Impact of Climate Variability on Rice Productivity in Pakistan

doi:10.21203/rs.3.rs-2606566/v1

Download PDF

Research Article

Impact of Climate Variability on Rice Productivity in Pakistan

https://doi.org/10.21203/rs.3.rs-2606566/v1

This work is licensed under a CC BY 4.0 License

Version 1

posted

You are reading this latest preprint version

Climate change is currently the most significant concern on the planet. This study examined the impacts of climate change on rice yield across rice growing agro-ecological zones in four districts (Swat from zone A, Mansehra from zone B, Mardan from zone C, and D.I. Khan from zone D) of Khyber Pakhtunkhwa, Pakistan based on the availability of data regarding rice production and meteorological variables. This study covered 33 years (1986–2018) data. Panel unit root tests of LLC and IPS suggested that all variable included in the model were stationary. Durbin Wu Hausman test favored fixed effect model as compared to random effect model. Diagnosis of fixed effect model results showed that it was plagued with problems of cross-sectional dependency, heteroscedasticity, and autocorrelation. Final findings were estimated using Feasible Generalized Least Squares (FGLS) model to overcome the problems in fixed effect model. Results revealed that area under crop has a positive and significant effect on rice yield. Estimated critical temperature for maximum yield of rice was 34.48 ºC in sowing stages, 35.85 ºC in vegetative stages, and 29.43 ºC in maturity stages and Rainfall in vegetative stages was 152 mm. It is concluded that rice yield showed a decline when temperature and rainfall exceed the critical levels. Government needs to use afforestation and other appropriate measures to keep the temperature from rising in these three zones. Extension department needs to provide information to rice growers for adaptation strategies to climate change.

Rice Productivity

Climate Change

Panel Data

FGLS Model

Pakistan

Climate change is happening all throughout the world, and no country is immune to its negative consequences (Hanif et al., 2010). Climate change is expected to lower crop productivity in several countries throughout the world in the following decade, according to experts (Falco et al., 2011). The rapid increase of greenhouse gases in the atmosphere has become a significant source of global warming (Aydinal and Cresser, 2008). It is known that the main source of these emissions of greenhouse gases are the developed nations of the world and around 75% of discharges related to greenhouse gases are from developed countries while developing countries contribute a low share in this emission (Farooqi et al., 2005). The greenhouse gases which include nitrogen oxide, methane, carbon dioxide and fluorinated gases [hydro fluorocarbon (HFCs), prefluorocarbons (PFCs), sulfur hexafluoride (SF6) and nitrogen trifluride (NF3)], directly affect the rainfall pattern, increase in temperature, which leads a negative effect on water and land resources. The risks are growing higher and higher due to human activity. The emission of greenhouse gases will ultimately leads toward the high temperature and high precipitation (Adams et al., 1989).

Temperatures are expected to rise roughly 3 degrees by 2040, and around 5–6 degrees by the end of the century, according to projections. Monsoon rains will be greatly reduced, but they will be much more intense. Droughts and floods are also expected over South Asia (IUCN. 2011). Evaporation and precipitation, as well as the frequency and severity of rainfall, are expected to increase, according to the climate model. Some areas may get wetter, while others may be impacted by an enhanced hydrological cycle, resulting in soil moisture loss and increased erosion (Bossello et al., 2005).

Around the globe the main challenges caused by climate change are water, energy, and food security of a country (GOP, 2017). Water supply for agriculture will be altered by climate change in a variety of ways, including shifting rainfall and production patterns. High temperatures can reduce production by promoting crop development and causing plant cell damage (Myers et al. 2017 and Bossello et al., 2005). The growth and health of some plants may be improved because of chilling and freezing environment, on the other hand few crops health, may damage due to the high temperature (Bossello et al., 2005).

Now a days in Pakistan, climatic changes has also become big concern to the researchers, water authorities, agriculturists, policy makers and other stakeholders from public and private organizations, and NGO’s have shown serious concerns over climate change issue (Rehman et al., 2017). Every year, many of glaciers in legendary mountainous region are already melting up to a meter. The rapid melting of these glaciers combines with the consequence of global warming give rise to intense precipitation which is expected to be the one of the greatest sources of river flooding. As soon as the mountainous glaciers melts completely the river flow would be projected to decrease dramatically (Ahmed and Javed, 2016). The impact of this change is now visible on agriculture sector and its production (Rehman et al., 2017)

Pakistan is one of world's largest rice producer and rice is the 2nd staple food consumed (Memon, 2013). Moreover, Pakistan is also one of the victim of severe climate change and the future climatic scenarios for Pakistan are even more challenging (Rehman et al. 2017). In Pakistan, Khyber Pakhtunkhwa province has a vast climatic conditions and various crops are cultivated here. Many researchers studied impact of climate change using different approaches. Cross-section approach deals with one time point or for a short period of time e.g., Kumar (2009), Ernest et al. (2007), Khan et al. (2018) used Ricardian cross-sectional model to examine the effect of climate on crop revenue. Time series methods deals with a set of observations each one being recorded at a specific time. Perron and Estrada (2012) and Ali et al. (2017) examined the changing climate upshots on agriculture key crops by time series method. Panel data approach deals with behaviour of entities observed across time. As Panel data approach has plentiful boons than cross section and time series, it is therefore considered best. Moreover, panel approach has more observations that give us more precise estimates (Wang, 2010). Thus, Panel data will be used to estimate the impact of climate variability on rice productivity in the study area. No analytic research work in the knowledge of this researcher has yet been undertaken to quantify climate variability impact on rice productivity in Khyber Pakhtunkhwa through panel data approach. Rice grower can be benefitted from the adaptation of changing environments, investigators can equate their results with this study and most of the important, from the finding of this study policy maker may get help in framing policies for better production of rice crop.

