change
indicators on production of four major crops in Punjab, Pakistan. The
results show that in the short run the increase in temperature is
expected to affect the wheat productivity but in long term the increase
in temperature has positive affect on wheat productivity. Similarly, the
increase in precipitation has negative impact in both short and long
term. A rise in temperature is beneficial for rice production initially.
However, beyond a certain optimal temperature, further increase in
temperature becomes harmful for rice production. Interestingly, the
increase in precipitation does not harm the rice productivity. It has
been evident that the change in climate variables (temperature,
precipitation) has a significant negative impact on production of
cotton. Finally, the increase in temperature also harms the sugarcane
productivity in long term.
The major conclusions of the study are:
First: The impact of changes in temperature and precipitation
varies significantly with the timing and production stages of the crops.
Second: The impact varies from crop to crop.
Finally: The districts variations in crop productivity are
significant.
5. LIMITATION OF THE STUDY
The limitations are:
(1) The analysis is limited to the province of Punjab; we are in
the process of finalising the results for other provinces of Pakistan.
(2) The study considers two important climate change variables
namely temperature and precipitation but other explanatory variables
like humidity, soil fertility, and other inputs variables are not
consider due to nonavailability of districts wise data. A district level
survey is required to include these variables in the analysis.
(3) The simulation analyses consider temperature increases by 1C
and 2 C respectively, and the precipitations scenarios are kept
constant. The simulation results for precipitation are in the process.
ANNEX
Simulation Results for Wheat Production (000 tonnes)
Year
Temperature Wheat Wise Cumulative
Years 1C Production Gain Gain
2008 63.24209
2009 63.29225 0.050168 0.050168
2010 63.34273 0.050475 0.100643
2011 63.39351 0.050783 0.151426
2012 63.4446 0.05109 0.202515
2013 63.496 0.051397 0.253913
2014 63.5477 0.051704 0.305617
2015 63.59971 0.052012 0.357629
2016 63.65203 0.052319 0.409948
2017 63.70466 0.052626 0.462574
2018 63.75759 0.052934 0.515507
2019 63.81083 0.053241 0.568748
2020 63.86438 0.053548 0.622296
2021 63.91824 0.053855 0.676152
2022 63.9724 0.054163 0.730315
2023 64.02687 0.05447 0.784785
2024 64.08165 0.054777 0.839562
2025 64.13673 0.055085 0.894647
2026 64.19212 0.055392 0.950039
2027 64.24782 0.055699 1.005738
2028 64.30383 0.056007 1.061745
2029 64.36014 0.056314 1.1 18058
2030 64.41677 0.056621 1.17468
% Gain
1.857433
Year
Temperature Wheat Wise Cumulative
Years 2C Production Gain Gain
2008 63.24209
2009 63.34273 0.100643 0.100643
2010 63.4446 0.101872 0.202515
2011 63.5477 0.103101 0.305617
2012 63.65203 0.104331 0.409948
2013 63.75759 0.10556 0.515507
2014 63.86438 0.106789 0.622296
2015 63.9724 0.108018 0.730315
2016 64.08165 0.109247 0.839562
2017 64.19212 0.110477 0.950039
2018 64.30383 0.111706 1.061745
2019 64.41677 0.112935 1.17468
2020 64.53093 0.114164 1.288844
2021 64.64632 0.115393 1.404237
2022 64.76295 0.116623 1.52086
2023 64.8808 0.1 17852 1.638712
2024 64.99988 0.119081 1.757793
2025 65.12019 0.12031 1.878103
2026 65 24173 0.121539 1.999642
2027 65.3645 0.122769 2.122411
2028 65.48849 0.123998 2.246408
