Impact of Climate Change on Rice and Wheat Yield in Punjab State of India: A District-Level Analysis


 The present study aims to examine the impact of climate change on wheat and rice yield of the Punjab state of India. Using district-level panel data from 1981 to 2017, the study employs fully modified ordinary least squares (FMOLS), dynamic ordinary least squares (DOLS), and pooed mean group (PMG) approaches. The Pedroni cointegration has established a long-run relationship of climate variables with rice and wheat crops. The results of FMOLS and DOLS show that minimum temperature has a positive effect on both wheat and rice, while maximum temperature is found to be negatively contributing to both the crops. Rainfall has a significant adverse effect on wheat yield. Seasonal rainfall has been detrimental to wheat and rice yield in the study period, indicating that excess rainfall proved counterproductive. Pooled mean group (PMG) model confirms the robustness of the results obtained by FMOLS and DOLS techniques. Moreover, Dumitrescu-Hurlin causality test has revealed a unidirectional causality running from minimum temperature, rainfall & maximum temperature to rice and wheat yield. The findings of the study suggest that the government should invest in developing stress-tolerant varieties of wheat and rice, managing crop residuals to curb further environmental effect and sustain natural resources for ensuring food security.

with sea levels and weather extremes (Brown and Funk, 2008;Kotschi 2007). Extreme weather events (flooding 79 & droughts) often lead to the obliteration of crops and food shortage (IPCC, 2018). As a result, both developed 80 and developing countries are affected by climate change. India is known to be one of the developing nations that 81 are most vulnerable to climate change owing to its immense reliance on agriculture for jobs. The dependency on 82 climate variables like temperature and rainfall makes agriculture production very sensitive to changes in climate.

83
In India, more than half the agricultural land and about 70-80 per cent of the total land is irrigated through

91
When Indian agriculture is considered, Punjab is found to be among the front-runner states in terms of total 92 production and the producer of major food grains (wheat and rice). So, studying the regional conditions of the

119
This section deals with a summary of the literature related to climate change and agricultural production. We have 120 included those studies which directly or indirectly investigated the effects of climate change on wheat, rice, and 121 cereal crops at regional, single country, and multiple countries levels.

143
Given the importance of wheat and rice as major crops of Punjab in providing food security, we take the data for 144 wheat and rice separately for the analysis. Our panel data covers 12 major wheat and rice producing districts of 145 Punjab during the period of 1981-2017. Table 2 describes the variables, their symbols, units and their sources 146 through which data have been collected. From the literature, maximum and minimum temperature along with 147 rainfall have been identified as the key determinants of rice and wheat yield. The trends of these key dependent 148 and independent variables have been shown in Figure 1-8. Apart from that, the production of these crops also 149 depends on the cultivated area, which relates to how much area these crops are sown in. Therefore, based on these 150 variables, the following model specifications are framed:

179
The equation for LLC is following:

181
Where y refers to the variable being tested for unit root, ∆ denotes the differentiated form of the variable, is less 182 than zero for the non-existence of unit root against the null hypothesis of ≥ 0.

183
The equation for IPS is following: Where the null hypothesis is that = 1.

186
The null hypothesis under the above methods has stated the existence of unit root (stationarity), while the 187 alternative has shown non-stationarity in the panel data (Akpolat, 2014). If non-stationarity exists, then 188 cointegration is estimated to get a consistent and efficient estimation. For cointegration, the Pedroni cointegration 189 method is applied, which has developed seven different tests to determine the existence of panel cointegration.

194
There is a presence of cointegration when is significantly different from zero.

206
Here ̂ * represents DOLS regression parameter applied to cross-sections n.

207
The FMOLS is a non-parametric technique, and the DOLS is a parametric procedure used to eliminate the  230 ) α i represents individual effects, which are supposed to be fixed in the time 232 dimension, k denotes the lag orders and is assumed the same for all cross-sectional units, γ i (k) and β i (k) , 233 respectively, represent lag and slope parameters that differ across groups.

Results and Discussion
236 Table 3 and 4 presents the descriptive statistics for all the variables corresponding to wheat and rice from 1980-

248
In order to determine the long-run relationship among the variables, it is necessary to fulfil the pre-conditions of 249 possessing unit root (non-stationarity) among all variables, either at level or at first difference.

276
Insert [ Table 8] 277 Table 8 has shown FMOLS & DOLS long-run estimations or the impact of various climatic variables described 278 in Table 1 on the productivity of rice. The empirical estimations of FMOLS have presented that the coefficient of 279 minimum temperature has shown a positive change in rice yield, which is statistically significant at 1% level.

280
Additionally, a 1° C increase in temperature has increased rice yield by 2.309%. The findings of the paper can be 281 supported by various national, regional and international empirical estimations. As the countries possessing

289
The coefficient of maximum temperature is found to be negatively affected the rice yield, where a 1°C increase

314
Impact on Wheat Yield 315

331
The coefficient of rainfall is negatively determining the wheat yield at 5%, which is empirically tested or discussed 332 under various studies and discussion, because of change in rainfall intensity which is significantly correlated with

370
The results of Table 13 have shown a unidirectional causal relationship between wheat and various variables like 371 yield to minimum temperature, maximum temperature to wheat yield, rainfall to yield, wheat area to wheat yield, 372 maximum to minimum temperature, wheat rainfall to minimum temperature and maximum temperature at 1% 373 level.

375
The study aims to examine the impact of climatic variables on the major crops, i.e., rice and wheat, in the Punjab

387
Dumitrescu-Hurlin causality test reveals a bidirectional causality between rice yield and harvested area of rice, 388 while a unidirectional causality runs from the harvested area of wheat to wheat yield.

389
As shown above, most of the climatic factors except minimum temperature and control variable have a negative 390 effect on rice and wheat. So, it is an alarming stage to address the issue of agricultural sustainability. Because if 391 rainfall is affecting the yield adversely, either in the case of wheat or rice, then the concern for the government 392 and policymakers is to follow adaptative and mitigation policies which will help farmers to sustain their income          Note: *, **, and *** show the significance level at 10%, 5%, and 1% respectively.