Universe of the study

This research was carried out in Pakistan's Khyber Pakhtunkhwa province. Khyber Pakhtunkhwa is a province in Pakistan's northwest (Khyber Pakhtunkhwa Climate Change Policy. 2016). The total area of Khyber Pakhtunkhwa is 74521 square kilometres, with mountains and rocks accounting for 70% of the land (White Paper, 2015). Extreme climatic variation is found in Khyber Pakhtunkhwa weather, northern area of Khyber Pakhtunkhwa is having extreme cold and snowy winter with a huge rate of rainfall and a pleasant summer as compared to the southern region of Khyber Pakhtunkhwa which have less severe winter, rainfall and a very hot summer. Khyber Pakhtunkhwa is divided into 4 zone namely zone A, zone B, zone C and zone D according to their climatic nature based on rainfall, temperature altitude (Khyber Pakhtunkhwa Climate Change Policy 2016) (Table 1).

Table 1

*Agro ecological zones of Khyber Pakhtunkhwa with districts*
Climatic zones	Description	Districts
Northern (A)	Higher northern mountains, Northern mountain,	Chitral, Swat, Bunir, Shangla, Upper Dir, Lower Dir
Eastern (B)	Wet mountains, Sub humid eastern mountains	Batagram, Haripur, Mansehra, Torghar, Kohistan, Abbottabad
Central (C)	Central plain valley	Peshawar, Charsadda, Mardan, Swabi, Nowshera, Kohat
Western (D)	Piedmont plain, Suleiman piedmont	Bannu, Karak, Lukky Marwat, Tank, D. I. Khan

Source: Environmental Protection Agency of Khyber Pakhtunkhwa, 2016.

Zone A which are higher mountains or northern mountains consists of six districts which are Chitral, Bunir, Swat, Shangla, Upper Dir, and Lower dir. While Batagram, Haripur, Mansehra, Torghar, Kohistan and Abbottabad comes under Zone B which are sub-humid eastern mountains or wet mountains. Similarly, the Central plain valley which include Swabi, Mardan, Charsadda, Peshawar Nowshera, and Kohat lies in Zone C. while Zone D known as Piedmont plain, Suleiman piedmont include Bannu, Karak, Lakki Marwat, Tank, D. I. Khan (Fig. 1).

Data and data sources

Panel data over time and across major rice producing districts in four climatic zones of Khyber Pakhtunkhwa was used in this study. Time period and major rice producing districts across four climatic zones were selected for which data on non-climatic and climatic variables were available. Data on rice production in thousand tons, area under rice crop in thousand hectares, yield of rice in kilograms per hectare, temperature in 0C and precipitation in mm were collected. Production, area and yield of rice data was gathered from Pakistan Bureau of Statistics and Development Statistics of Khyber Pakhtunkhwa. Data on climatic variables was collected from Pakistan Meteorological Department, Peshawar.

Because of fluctuations in climatic conditions throughout in different phases of crop growth, have varied effects on crop production, the temperature and precipitation variables were computed using three phonological stages of rice crop (Auffhammer et al., 2012). Nursery growth, transplanting, and tillering are covered in the first stage, vegetative growth, blooming, and milking are covered in the second stage, and rice maturity and harvesting are covered in the third stage. The first stage of rice lasts from June to July. The second from August to September, and the third from October to November (Ahmad et al., 2016), (Segerson and Dixon, 1999), (Cabas et al. 2010) and (Cheng and Chang, 2002).

Conceptual framework

Traditionally, Influence of climate alteration was measured by many researchers. The production function which is also known as product modelling is based on empirical or experimental production, and many researcher use this model to investigate the relationship among the yield and climatic variables (Deressa et al., 2005). In the same way, The Ricardian approach is another useful tool for determining the total climatic influence on a specific geographical area. It has been used in both developed and developing countries in a variety of geographical areas (Salvo et al., 2013) Mishra and Sahu (2014), Salvo et al. (2013), Deressa and Hassan (2009), Kabubo-Mariara and Karanja (2006), Deressa et al. (2005), Gbetibouo and Hassan (2005), Mendelsohn and Dinar (2003), Mendelsohn and Dinar (2003), Mendelsohn and Dinar (2003), Mendelsohn and Dinar (2003), Mendelsohn and Din

At the global, country, and regional levels, the time series approach has been widely used to explore the impact of climate variables on crop yields (Maharjan and Joshi, 2012). Rahim and Puay (2017), Zaied and Zouabi (2015), Amponsah et al. (2015), and Alam (2013) used time series analysis to investigate the relationship between climate variables and agricultural product yield.

Advanced Ricardian (Panel Data) Approach is also used by researcher to assess the impact of rainfall and temperature change on agriculture production such as Loum and Fogarassy (2015), Sarker et al. (2014), Dasgupta (2013), Barnwal and Kotani (2013) Dell et al. (2012), Akram (2012) Lobell et al. (2011), Brown et al. (2010), Guiteras (2007). The basic advantage of this approach is that it takes into account, the fluctuations that occur randomly year-to-year in the weather conditions (Deschenes and Greenstone, 2007).

Panel data approach

Panel data is the blend of both time series and cross-sectional data. When data is collected over more than two dimensions i.e., different cross sections and over time is known as Panel data. Variables which cannot be observed or measured, and it changes over time but not across entities, these variables can be controlled through Panel data. Moreover, Panel data can include variables at different stages of analysis.

Efficiency of the panel data is more because it has variety of advantages than cross sectional and time series data. In panel data there are more observation which gives more precise estimates, more information, less collinearity issues in data. It also resolves the misspecification problem that arises from omitted variable (Jintian, 2010).

Panel data approach can be simply presented as follows:

$$Yit=\alpha + \beta xit + \epsilon it$$

Y = dependant variable

X = independent variable

α and β = coefficients

i and t = directories for cross section and time

ε_it = error term

The error (ε_it) component is the most important factor in the panel data method equation because it tells us whether to use a fixed effect model or a random effect model (Gardinar et al. 2009).