2029 65.61372 0.125227 2.371635
2030 65.74018 0.126456 2.498091
% Gain
3.950046
Simulation Results for Cotton Production (000 Bales)
Year
Temperature Cotton Wise Cumulative
Years 1C Production Loss Loss
2008 371.9732
2009 369.8384 2.134754 2.134754
2010 367.6929 2.145498 4.280251
2011 365.5367 2.156241 6.436493
2012 363.3697 2.166985 8.603478
2013 361.192 2.177729 10.78121
2014 359.0035 2.188473 12.96968
2015 356.8043 2.199217 15.1689
2016 354.5943 2.20996 17.37886
2017 352.3736 2.220704 19.59956
2018 350.1422 2.231448 21.83101
2019 347.9 2.242192 24.0732
2020 345.6471 2.252936 26.32614
2021 343.3834 2.263679 28.58981
2022 341.109 2.274423 30.86424
2023 338.8238 2.285167 33.1-494
2024 336.5279 2.295911 35.44532
2025 334.2212 2.306655 37.75197
2026 331.9038 2.317398 40.06937
2027 329.5757 2.328142 42.39751
2028 327.2368 2.338886 44.7364
2029 324.8872 2.34963 47.08603
2030 322.5268 2.360374 49.4464
%Loss
13.293
Year
Temperature Cotton Wise Cumulative
Years 2C Production Loss Loss
2008 371.9732
2009 367.6929 4.280251 4.280251
2010 363.3697 4.323226 8.603478
2011 359.0035 4.366202 12.96968
2012 354.5943 4.409177 17.37886
2013 350.1422 4.452152 21.83101
2014 345.6471 4.495127 26.32614
2015 341.109 4.538102 30.86424
2016 336.5279 4.581078 35.44532
2017 331.9038 4.624053 40.06937
2018 327.2368 4.667028 44.7364
2019 322.5268 4.710003 49.4464
2020 317.7738 4.752979 54.19938
2021 312.9779 4.795954 58.99533
2022 308.1389 4.838929 63.83426
2023 303.257 4.881904 68.71617
2024 298.3322 4.924879 73.64104
2025 293.3643 4.967855 78.6089
2026 288.3535 5.01083 83.61973
2027 283.2997 5.053805 88.67353
2028 278.2029 5.09678 93.77031
2029 273.0631 5.139755 98.91007
2030 267.8804 5.182731 104.0928
%Loss
27.98385
Simulation Results for Sugarcane Production (000 Tonnes)
Year
Temperature Sugarcane Wise Cumulative
Years 1C Production Loss Loss
2008 936.464
2009 933.3288 3.135187 3.135187
2010 929.9425 3.3863 6.521487
2011 926.3051 3.637413 10.1589
2012 922.4166 3.888526 14.04743
2013 918.277 4.139639 18.18707
2014 913.8862 4.390752 22.57782
2015 909.2443 4.641865 27.21968
2016 904.3514 4.892978 32.11266
2017 899.2073 5.144091 37.25675
2018 893.8121 5.395204 42.65196
2019 888.1658 5.646317 48.29827
2020 882.2683 5.89743 54.1957
2021 876.1198 6.148543 60.34425
2022 869.7201 6.399656 66.7439
2023 863.0694 6.65077 73.39467
2024 856.1675 6.901883 80.29656
2025 849.0145 7.152996 87.44955
2026 841.6104 7.404109 94.85366
2027 833.9551 7.655222 102.5089
2028 826.0488 7.906335 110.4152
2029 817.8914 8.157448 118.5727
2030 809.4828 8.408561 126.9812
% Loss
13.56
Temperature Sugarcane Wise Cumulative
Years 2C Production Loss Loss
2008 936.464
2009 929.9425 6.521487 6.521487
2010 922.4166 7.525939 14.04743
2011 913.8862 8.530391 22.57782
2012 904.3514 9.534843 32.11266
2013 893.8121 10.5393 42.65196
2014 882.2683 11.54375 54.1957
2015 869.7201 12.5482 66.7439
2016 856.1675 13.55265 80.29656
2017 841.6104 14.5571 94.85366
2018 826.0488 15.56156 110.4152
2019 809.4828 16.56601 126.9812
2020 791.9123 17.57046 144.5517
2021 773.3374 18.57491 163.1266
2022 753.7581 19.57936 182.706
2023 733.1742 20.58382 203.2898
2024 711.586 21.58827 224.8781
2025 688.9933 22.59272 247.4708
2026 665.3961 23.59717 271.0679
2027 640.7945 24.60163 295.6696
2028 615.1884 25.60608 321.2756
2029 588.5779 26.61053 347.8862
2030 560.9629 27.61498 375.5012
% Loss
40.098
Simulation Results for Rice Production (000 Tonnes)
Year
Tempera- Rice Wise Cumulative
