Table 3
LnCP
|
LnCP
|
LnCO2
|
LnTEM
|
LnLF
|
LnFE
|
LnWA
|
Mean
|
4.385
|
0.192
|
3.023
|
17.689
|
18.485
|
4.895
|
Median
|
4.391
|
0.167
|
3.022
|
17.703
|
18.672
|
4.898
|
Maximum
|
4.670
|
0.116
|
3.086
|
18.118
|
19.519
|
4.933
|
Minimum
|
3.995
|
0.484
|
2.965
|
17.253
|
17.364
|
4.847
|
Std. Dev.
|
0.200
|
0.168
|
0.029
|
0.275
|
0.642
|
0.025
|
Skewness
|
0.359
|
0.012
|
0.355
|
0.033
|
0.338
|
0.449
|
Kurtosis
|
2.066
|
1.926
|
2.970
|
1.721
|
1.873
|
2.231
|
Jarque-Bera
|
1.737
|
1.442
|
0.632
|
2.048
|
2.159
|
1.746
|
Probability
|
0.419
|
0.486
|
0.728
|
0.359
|
0.339
|
0.417
|
Sum
|
131.555
|
-5.770
|
90.69
|
530.670
|
554.559
|
146.861
|
Sum Sq. Dev.
|
1.1623
|
0.820
|
0.024
|
2.200
|
11.954
|
0.018
|
Observations
|
30
|
30
|
30
|
30
|
30
|
30
|
Sources: Calculated by the Authors |
Descriptive statistics
Table 3 shows the initial summary statistics for the variables that are important for describing the properties of the raw data that is derived from the descriptive statistics. Climate variables have a mean (3.023) among them, with temperature having the greatest mean (3.023), followed by carbon dioxide (CO2) emissions having the lowest mean (0.192). The maximum value (3.086) and the minimum value (0.484) of the temperature and the carbon dioxide emissions (CO2) were found to be the highest and lowest, respectively. Carbon dioxide (CO2) emissions have the highest standard deviation (0.168) compared to temperature variables. In addition, the Jarque-Bera test suggests that the data are consistently distributed and identical and that the distribution is regular.
Table 4
Variables
|
LnCP
|
LnCO2
|
LnTEM
|
LnLF
|
LnFE
|
LnWA
|
LnCP
|
1.000
|
|
|
|
|
|
|
-----
|
|
|
|
|
|
|
-----
|
|
|
|
|
|
LnCO2
|
0.960
|
1.000
|
|
|
|
|
|
18.364
|
-----
|
|
|
|
|
|
[0.000]
|
-----
|
|
|
|
|
LnTEM
|
0.675
|
0.736
|
1.000
|
|
|
|
|
4.851
|
5.763
|
-----
|
|
|
|
|
[0.000]
|
[0.000]
|
-----
|
|
|
|
LnLF
|
0.966
|
0.969
|
0.711
|
1.000
|
|
|
|
20.069
|
20.990
|
5.358
|
-----
|
|
|
|
[0.000]
|
[0.000]
|
[0.000]
|
-----
|
|
|
LnFE
|
0.937
|
0.933
|
0.612
|
0.932
|
1.000
|
|
|
14.225
|
13.729
|
4.103
|
13.700
|
-----
|
|
|
[0.000]
|
[0.000]
|
[0.000]
|
[0.000]
|
-----
|
|
LnWA
|
0.537
|
0.440
|
0.165
|
0.494
|
0.673
|
1.000
|
|
3.373
|
2.596
|
0.888
|
3.011
|
4.822
|
-----
|
|
[0.002]
|
[0.014]
|
[0.381]
|
[0.005]
|
[0.000]
|
-----
|
Sources: Calculated by the Authors |
Correlation Analysis
Table 4 shows the correlation matrix in the same way. According to the correlation analysis, carbon dioxide (CO2) emissions, average temperature, agriculture labor, fertilizer use, and water availability are all positively connected with crop yield. This study also discovered a link between carbon dioxide (CO2) emissions and water availability, fertilizer consumption, and agricultural labor. Furthermore, the correlation asserts that the regressors are not multicollinear.