Three approaches are used in literature for analysis of panel data (Baltagi, 2008).

i. Pooled effect model

Pooled effect also known as common constant effect is same as simple regression. In pooled effect model no panel information are used as it is observed that every variable is uncorrelated with other, ignoring panel and time.

The simple pooled model can be expressed as follows:

$$Yit= {\beta }0+\beta xit+ \epsilon it$$

Pooled effect assumes, that every observation performs in the same way that never experience autocorrelation and heteroscedasticity that is why only simple regression model can be used to estimate the model. Mostly pooled effect model are restrictive than fixed effect and random effect. Pooled model should be used when fixed effect is not efficient.

ii. Fixed effect model

Deschenes and Greenstone (2007) presented the fixed model approach which removed the problems associated to hedonic approach and is considered the ideal model due to its quick response time to sudden change in weather condition and it also controls the effect of unobserved variable (Mendelsohn and Dinar, 2009).

Because fixed effect organises all time-invariant changes between entities, the projected coefficients of fixed effect cannot be skewed by excluding time invariant properties. Because it is constant for each cross section, a time-invariant feature cannot produce such a change (Torres-Reyna, 2007).

$Yit=\alpha i + \beta xit + \epsilon it$ t = 1… T and i = 1… N (3)

Where:

$Yit$ = Dependent variable

αi = Correlated with x and unobserved time invariant individual effect for every cross section

$\epsilon it$ = Error term.

$\left(\beta \right)$ Parameter representing slope (same for all cross sections, & doesn’t changeable).

For two or more than two time period data set fixed effect model is suitable. For estimation of fixed effect model least square dummy variable (LSDV) is used.

iii. Random effect model

In random effect model time variant variables are included. The logic behind random effect is that the variation across the entities are assumed to be random and are uncorrelated with the cross section or independent variables included in the models (Torres-Reyna. 2007). The constants for each part are treated as random parameters in the random effect model.

$$Yit=(\alpha i+vi)+ \beta xit + \mu it$$

$$Yit=\alpha i+ \beta xit + vi+\mu it$$

Individual effects are randomly spread crosswise in a random effect, and αi is uncorrelated with x. When the random impact is considered to be that the unit's error terms are not associated with the cross-section, time invariant variables play a function as an explanatory variable. Random effects should be employed if variations between entities have an impact on the dependent variable.

Empirical model

Yield of rice across districts and over time is expected to be a function of area under rice crop, minimum and maximum temperature and precipitation during the crop season. Model for panel data estimation is given as:

LNYIELDit = $\beta 0$+ ${\beta }_{1}$LNAREAit + ${\beta }_{2}$LNTmax_Sit + ${\beta }_{3}$LNTmax_Vit + ${\beta }_{4}$LNTmax_Mit + ${\beta }_{5}$(LNTmax_Sit)² + ${\beta }_{6}$(LNTmax_Vit)² + ${\beta }_{7}$(LNTmax_Mit)² + ${\beta }_{8}$LNRain_Sit + ${\beta }_{9}$LNRain_Vit + ${\beta }_{10}$LNRain_Mit + ${\beta }_{11}$(LNRain_Sit)² + ${\beta }_{12}$(LNRain_Vit)² + ${\beta }_{13}$(LNRain_Mit)² + ${\beta }_{14}T+$ $Uit$ (6)

Where:

LNYIELD = Natural log of rice yield (kg/ha)

LNAREA = Natural log of area (000 ha)

LNTmax_S = Natural log of maximum temperature for sowing stage

LNTmax_V = Natural log of max. temperature for vegetative stage

LNTmax_M = Natural log of max. temperature in maturity stage

(LNTmax_S)² = Natural log of max. temperature square in sowing stage

(LNTmax_V)² = Natural log of max. temperature square in vegetative stage

(LNTmax_M)² = Natural log of max. temperature square in maturity stage

LNRain_S = Natural Log of rainfall in sowing stage

LNRain_V = Natural Log of rainfall in vegetative stage

LNRain_M = Natural Log of rainfall in maturity stage

(LNRain_S)² = Natural Log of rainfall square in sowing stage

(LNRain_V)² = Natural Log of rainfall square in vegetative stage

(LNRain_M)² = Natural Log of rainfall square in maturity stage

β1–13 = Estimated parameters

T = Trend (Time in years)

U = Error term

i = Cross section

t = Time period

Model selection

Following test was conducted for the selection of appropriate model.

1. Durbin Wu Hausman test

For selection between random and fixed effect model Durbin Wu Hausman test was conducted (Gardinar et al. 2009). Durbin Wu Hausman test construct the following hypothesis

H_O: The random effect model is efficient

H₁: The fixed effect model is efficient

Following static are used in Hausman test:

H = (β^{^FE} - β^{^RE}) ^̸ - [Var (β^{^FE}) – Var (β^{^RE})] ⁻¹ (β^{^FE} - β^{^RE}) ͠ χ² (k) (7)

Hausman test follows chi square distribution. If the chi square value is found insignificant then null hypothesis of random effect model will be accepted and vice versa.

2. Breusch-Pagan Lagrange Multiplier (LM) test

Breusch-Pagan Lagrange Multiplier (LM) test assists in deciding between pooled effects and random effects model. The LM test's hypothesis:

H₀: Variance of the random effect is zero: Var[ui] = 0

H₁: Variance of the random effect is not zero: Var[ui] ≠ 0

LM test follows chi square distribution

Critical temperature and its impact on yield

Critical temperature is the temperature where the yield is maximum or minimum. This temperature can be calculated using the formula provided below.