Years ture 1C Production Loss Loss
2008 407.1121
2009 407.0383 0.073766 0.073766
2010 406.9713 0.067084 0.14085
2011 406.9109 0.060401 0.201251
2012 406.8571 0.053718 0.254969
2013 406.8101 0.047036 0.302005
2014 406.7697 0.040353 0.342358
2015 406.7361 0.033671 0.376029
2016 406.7091 0.026988 0.403018
2017 406.6888 0.020306 0.423323
2018 406.6752 0.013623 0.436947
2019 406.6682 0.006941 0.443888
2020 406.668 0.000258 0.444146
2021 406.6744 -0.00642 0.437722
2022 406.6875 -0.01311 0.424615
2023 406.7073 -0.01979 0.404826
2024 406.7338 -0.02647 0.378354
2025 406.7669 -0.03315 0.345199
2026 406.8067 -0.03984 0.305362
2027 406.8533 -0.04652 0.258843
2028 406.9065 -0.0532 0.205641
2029 406.9663 -0.05988 0.145757
2030 407.0329 -0.06657 0.07919
% Loss
0.01945
Tempera- Rice Year Wise Cumulative
Years ture 2C Production Loss/ Gain Loss/ Gain
2008 407.1121
2009 406.9713 -0.14085 -0.14085
2010 406.8571 -0.11412 -0.25497
2011 406.7697 -0.08739 -0.34236
2012 406.7091 -0.06066 -0.40302
2013 406.6752 -0.03393 -0.43695
2014 406.668 -0.0072 -0.44415
2015 406.6875 0.019531 -0.42461
2016 406.7338 0.046261 -0.37835
2017 406.8067 0.072991 -0.30536
2018 406.9065 0.099721 -0.20564
2019 407.0329 0.126451 -0.07919
2020 407.1861 0.153182 0.073992
2021 407.366 0.179912 0.253904
2022 407.5727 0.206642 0.460545
2023 407.806 0.233372 0.693917
2024 408.0661 0.260102 0.954019
2025 408.353 0.286832 1.240851
2026 408.6665 0.313562 1.554413
2027 409.0068 0.340292 1.894705
2028 409.3738 0367022 2.261728
2029 409.7676 0.393752 2.65548
2030 410.1881 0.420483 3.075963
% Gain
0.755557
REFERENCES
Asian Development Bank (2009) Building Climate Resilience in the
Agriculture Sector in Asia and in the Pacific. Asian Development Bank.
Annual Development Report, p. 9.
Chaudhary, R. C., J. S. Nanda, and D. V. Tran (2002) Guidelines for
Identification of Field Constraints to Rice Production. International
Rice Commission, Food and Agriculture Organisation of the United
Nations, Room.
Mendelsohn, R., W. Nordhaus and D. Shaw (1994) The Impact of Global
Warming on Agriculture: A Ricardian Analysis. The American Economic
Review 84, 753-771.
MoE (2009) Climate Change Vulnerabilities in Agriculture in
Pakistan. Ministry of Environment, Government of Pakistan. Annual
Report, pp. 1-6.
Pakistan, Government of (2011) Pakistan Economic Survey 2011-12,
Chapter No. 2.
Schlenker, W. and M. J. Roberts (2006) Nonlinear Effects of Weather
on Com Yields. Review of Agricultural Economics 28:3, 391-398.
Shakoor, Usman, Abdul Saboor, Ikram Ali, and A. Q. Mohsin (2011)
Impact of Climate Change on Agriculture: Empirical Evidence from Arid
Region. Pak. J. Agri. Sci. 48:4, 327-333.
Stern (2006) Stern Review on the Economics of Climate Change. H. M.
Treasury.
Comments
This is a research area of vital importance which has been explored
very little in Pakistan. Dr Rehana and her research team's
deliberations and empirical exploration regarding the impact of climate
change on major agricultural crops is highly commendable. It is a well
thought paper written with deep understanding of the issue. Mostly
people around the world are tracing the impact of climate change on
agriculture either through production function approach or through
Ricardian Regression. This study has employed Fixed Effect Model for the
first time in Pakistan's perspective. The work which is still in
progress is commendable. Some of the following points must be considered
before finalizing the paper.
