3.2 Risk in Wheat Production
First, to satisfy stationary condition we performed unit root test for time series data.
Both variations in precipitation and temperature were found stationary at level, I(0),
while all other climatic and non-climatic variables became stationary at first difference, I(1). The results of unit root test are presented in Appendix 3. Second,
Wald test was employed to assess about model significance as suggested by
Kohansla and Aliabadi (2014) and Ahmad et al. (2014). Wald test results are
presented in Appendix 4; however, insignificant P value rejected the null hypothesis and significant intercept was found. Hence, the model used in this study was
verified as significant. Third, as stochastic frontier analysis and production risk both
capture the variation in output, we tested the heteroscedasticity as suggested by
Wood and Mendelsohn (2015). The Breusch–Pagan test was applied and is presented in Appendix 5. Null hypothesis for Breusch–Pagan was that the model holds
homoscedasticity or constant variance and is rejected due to insignificant P value.
The test revealed the presence of heteroscedasticity (Production Risk). Finally,
two-step stochastic production function was estimated and the estimates of risk in
wheat crop production are presented in Table 4.
The estimation results for risk in yield revealed that the relationship between risk
and time (technology) was significant and negative sign showed a decrease in risk
due to improvement in technology every year, as improved seed varieties and
advance cultivation technology are encouraging the wheat production and declining
the level of risk. Similar significant and risk-decreasing trend by technology was
reported by Holst et al. (2011) for wheat production in China. Area under cultivation and amount of fertilizer applied were not significant but the positive coefficient exposed increasing risk by increasing both area and fertilizer application.
This may be because of limited management skills and inadequate
information/knowledge of balance use of fertilizer to the farmers in the study area,
as the same was observed by Shakoor et al. (2015) for rice crop, and Zulfiqar and
Ashfaq (2014) for wheat in Pakistan. Holst et al. (2011) reported decreasing but
non-significant risk due to increase in area in China. Conversely, Poudel and Kotani
(2013) reported a significant decreasing trend in risk for wheat in Nepal.
Moreover, no significant production risk was found due to temperature during all
the three growth stages, though coefficient is positive for sowing stage and flowering stage and negative for harvesting stage. It means that increasing temperature
during flowering and sowing stage raised the level of risk while rise in temperature
during harvesting stage caused decline in wheat yield risk. The optimum temperature during sowing stage is 12–25 °C; increase in temperature during flowering
stage leads to shorten length of growing season and instigates decline in wheat
yield. Increasing temperature has positive impact on mean yield of wheat crop
during harvesting stage in Pakistan, as Ashfaq et al. (2011) found that high temperature during harvesting stage helps in the formation of healthy seed. On the other
hand, precipitation during sowing and flowering stage showed a positive and significant effect on wheat yield risk. Similarly, significant increasing risk due to
Climate Change, Risk and Food Security: An Analysis …
53
First, to satisfy stationary condition we performed unit root test for time series data.
Both variations in precipitation and temperature were found stationary at level, I(0),
while all other climatic and non-climatic variables became stationary at first difference, I(1). The results of unit root test are presented in Appendix 3. Second,
Wald test was employed to assess about model significance as suggested by
Kohansla and Aliabadi (2014) and Ahmad et al. (2014). Wald test results are
presented in Appendix 4; however, insignificant P value rejected the null hypothesis and significant intercept was found. Hence, the model used in this study was
verified as significant. Third, as stochastic frontier analysis and production risk both
capture the variation in output, we tested the heteroscedasticity as suggested by
Wood and Mendelsohn (2015). The Breusch–Pagan test was applied and is presented in Appendix 5. Null hypothesis for Breusch–Pagan was that the model holds
homoscedasticity or constant variance and is rejected due to insignificant P value.
The test revealed the presence of heteroscedasticity (Production Risk). Finally,
two-step stochastic production function was estimated and the estimates of risk in
wheat crop production are presented in Table 4.
The estimation results for risk in yield revealed that the relationship between risk
and time (technology) was significant and negative sign showed a decrease in risk
due to improvement in technology every year, as improved seed varieties and
advance cultivation technology are encouraging the wheat production and declining
the level of risk. Similar significant and risk-decreasing trend by technology was
reported by Holst et al. (2011) for wheat production in China. Area under cultivation and amount of fertilizer applied were not significant but the positive coefficient exposed increasing risk by increasing both area and fertilizer application.
This may be because of limited management skills and inadequate
information/knowledge of balance use of fertilizer to the farmers in the study area,
as the same was observed by Shakoor et al. (2015) for rice crop, and Zulfiqar and
Ashfaq (2014) for wheat in Pakistan. Holst et al. (2011) reported decreasing but
non-significant risk due to increase in area in China. Conversely, Poudel and Kotani
(2013) reported a significant decreasing trend in risk for wheat in Nepal.
Moreover, no significant production risk was found due to temperature during all
the three growth stages, though coefficient is positive for sowing stage and flowering stage and negative for harvesting stage. It means that increasing temperature
during flowering and sowing stage raised the level of risk while rise in temperature
during harvesting stage caused decline in wheat yield risk. The optimum temperature during sowing stage is 12–25 °C; increase in temperature during flowering
stage leads to shorten length of growing season and instigates decline in wheat
yield. Increasing temperature has positive impact on mean yield of wheat crop
during harvesting stage in Pakistan, as Ashfaq et al. (2011) found that high temperature during harvesting stage helps in the formation of healthy seed. On the other
hand, precipitation during sowing and flowering stage showed a positive and significant effect on wheat yield risk. Similarly, significant increasing risk due to
Climate Change, Risk and Food Security: An Analysis …
53