Critical temperature = exp (-(β₂/2* β₃)) (8)

Summary statistics of variables

Table 2 shows the summary statistics for the variables included in the model. The total number of observations is 132, which is equal to N = 4, T = 33, and N*T = 132. Having standard deviation of 0.22, with the natural log (ln) mean of yield was 7.67 kg/ha. The natural log of area for the second variable ranged from 0.26–3.02.53 hac, having mean of 1.37 and a standard deviation of 0.75. Log of maximum temperature for sowing months mean is 3.53 with standard deviation of 0.84 and has a mean range is 3.20–3.71. Log of maximum temperature for vegetative months ranges in between 3.33–3.68 with mean value of 3.51 having standard deviation of 0.67. log of maximum temperature for maturity months mean value is 3.40, has standard deviation of 0.75, mean ranging in between 3.07–3.55. Temperature maximum square for sowing months mean is 12.51 with standard deviation of 0.59, range of mean is 10.24–13.80. Square of Temp-max for vegetative months range is 11.09–13.55 with mean value of 12.33 having standard deviation of 0.47. Temp-max square for maturity months mean is 11.60 with standard deviation of 0.50, mean range is 9.42–12.60. Log of rainfall for sowing months mean is 4.42 with standard deviation of 0.49 and has a mean range is 3.26–5.21. LNRain_V ranges in between 4.38–5.22 with a mean value of 4.92 having standard deviation of 0.17. LNRain_M mean value is 3.80, has standard deviation of 0.49, mean ranging in between 2.75–4.61.

Table 2

*Summary statistics*
Variables	Obs	Mean	Std. Dev.	Min.	Max.
LNYIELD (YIELD)	132	7.67 (2196.17)	0.22 (458.35)	6.97 (1070.40)	8.07 (3203.70)
LNAREA (AREA)	132	1.37 (5.14)	0.75 (3.62)	0.26 (1.3)	3.02 (20.6)
LNTmax_S (Tmax_S)	132	3.53 (34.46)	0.84 (2.85)	3.20 (24.55)	3.71 (41.1)
LNTmax_V (Tmax_V)	132	3.51 (33.58)	0.67 (2.27)	3.33 (27.95)	3.68 (39.7)
LNTmax_M (Tmax_M)	132	3.40 (30.23)	0.75 (2.23)	3.07 (21.55)	3.55 (34.85)
(LNTmax_S)2 (Tmax_S)2	132	12.51 (1195.67)	0.59 (196.06)	10.24 (602.70)	13.80 (1689.21)
(LNTmax_V)2 (Tmax_V)2	132	12.33 (1133.38)	0.47 (154.84)	11.09 (781.20)	13.55 (1576.09)
(LNTmax_M)2 (Tmax_M)2	132	11.60 (919.28)	0.50 (134.83)	9.42 (464.40)	12.60 (1214.523)
LNRain_S (Rain_S)	132	4.42 (92.86)	0.49 (37.59)	3.26 (26.25)	5.21 (184.5)
LNRain_V (Rain_V)	132	4.92 (139.01)	0.17 (23.40)	4.38 (80)	5.22 (185)
LNRain_M (Rain_M)	132	3.80 (50.13)	0.49 (22.99)	2.75 (15.65)	4.61 (101)
(LNRain_S)2 (Rain_S)2	132	19.81 (3615.49)	4.20 (5023.82)	10.67 (15.60)	27.22 (34040.25)
(LNRain_V)2 (Rain_V2)	132	24.24 (31864.74)	1.74 (53353.03)	19.20 (12.60)	27.25 (373932.3)
(LNRain_M)2 (Rain_M)2	132	14.69 (4187.34)	3.67 (7261.17)	7.56 (3.61)	21.29 (43701.9)
Source: Authors’ estimates from panel data, 1986–2018

Rainfall square for sowing months mean is 19.81 with standard deviation of 4.20 range of mean is 10.67–27.22. LNRain_V2 range is 19.20-27.15 with mean value of 24.24 having standard deviation of 1.74. Rainfall square for maturity months mean is 14.69 with standard deviation of 3.67, mean range is 7.56–21.29.

Panel un i t root tests

Panel unit root tests were used to ensure that the series were stationary as shown below in Table 3. All fourteen variables in the model were tested for stationarity. Unit root testing for the LLC (Levin-Lin-Chu) and IPS (Im-Pesaran-Shin) panels. The LLC and IPS tests were used to determine yield stationarity. When the yield was tested, both with and without trend, the results showed that the yield was stagnant. Both the LLC and IPS tests were used to confirm stationarity in the area, which is the model's second variable. When LLC tests were conducted using merely a trend, the results revealed that the area is non-stationary. When the intercept was combined with the trend, the variable became stationary. The IPS p-value for trend and intercept is extremely high, indicating that the variable is stationary. The same test was carried out to see the stationarity of all the temperature variables (LNTmax S, LNTmax V, LNTmax M, and their Squares) for all the phonological phases, including sowing, vegetative, and maturity months. In both situations, the panel unit root tests indicate that this variable is stationary with trend and with trend and intercept. The discovered P-values are extremely significant.