(1) It is always some standard economic theory that should be
operated in every climate impact studies. In this paper, methodological
considerations must be accompanied by reasonable theoretical back
ground.
(2) The reader is not comfortable to understand the need of the
study specifically as we see weak linkages of what has been done up till
now and what further research is required. The review of literature is
not as appropriate as it should be for a standard research paper.
Further literature should be explored particularly keeping production
function approach and Ricardian approach in view. Some impact studies
have been made in India, Bangladesh and Sri Lanka. A thorough scanning
of such studies is required to arrive at the justification this study is
being conducted like this way.
(3) A sound justification must be given regarding the use of FEM as
against other standard approaches. The salient advantages of this model
must be the part of this paper so that the reader could know how far the
FEM is better than the traditional approaches.
(4) It should be clearly mentioned how many districts of Punjab
have been taken for the analysis of this issue. Similarly, the reader
would be interested in knowing why of the districts have been added in
the research for each crop. Apparently the picture is vague.
(5) Abstract reflects that the study is from 1980 to 2008 while the
main text says that it is from 1980 to 2009. Also give appropriate
reason of selecting this particular time period. There might be an odd
or extreme event in this time series. How the model is adjusting these
extreme events. It should have been mentioned in the explanation of
data.
(6) It must be addressed in the methodology why non-linear impact
is being explored and some due references must be given. Why other
functional forms are not testable.
(7) It should be clearly stated in the text that the temperature
and precipitation has been taken either on district basis of the
specific months or just the average of all the sample districts or the
average of overall Punjab in each of the time series.
(8) There is need to give some logical reason of selecting these
districts while ignoring the other important districts. The inclusion of
districts with very small proportion of a specific crop (Jehlem in
Cotton FEM) should be clearly justified. The robustness of the results
would highly depend on the selection of districts. By dropping an
important district or by including an unimportant district, we cannot
arrive at appropriate conclusion.
(9) The procedure of Simulation Analysis must be given in the
paper.
In the end I would say that since this paper is the part of a
continuous research effort. I do hope that the final results at Pakistan
level would be quite helpful for policy perspective.
Abdul Saboor
PMAS Arid Agriculture University, Rawalpindi.
Rehana Siddiqui <rehana@pide.org.pk> is Joint Director at the
Pakistan Institute of Development Economics, Islamabad. Ghulam Samad
<samad@pide.org.pk> is Research Economist at the Pakistan
Institute of Development Economics, Islamabad. Muhammad Nasir
<nasir84@pide.org.pk> is Staff Economist at the Pakistan Institute
of Development Economics, Islamabad. Hafiz Hanzla Jalil
<hanzala@pide.org.pk> is Research Economist at the Pakistan
Institute of Development Economics, Islamabad.
(1) Chaudhary, et al. (2002) gives the optimal temperatures range
from 20[degrees]C-35[degrees]C for the first stage, where as
25[degrees]C-31[degrees]C for the second stage. However, based on our
results, we may say that the starting pint of the optimal temperature
range varies between 26.75[degrees]C from 28[degrees]C in the second
stage.
(2) Arshad and Anwar [undated] in their online article titled
"Best Methods/ Practices to Increase per Acre Cotton Yield" on
the website of Ministry of Textile Industry gives the maximum
temperature range of 30[degrees]C-35[degrees]C. However, other online
sources have consensus upon the maximum limit of 32[degrees]C.
