How Econometrics Can Help Us Understand the Effects …
25
However, as a long-run concept, weak exogeneity can give different results from
those obtained analysis Granger causality. Ahumada and Cornejo (2019) found that
all variables adjust to deviations from the long-run equilibria. This finding implies
that climate variables are not weak exogenous which indicates that a system approach
should be followed instead of estimating a single-equation model.
Although this result may be unexpected at first sight, it may be properly interpreted
when we take into account the effect of deforestation. The soybeans upward trend
in Argentina, a behavior also shown by other soybean producers such as Brazil
(see Fig. 1), could have given incentives to the expansion of agriculture through
the use of new lands coming from deforestation. Using data from NASA’s Moderate
Resolution Imaging Spectrometer (MODIS) on the Terra and Aqua satellites, Morton
et al. (2006) have shown that in 2003, the peak year of deforestation in Matto Grosso
(Brazilian state with the highest deforestation and soybean production rates) more
than 20% of the state’s forests were converted to cropland.
Deforestation contributes to global climate warming since it is responsible for
not compensating the anthropogenic emissions of carbon dioxide to the atmosphere,
and, in fact, deforestation releases CO 2 to the atmosphere. Therefore, from this
perspective, we could think that global temperature anomalies and CO 2 emissions
may also adjust to deviations from the long run in our estimated climate system.
Measuring Nonlinear Effects
Linearity is usually a starting functional form of many econometric models although
there are several routes to relax this assumption by including polynomial terms, asymmetries, thresholds, etc. Different climate variables may have nonlinear relationships
with crop yields.
Using different spatial panel econometric techniques, Chen et al. (2013) found
nonlinearities and asymmetric relationships between yields and weather variables as
it has been suggested in the literature. It is usually considered that the best predictor
of crop yield is some measure of extreme heat during the growth period of the plant,
considering a temperature threshold above 29 or 30
◦ C (Schlenker and Roberts 2009),
depending on the analyzed crop. Furthermore, extreme high temperatures are harmful
for crop growth, particularly during the phases of the growth cycle (Auffhammer et al.
2012; Welch et al. 2010). Therefore, the effect of temperature could be nonlinear but
with a threshold at certain high levels.
Using daily data of maximum temperature from 54 meteorological stations of the
Argentine soybean production area from 1973 to 2015, Ahumada and Cornejo (2019)
constructed different variables that measure the number of days during the growing
phase of the crop (from December to April) in which the temperature exceeded a
threshold of 28, 29, 30, or 31
◦ C. They evaluated which of these different temperature thresholds has the most significant impact on Argentine soybean yields. The
maximum temperature of each meteorological station was weighted by its share in
the total soybean planted area. Those weights were annually updated to account for
25
However, as a long-run concept, weak exogeneity can give different results from
those obtained analysis Granger causality. Ahumada and Cornejo (2019) found that
all variables adjust to deviations from the long-run equilibria. This finding implies
that climate variables are not weak exogenous which indicates that a system approach
should be followed instead of estimating a single-equation model.
Although this result may be unexpected at first sight, it may be properly interpreted
when we take into account the effect of deforestation. The soybeans upward trend
in Argentina, a behavior also shown by other soybean producers such as Brazil
(see Fig. 1), could have given incentives to the expansion of agriculture through
the use of new lands coming from deforestation. Using data from NASA’s Moderate
Resolution Imaging Spectrometer (MODIS) on the Terra and Aqua satellites, Morton
et al. (2006) have shown that in 2003, the peak year of deforestation in Matto Grosso
(Brazilian state with the highest deforestation and soybean production rates) more
than 20% of the state’s forests were converted to cropland.
Deforestation contributes to global climate warming since it is responsible for
not compensating the anthropogenic emissions of carbon dioxide to the atmosphere,
and, in fact, deforestation releases CO 2 to the atmosphere. Therefore, from this
perspective, we could think that global temperature anomalies and CO 2 emissions
may also adjust to deviations from the long run in our estimated climate system.
Measuring Nonlinear Effects
Linearity is usually a starting functional form of many econometric models although
there are several routes to relax this assumption by including polynomial terms, asymmetries, thresholds, etc. Different climate variables may have nonlinear relationships
with crop yields.
Using different spatial panel econometric techniques, Chen et al. (2013) found
nonlinearities and asymmetric relationships between yields and weather variables as
it has been suggested in the literature. It is usually considered that the best predictor
of crop yield is some measure of extreme heat during the growth period of the plant,
considering a temperature threshold above 29 or 30
◦ C (Schlenker and Roberts 2009),
depending on the analyzed crop. Furthermore, extreme high temperatures are harmful
for crop growth, particularly during the phases of the growth cycle (Auffhammer et al.
2012; Welch et al. 2010). Therefore, the effect of temperature could be nonlinear but
with a threshold at certain high levels.
Using daily data of maximum temperature from 54 meteorological stations of the
Argentine soybean production area from 1973 to 2015, Ahumada and Cornejo (2019)
constructed different variables that measure the number of days during the growing
phase of the crop (from December to April) in which the temperature exceeded a
threshold of 28, 29, 30, or 31
◦ C. They evaluated which of these different temperature thresholds has the most significant impact on Argentine soybean yields. The
maximum temperature of each meteorological station was weighted by its share in
the total soybean planted area. Those weights were annually updated to account for