Table 3

*Panel unit root tests*
		Without Trend		With Trend
Variables		Statistic	P-values	Statistic	P-values
Yield (kg/ha)	LLC	-3.4765	0.0003***	-3.4884	0.0002***
Yield (kg/ha)	IPS	-4.4368	0.0000***	-5.1596	0.0000***
Area (ha)	LLC	-2.2174	0.0133***	-4.2094	0.0000***
Area (ha)	IPS	-1.3381	0.0904	-3.3055	0.0005***
LNTmax_S (ºC)	LLC	-6.3515	0.0000***	-5.6485	0.0000***
LNTmax_S (ºC)	IPS	-6.5976	0.0000***	-6.5932	0.0000***
LNTmax_V (ºC)	LLC	-3.2995	0.0005***	-3.1185	0.0009***
LNTmax_V (ºC)	IPS	-6.2707	0.0000***	-6.8737	0.0000***
LNTmax_M (ºC)	LLC	-2.9292	0.0017***	-2.5273	0.0057***
LNTmax_M (ºC)	IPS	-4.6180	0.0000***	-4.9888	0.0000***
LNTmax_S2 (ºC)	LLC	-6.4391	0.0000***	-6.4391	0.0000***
LNTmax_S2 (ºC)	IPS	-6.5702	0.0000***	-6.5709	0.0000***
LNTmax_V2 (ºC)	LLC	-3.3158	0.0000***	-3.1350	0.0009***
LNTmax_V2 (ºC)	IPS	-6.2780	0.0000***	-6.8801	0.0000***
LNTmax_M2 (ºC)	LLC	-2.9375	0.0017***	-2.5341	0.0056***
LNTmax_M2 (ºC)	IPS	-4.6173	0.0000***	-4.9876	0.0000***
LNRain_S (mm)	LLC	-5.1500	0.0000***	-3.8847	0.0001***
LNRain_S (mm)	IPS	-6.7194	0.0000***	-6.7728	0.0000***
LNRain_V (mm)	LLC	-3.7117	0.0001***	-3.6833	0.0001***
LNRain_V (mm)	IPS	-4.6492	0.0000***	-5.0655	0.0000***
LNRain_M (mm)	LLC	-3.6349	0.0001***	-3.3057	0.0005***
LNRain_M (mm)	IPS	-4.6835	0.0000***	-5.2976	0.0000***
LNRain_S2 (mm)	LLC	-5.1252	0.0000***	-3.8743	0.0001***
LNRain_S2 (mm)	IPS	-6.7331	0.0000***	-6.7906	0.0000***
LNRain_V2 (mm)	LLC	-3.7379	0.0001***	-3.7124	0.0001***
LNRain_V2 (mm)	IPS	-4.6520	0.0000***	-50681	0.0000***
LNRain_M2 (mm)	LLC	-3.5623	0.0002***	-3.2469	0.0006***
LNRain_M2 (mm)	IPS	-4.7020	0.0000***	-5.3317	0.0000***

Source: Valued from data, 1986–2018.

Note level of significance, ***p < 0.01, **p < 0.05

Model specification test

Durb i n Wu Hausman test (FE vs RE)

To choose between fixed effects and random effects models, the Hausman test was used. The outcome of this test is shown in the table below.

Prob > chi2 = 0.0000

Due to the high significance of the p-value, the best fitting model is the fixed effects model, which was selected for further research.

Test for cross-sectional dependence

Pesaran's test of cross-sectional dependence was used to determine whether data were cross-sectionally dependent. The following are the outcomes.

Pesaran's test of cross-sectional independence = -0.256, Pr = 0.7981

Average absolute value of the off-diagonal elements = 0.210

The test's non-significant value indicates that there is no cross-sectional dependence in the data.

Wald test for heteroskedasticity

Using STATA 12, the Wald test for heteroskedasticity was performed, giving the following findings.

chi2 (4) = 41.82

Prob > chi2 = 0.0000

The presence of a very significant value indicates that the data set has an issue with heteroscedasticity.

Serial correlation

The autocorrelation of panel data was also examined using the Wooldridge autocorrelation test. The following are the outcomes.

Prob > F = 0.1524

This demonstrated that there is no autocorrelation in data.

FGLS estimates for panel data (1986–2018)

The findings of the variables used in the study in the model are shown in Table 4. Results revealed that, 1% increase in area has a positive impact on yield of rice, increasing it by 0.15 percent, the result are in line to Hussain, A., & Bangash, R. (2017). Rice yield is affected by the average maximum temperature for sowing months and the average maximum temperature for sowing month square.

Table 4

*FGLS estimates for panel data (1986–2018)*
Coefficients			Generalized least squares
Panels			Heteroskedastic
Correlation			No autocorrelation
Estimated covariance’s		04	No. of obs	132
Estimated autocorrelations		0	No. of groups	4
Estimated coefficients		15	Time period	33
			Wald chi2(14)	199.39
			Prob > chi2	0.0000
Variables	Coefficients	Std. Err.	z-values	P>\|z\|
LNAREA	.1566248	.0257844	6.07	0.000***
LNTmax_S	23.88759	7.271321	3.29	0.001***
LNTmax_V	− 56.28832	24.3481	-2.31	0.021**
LNTmax_M	− 57.00566	18.93379	-3.01	0.003***
(LNTmax_S)²	− 3.373466	1.043716	-3.23	0.001***
(LNTmax_V)²	7.862926	3.439045	2.29	0.022**
(LNTmax_M)²	8.426952	2.764805	3.05	0.002***
LNRain_S	.1950235	.5702567	0.34	0.732^ns
LNRain_V	15.62666	3.94158	3.96	0.000***
LNRain_M	− .3036746	.3997954	-0.76	0.448^ns
(LNRain_S)²	− .0170238	.070239	-0.24	0.808^ns
(LNRain_V)²	− 1.554174	.4025644	-3.86	0.000***
(LNRain_M)²	.0394931	.0530368	0.74	0.456^ns
Trend	.0089139	.0014007	6.36	0.000***
Constant	122.8388	40.9811	3.00	0.003***

Source: Projected from panel data, 1986–2018.

Note level of significance, ***p < 0.01, **p < 0.05, ns shows non-significant

The co-efficient of LNTmax S is positive, whereas the co-efficient of its square (LNTmax S2) is negative, the same results were also observed by the Ahmed et al. 2016 & Cabas et al. (2010). This demonstrates that a rise in temperature causes an increase in rice yield at first. When the temperature exceeds above the critical temperature i.e., 34.48 ºC. it shows a fall after reaching its maximum at the critical temperature.

Temperature maximum of vegetative months and maturity months P-values are significant, and are in opposite direction as compare to the Temperature Maximum in Sowing months, as the LNTmax_V and LNTmax_M has a negative co-efficient and there Squares has positive co-efficient. This mean that temperature in vegetative and maturity months has a negative impact on rice yield. Rainfall in Vegetative months P-value is highly significant and has a positive effect on rice yield. Co-efficient of LNRainfall_V is positive and its square (LNRainfall_V2) is negative. The results are in line with Ahmed et al. 2016, Cabas et al. (2010), Chaudhary, et al. (2002), & Siddiqui et al. (2012).