Table 1
Estimation Results for Wheat Production
Variable Model 1 Model 2
Contant 749.56 *** 730.09 ***
First Stage Temperature -43.11 *** -46.95 ***
First Stage Temperature^2 1.45 *** 1.66 **
Second Stage Temperature -4.58
Second Stage Temperature^2 0.16
Third Stage Temperature 0.09
Third Stage Temperature^2 -0.0004
First Stage Precipitation 0.44 *** 0.45 ***
First Stage Precipitation^2 -0.002 ** -0.002 *
Second Stage Precipitation 0 34 *** q 39 ***
Second Stage Precipitation^2 -0.002 ** -0.002 ***
Third Stage Precipitation -0.006 -0.06
Third Stage Precipitation^2 -0.0002 0.0001
Bahawalpur 306.21 *** 302.72 ***
Faisalabad 338.69 *** 339.52 ***
Jhelum -325.69 *** -324.47 ***
Lahore -324.13 *** -325.37 ***
Mianwali -108.92 *** -108.37 ***
Multan 41.65 *** 42.17 ***
Sialkot 72.18 *** 73.80 ***
[R.sup.2] 0.90 0.90
DW-Statistic 1.98 1.98
F-Statistic 58.22 *** 77.24 ***
Note: ***, ** and * represents significance at 1 percent,
5 percent and 10 percent level of significance respectively.
Table 2
Estimation Results for Rice Production
Variable Model 1 Model 2
Constant 83.64 *** 96.00 ***
First Stage Temperature 2.70 * 1.70 *
First Stage Temperature^2 -0.05 * -0.03 **
Second Stage Temperature -5.35 *** -5.06 ***
Second Stage Temperature^2 0.10 *** Q Q9 ***
Third Stage Temperature 0.12 0.65
Third Stage Temperature^2 0.02 -0.005
First Stage Precipitation 0.004
First Stage Precipitation^2 -0.00001
Second Stage Precipitation 0.0093
Second Stage Precipitation^2 -0.0001
Third Stage Precipitation -0.032
Third Stage Precipitation^2 0.0003
Bahawalpur -58.51 *** -58.62 ***
Faisalabad -45.56 *** -47 19 ***
Jhelum -60.18 *** -61.40 ***
Lahore -10.04 *** -10.00 ***
Mianwali -56.08 *** -56.78 ***
Multan -44.63 *** -44.63 ***
Sialkot 275.03 *** 278.64 ***
[R.sup.2] 0.96 0.95
DW-Statistic 2.09 2.00
F-Statistic 175.28 *** 193.90 ***
Note: ***, ** and * represents significance at 1 percent,
5 percent and 10 percent level of significance
respectively.
Table 3
Estimation Results for Cotton Production
Variable Model 1 Model 2
Constant 411.42 *** 403.52 ***
DFMT -47.46 ** -12.33 **
DFMT^2 -2.60
DFMP -1.46 * -0.50 *
DFMP^2 0.007
Bahawalpur 720.36 *** 735.1092 ***
Faisalabad -286.06 *** -289.203 ***
Jhelum -397.61 *** -406.731 ***
Mianwali -338.28 *** -355.775 ***
Multan 301.60 *** 316.5995 ***
[R.sup.2] 0.95 0.95
DW-Statistic 1.98 1.98
F-Statistic 208.74 *** 264.70 ***
Note: DFMT = Deviation from Maximum Temperature,
DFMP = Deviation from Maximum Precipitation.
***, ** and * represents significance at 1 percent, 5
percent and 10 percent level of significance respectively.
Table 4
Estimation Results for Sugarcane Production
Variable Results
Constant -30892.39 **
First Stage Temperature 165.41
First Stage Temperature^2 -3.85
Second Stage Temperature -1.92
Second Stage Temperature^2 0.079
Third Stage Temperature 133.58
Third Stage Temperature^2 -2.65
Fourth Stage Temperature 2491.88 **
Fourth Stage Temperature^2 -54.35 **
First Stage Precipitation 4.11
First Stage Precipitation^2 -0.026
Second Stage Precipitation -5.28
Second Stage Precipitation^2 0.074
Third Stage Precipitation 2.00
Third Stage Precipitation^2 -0.0039
Fourth Stage Precipitation -2.73
Fourth Stage Precipitation^2 0.013
Bahawalpur -402.95 **
Faisalabad 4656.8 **
Jhelum -960.94 **
Lahore -889.71 **
Mianwali -820.44 **
Multan -789.13 **
Sialkot -793.61 **
[R.sup.2] 0.98
DW-Statistic 1.80
F-Statistic 235.70 ***
Note: ***, ** and * represents significance at 1 percent,
5 percent and 10 percent level of significance respectively.