This means that a rise in rainfall causes an increase in rice yield at first. When the rainfall exceeds above the critical rainfall i.e., 152.55 mm. It shows a fall after reaching its maximum at the critical rainfall.

Impact of temperature on rice yield

Rice yield in response to temperature in sowing stage

Calculating the rice yield in sowing months using the formula (8).

Critical temperature = 34.48 ºC

According to the critical temperature, the yield of rice reached a maximum temperature of 34.48 ºC during the sowing stages of the crop. Over all districts, the highest yield at critical temperature was assessed to be 2218.493 kg/ha. When the temperature exceeds this critical temperature, the yield begins falling. The Fig. 2 below was made to demonstrate the relationship.

Rice yield in response to temperature in vegetative stage

Calculating the rice yield in sowing months using the formula (3.9).

Critical temperature = 35.85 ºC

The critical temperature in vegetative stage demonstrates that the rice yield would be minimum when temperature reaches 35.85ºC during vegetative months of the crop. However, the yield would increase when the value exceeds the critical temperature. The highest yield at critical temperature for over all the districts is expected to be 2028.23 kg/ha, as shown below in Fig. 3.

Rice yield in response to temperature in maturity stage

Formula (8) was used to calculate the critical temperature in maturity stage.

Critical temperature = 29.43 ºC

Figure 4 shows that rice yield would be highest when the temperature reaches 29.43°C during the crop's maturity stage. It is projected that the maximum yield at critical temperature is 1990 kg/ha. When the temperature rises above the critical temperature, the yield will begin to rise.

Rice yield in response to rainfall in vegetative stage

Formula (8) was used to calculate the critical rainfall in vegetative stage for rice yield.

Critical rainfall = 152.52 mm

According to the critical rainfall, the rice yield would be the highest when rainfall hits 152.52 mm during the vegetative stage of the crop. The maximum yield at critical rainfall is expected to be 2190 kg/ha. When the rainfall exceeds critical rainfall, the yield begins to decline. The Fig. 5 below was made to demonstrate the relationship.

FGLS estimation technique was used for estimating the impact of climate variability on rice productivity among the nominated districts of Khyber Pakhtunkhwa. Panel data from 1986–2018 was used, included variables i.e area, average maximum temperature in sowing, vegetative and harvesting stages, average maximum temperature square in sowing, vegetative, and harvesting stages, average rainfall, and rainfall square in sowing, vegetative and harvesting stages.

Results revealed that area and maximum temperature in sowing stage have positive effect on rice yield. Temperature square in sowing stage was detected as negatively effecting the rice yield. Maximum Temperature in vegetative and maturity stages and its square has also a negative impact. While rainfall in vegetative stages have a positive impact on yield of rice in nominated districts. Critical temperature for maximum yield of rice was calculated to be 34.48 ºC in sowing stages, 35.85 ºC in vegetative stages, and 29.43 ºC in maturity stages and Rainfall vegetative stages was 152 mm. It was projected that yield in the study districts showed a decline when temperature and rainfall rise above critical level.

Policymakers need to encourage rice cultivators in Mansehra to allot more cultivated land for rice cultivation, as the district's uppermost temperature in the sowing month is 33.80 ºC, which is below the critical temperature (34.48 ºC), implying that rice yield will rise as the temperature rises.

The highest maximum temperature in the districts of D.I. Khan and Mardan is 41.20 ºC and 38.32 ºC, respectively, which is considerably over the average critical temperature. As a result, the government needs to use afforestation and other appropriate measures to keep the temperature from rising.

Extension department needs to provide information to rice growers for adaptation strategies regarding climate change and its effects on rice yield.

Funding

The authors declare that no funds, grants, or other support were received during the preparation of this manuscript.

Competing Interests

The authors have no relevant financial or non-financial interests to disclose

Author Contributions

Data collection and first draft of the manuscript was written by Hikmatyar Khan. Data analysis was performed by Shahid Ali. Manuscript final review and editing was performed by Farhana Gul.

Ethical Approval N/A

Consent to Participate

All authors are willing to participate

Consent to Publish

All authors are willing to pulish their article in your journal.

Availability of data and materials

Data and materials are available and will be provided upon request.

Akram, N. 2012. Is climate change hindering economic growth of Asian economies? Asia-Pacific Development Journal. 19(2): 1-18.
Adams, R. M. 1989. Global climate change and agriculture: An economic perspective. American Journal of Agricultural Economics. 71(5): 1272–79.
Adedeji, I. A., N. E. Tiku, P. R. Waziri-Ugwu and S. O. Sanusi. 2017. The eﬀect of climate change on rice production in Adamawa State, Nigeria. Agroeconomia Croatica. 7(1): 1-13.
Ahmed, V. and A. Javed. 2016. National study on agriculture investment in Pakistan. https://www.researchgate.net/publication/327112239.
Ahmad, M., M. Nawaz, M. Iqbal.,S. A. Javed. 2014. Analysing the Impact of Climate Change on Rice Productivity in Pakistan. MPRA Paper No. 72861, posted 10 Aug 2016.
Ali, S., Y. Liu, M. Ishaq, T. Shah, Abdullah, A. Ilyas and Izhar Ud Din. 2017. Climate Change and Its Impact on the yield of major food crops: evidence from Pakistan. Foods. 6(6): 39.
Alam, Q. 2013.Climate change, agricultural productivity and economic growth in India. The Bounds Test Analysis. International Journal of Applied Research and Studies. 2(11): 1-14.
Amponsah, L., G. K. Hoggar and S. Y. Asuamah. 2015. Climate change and agriculture, modeling the impact of carbon dioxide emission on cereal yield in Ghana. Agriculture and Food Sciences Research. 2(5): 32-38.
Auffhammer, A., V. Ramanathan, and J. R. Vincent (2012) Climate Change, and the Monsoon, and Rice Yield in India. Climate Change 111, 411–424.
Aydinalp, C. and M. S. Cresser. 2008. The effects of global climate change on agriculture. American-Eurasian Journal of Agriculture and Environmental Science. 3(5): 672-676.
Barnwal, P., & K, Kotani. 2010. Impact of variation in climatic factors on crop yield: A case of rice crop in Andhra Pradesh, India. Economics and Management series, 17.
Barnwal, P. and K. Kotani. 2013. Climatic impacts across agricultural crop yield distributions, an application of quantile regression on rice crops in Andhra Pradesh India. Ecological Economics. 87: 95-109.
Bosello, F. and J. Zhang. 2005. Assessing climate change impacts on agriculture. Nota di Lavoro, Fondazione Eni Enrico Mattei, No. 94.2005.
Brown, C., R. Meeks., Y. Ghile and K. Hunu. 2010. Empirical analysis of the effects of climate variables on national level economic growth. World Bank’s World Development, policy research working paper 5357.
Cabas, J., A. Weersink, and E. Olala (2010) Crop Yield Response to Economic, Site and Climatic Variables. Climatic Change 101, 599-616
Chaudhary, R. C., Nanda, J. S. and Tran, D.V. (2002), “Guidelines for Identification of Field Constraints to Rice Production”, International Rice Commission, Food and Agriculture Organization of the United Nations, Room
Cheng, C. and Chang. 2002. The potential impact of climate change on Taiwan’s agriculture. Agricultural Economics. 27: 51-64.
Ernest, L., C. Molua, and M. Lambi. 2007.The Economic Impact of Climate Change on Agriculture in Cameroon. Policy Research Working Paper, 4364: 22.
Dasgupta, S. 2013. Impact of climate change on crop yields with implications for food security and poverty alleviation, http://www.climateimpacts-2013.org/files/cwi_dasgupta.
Dell, M., B.F. Jones and B. Olken. 2012. Temperature shºCks and economic growth, evidence from the last half century, American Economic Journal, vol.4 (3): 66-95.
Deressa, T., R. Hassan and D. Poonyth. 2005. Measuring the impact of climate change on South African agriculture, the case of sugarcane growing regions, Agrekon. 44(4): 524-542.
Deressa, T. and R. M. Hassan. 2009, Economic impact of climate change on crop production in Ethiopia: Evidence from cross-section measures. Journal of African Economies.18: 529-554.
Farooqi, A.B., A. H. Khan and H. Mir. 2005. Climate change perspective in Pakistan. Pakistan Journal of Metrology. 2: 11-21.
Francesco, B. & Z, Jian. 2005. Assessing Climate Change Impacts: Agriculture, Notadi Lavoro, Fondazione Eni Enrico Mattei, No. 94.
Gbetibouo, G. A and R.M. Hassan. 2005. Measuring the economic impact of climate change on major South African field crops. A Ricardian Approach, Global and Planetary Change. 47(2-4): 143-152.
Guiteras, R. 2007. The impact of climate change on Indian agriculture, job market paper. Department of Economics, University of Maryland, College Park, pp.1-52.
GOP. 2017. Economic Survey of Pakistan 2016-2017. Ministry of Finance, Economic Advisory Wing, Finance Division, Islamabad, Pakistan
GOP. 2018. Economic Survey of Pakistan 2017-2018. Ministry of Finance, Economic Advisory Wing, Finance Division, Islamabad, Pakistan
Government of Khyber Pakhtunkhwa. 2016. Khyber Pakhtunkhwa climate change policy. 2016. Environmental Protection Agency Government of Khyber Pakhtunkhwa. Forestry, Environment & Wildlife Department.
Hanif, U., S. H. Syed., R. Ahmad., and K. A. Malik. 2010. Economic Impact of Climate Change on the Agricultural Sector of Punjab. The Pakistan Development Review. 49: 771–798.
Hussain, A., & Bangash, R. (2017). Impact of climate change on crops’ productivity across selected agro-ecological zones in Pakistan. The Pakistan Development Review, 56(2), 163-187.
Iqbal, K. and A. Siddique. 2008. The impacts of climate change on agricultural productivity: evidence from panel data of Bangladesh. The University of Western Australia. Discussion paper 14: 29.
IUCN. 2011. International union for conservation of nature. Climate Change Vulnerabilities in Agriculture Sector of Pakistan, IUCN and Government of Pakistan, Islamabad.
Khan, A., S. Ali, S.A. Shah and M. Fayaz. 2018. Impact of temperature and precipitation on net revenue of maize growers in Khyber Pakhtunkhwa, Pakistan. Sarhad Journal of Agriculture. 34(4): 729-739.
Kumar, K. S. 2009. Climate sensitivity of Indian agriculture: do spatial effects matter. SANDEE Working Paper No 45-09.
Kurukulasuriya, P., R. Mendelsohn, R. Hassan, J. Benin, T. Deressa, M. Diop, H. M. Eid, K. Y. Fosu, G. Gbetibuo, and S. Jain. 2006. Will African agriculture survive climate change? World Bank Economy. 20: 367–388.
Lee, J., D. Nadolnyak., and V. Hartarska. 2012 Impact of Climate Change on Agricultural Production in Asian Countries Evidence from Panel. Presentation at the Southern Agricultural Economics. February 4-7.
Lobell, D. B., W. Schlenker and J. Costa-Roberts. 2011. Climate trends and global crop production since 1980. Science. 333: 616-620.
Loum, A. and C. Fogarassy. 2015. The effects of climate change on cereals yield of production and food security in Gambia, Applied Studies in Agribusiness and Commerce- APSTRACT. 9(4): 83-92.
Luo, X. S., D. Muleta., Z. Hu., H. Tang., Z. Zhao., S. Shen and B. L. Lee. 2017. Intensive development and agricultural adaptation to climate change. Current Opinion in Environmental Sustainability. 24: 78-83.
Maharajan, K. L. and J. N. Prakash. 2012. Climate change, agriculture and rural livelihoods in developing countries with references to Nepal, Hiroshima international center for environmental cooperation, graduate school for international development and cooperation, Hiroshima University, pp. 1-203.
Memon, N. A. 2013. Rice: Important cash crop of Pakistan. Pakistan Food Journal, 21-23.
Mendelsohn, R and A. Dinar. 2003. Climate, Water, and Agriculture. Land Economics. 79(3): 328-341.
Mendelsohn, R., A. Dinar., and L. Williams. 2006. The distributional impact of climate change on rich and poor countries. Environment and Development Economics. 11(2): 159-178.
Mendelsohn, R. 2014. The impact of climate change on agriculture in Asia. Journal of Integrative Agriculture. 13(4): 660-665.
Mishraa, D., and N. C. Sahub. 2014. Economic Impact of Climate Change on Agriculture Sector of Coastal Odisha. APCBEE PrºCedia. 10: 241–245.
Myers, S. S., M. R. Smith., S. Guth., C. D. Golden., B. Vaitla., N. D. Dangour and P. Huybers. 2017. Climate change and global food system: potential impacts on food security and under nutrition. Annual of Public Health. 38: 259-277.
Ochieng, J., L. Kirimi., and M. Mathenge. 2016. Effects of climate variability and change on agricultural production: The case of small scale farmers in Kenya. NJAS-Wageningen Journal of Life Sciences. 77: 71-78.
Kabubo-Mariara, J. and F. K. Karanja. 2006. The economic impact of climate change on Kenyan crop agriculture. A Ricardian Approach, Third World Congress of Environmental and Resource Economics: Kyoto Japan, July 3-7.
Peng, S., J. Huang., J. E. Sheehy., R. E. Laza., R. M. Visperas., X. Zhong., G. S. Centeno., G. S. Khush and K. G. Cassman. 2004. Rice yields decline with higher night temperature from global warming. P. Natl. Acad. Sci. 101: 9971-9975.
Perron, P., and F. Estrada. 2012. Breaks, trends and the attribution of climate change: a time-series analysis (No. WP2012-013). Boston University-Department of Economics.
Rahim, S. and T. G. Puay. 2017. The impact of climate on economic growth in Malaysia. Journal of Advanced Research in Business and Management Studies. 6(2): 108-119.
Rahman, M. A., A. Saboor., I. A. Baig., U. Shakoor and H. Kanwal., 2017. An investigation of the impact of climate change on rice crop in Pakistan: a multivariate analysis. Pakistan Journal of Agricultural Sciences. 54(3): 561-566.
Raza, S. H, and S. Anwar. 2015. Effect of climate change on Rice Production in Punjab. Journal of Biology, Agriculture and Healthcare. 5(12): 831-839.
Salvo, M.D., R. Raffaelli and R. Moser. 2013. The impact of climate change on permanent crops in an Alpine region: A Ricardian analysis. Agricultural Systems. 118: 23-32.
Sarker, M. D., A. Rashid., K. Alam and J. Gow. 2014. Assessing the effects of climate change on rice yields, an econometric investigation using Bangladeshi panel data. Economic Analysis and Policy. 44: 405-416
Segerson, K. and B. L. Dixon (1999) Climate Change and Agriculture: The role of farmer adaptation. In R. Mendelsohn and J. E. Neumann (eds.) The Impact of Climate Change on the United States Economics. Chap 4. Cambridge University Press, Cambridge, pp. 75-93.
Shakoor, U., A. Saboor, I. Baig, A. Afzal and A. Rahman. 2015. Climate variability impacts on rice crop production in Pakistan. Pakistan Journal Agriculture and Research. 28(1): 44-49.
Siddiqui, R., Samad, G., Nasir, M., & Jalil, H. H. (2012). The impact of climate change on major agricultural crops: evidence from Punjab, Pakistan. The Pakistan Development Review, 261-274.
Torres-Reyna, O. 2007. Panel data analysis fixed and random effects using Stata (v. 4.2). Data & Statistical Services, Princeton University.
Umer, U. J. B., and R. A. Ranjan. 2014. Deterioration of Agricultural Productivity Due to Climate Change in Haryana. University Library of Munich, Germany. Munich Personal RePEc Archive. https://mpra.ub.uni-muenchen.de/72654.
Wang, J. 2010. Food security, food prices and climate change in China: A dynamic panel data analysis. Agriculture and Agricultural Science PrºCedia. 1: 321–324.
Zaied, Y. B. and O. Zouabi. 2015. Climate change impacts on agriculture, a panel co-integration approach and application to Tunisia, MPRA Paper No. 64711.
Zhang, P., J. Zhang and M. Chen. 2017. Economic impact of climate change on agriculture: The importance of additional climatic variables other than temperature and precipitation in china. Journal of Environmental Economics and Management. 83: 8-31.

No competing interests reported.

Download PDF

Version 1

posted

You are reading this latest preprint version

Impact of Climate Variability on Rice Productivity in Pakistan

Status:

Version 1

Abstract

Figures

Introduction

Materials And Methods

Universe of the study

Data and data sources

Conceptual framework

Panel data approach

i. Pooled effect model

ii. Fixed effect model

iii. Random effect model

Empirical model

Model selection

1. Durbin Wu Hausman test

2. Breusch-Pagan Lagrange Multiplier (LM) test

Critical temperature and its impact on yield

Results And Discussion

Summary statistics of variables

Model specification test

Test for cross-sectional dependence

Wald test for heteroskedasticity

Serial correlation

FGLS estimates for panel data (1986–2018)

Impact of temperature on rice yield

Rice yield in response to temperature in sowing stage

Rice yield in response to temperature in vegetative stage

Rice yield in response to temperature in maturity stage

Rice yield in response to rainfall in vegetative stage

Conclusions And Recommendations

Declarations

References

Additional Declarations

Status:

Version 1