Table 5
Unit Root Test (Augmented Dickey–Fuller)
Variables
|
ADF (Akaike Info Criterion)
|
|
PP (Philips–Perron)
|
|
|
Level
|
|
Level
|
|
|
Intercept
|
Intercept and trend
|
Intercept
|
Intercept and trend
|
LnCP
|
-1.058
[0.718]
|
-3.746
[0.034]
|
-1.078
[0.710]
|
-3.620
[0.045]
|
LnCO2
|
0.552
[0.985]
|
-6.598
[0.000]
|
0.606
[0.987]
|
-2.118
[0.514]
|
LnTEM
|
-1.700
[0.420]
|
-3.436
[0.066]
|
-1.495
[0.521]
|
-3.426
[0.067]
|
LnLF
|
3.713
[1.000]
|
-2.462
[0.342]
|
3.686
[1.000]
|
-2.667
[0.256]
|
LnFE
|
-0.798
[0.803]
|
-5.010
[0.001]
|
-1.888
[0.332]
|
-4.943
[0.002]
|
LnWA
|
-2.102
[0.245]
|
-1.183
[0.895]
|
-1.923
[0.317]
|
-0.984
[0.930]
|
|
1ST difference
|
|
1ST difference
|
|
|
Intercept
|
Intercept and trend
|
Intercept
|
Intercept and trend
|
LnCP
|
-7.387
[0.000]
|
-7.325
[0.000]
|
-11.611
[0.000]
|
-12.07
[0.000]
|
LnCO2
|
-3.755
[0.010]
|
-3.625
[0.050]
|
-5.438
[0.000]
|
-5.421
[0.000]
|
LnTEM
|
-7.779
[0.000]
|
-7.734
[0.000]
|
-8.030
[0.000]
|
-8.153
[0.000]
|
LnLF
|
-3.717
[0.009]
|
-5.183
[0.001]
|
-3.702
[0.009]
|
-5.182
[0.001]
|
LnFE
|
-7.786
[0.000]
|
-7.544
[0.000]
|
-9.160
[0.000]
|
-9.055
[0.000]
|
LnWA
|
-4.725
[0.000]
|
-6.000
[0.000]
|
-5.180
[0.000]
|
-6.380
[0.000]
|
Sources: Calculated by the Authors |
Unit Root Test
The variables in this study were tested for stationarity using the ADF test, and the estimated findings of the ADF test were then confirmed using the PP and KPSS unit root tests. According to Table 5's ADF unit root test results, crop production (lnCP), CO2 emissions (lnCO2), temperature (lnTEM), land forest (lnLF), water availability (lnWA), and fertilizer consumption (lnFE) have stabilised at the I(0) level. However, lnCO2 is the first-order I(1) combined with the intercept and trend, whereas CO2 emissions will be the first I(1) separately. Using the ARDL co-integration method, the impact of climate change on agricultural output in Pakistan from 1990 to 2019 was examined.
Table 6
Optimal lag length results
Lag
|
LogL
|
LR
|
FPE
|
AIC
|
SC
|
HQ
|
0
|
249.964
|
NA
|
5.71e-16
|
-18.071
|
-17.783
|
-17.985
|
1
|
382.854
|
196.873*
|
4.69e-19*
|
-25.248
|
-23.232*
|
-24.649
|
2
|
415.753
|
34.117
|
9.25e-19
|
-25.018
|
-21.275
|
-23.905
|
3
|
469.254
|
31.703
|
1.17e-18
|
-26.315*
|
-20.843
|
-24.688*
|
* Shows lag order selected by the criterion |
Sources: Calculated by the Authors |
Selection Of Lag Order Vector Error Correction Model
We used the ARDL contour method to see if there was co-integration between the significant variables in this study, such as crop production (lnCP), emissions of carbon dioxide (lnCO2), average temperature (lnTEM), agricultural labor force (lnLF), fertilizer consumption (lnFE), and water availability (lnWA), due to the unique order of integration of the time series properties of the selected study variables. The next stage in this study was to select the proper lag order for the variables to apply the ARDL approach after assessing the integration of the series. The ideal delay order is determined by LR (Sequence Corrected LR Test Statistics), FPE (Final Prediction Error), AIC, SBC, and HQC. Table 6's forecast results demonstrate that the majority of criteria have the delay order set to 1.
Furthermore, in Fig. 4, the validation of identifying the appropriate lag length under the VAR method is determined by showing a polynomial plot; all the blue dots in this plot are within circles, indicating that estimation at lag 1 provides satisfactory results.
LR (Sequence Corrected LR Test Statistics), FPE (Final Prediction Error), AIC, SBC, and HQC determine the appropriate lag order. The prediction results in Table 6 show that most criteria set the lag order to 1.
Furthermore, in Fig. 4, the validation of identifying the appropriate lag length under the VAR method is determined by showing a polynomial plot; all the blue dots in this plot are within circles, indicating that estimation at lag 1 provides satisfactory results.
Table 7
The ARDL bounds tests to co-integration Sources: Calculated by the Authors
Function
|
F-statistic
|
k
|
FLNCP(LNCP│LNCO2,LNTEM,LNLF,LNFE,LNWA)
|
6.181
|
5
|
C-value bounds
|
|
|
Significant Level
|
I(0) bounds
|
I(1) bounds
|
10%
|
2.08
|
3
|
5%
|
2.39
|
3.38
|
2.5%
|
2.7
|
3.73
|
1%
|
3.06
|
4.15
|
The Ardl Bounds Tests Of Co-integration
As a result of identifying the unit root of the model and having selected the best lag model, we can regress Eq. (2) to determine whether the variables are co-integrated over the long run by finding Wald F-statistics. Based on the Wald F-statistics and the critical value, presented in Table 7, it is important to note that the crucial values proposed by (Pesaran et al., 2001), cannot be used when there is a small sample size. We used the crucial values to calculate the cointegration with just 30 observations in the study due to the very small sample size. In order to assess the connection between the variables chosen for this research, and in light of the results of the estimation of unit root testing, which demonstrate that all variables are integrated at I(1), we now used the ARDL method of co-integration (bounds testing) to estimate the relationship between the variables. A summary of the results of the ARDL bound testing is shown in Table 7. The results show that, at the 1 percent significance level for the I(0) bound, the f-statistic value (6.181) exceeds the threshold of 4.15, based on the I(0) bound. Based on the findings of the limits test, it is evident that the null hypothesis that there is no cointegration correlation among LCP, LCO2, LTEM, LLF, LWA, and LFE has been rejected. It also validates that there are substantial interactions between variables in the long run.
Table 8
Short-run and long-run estimates
Panel A: short-run results
|
|
|
|
|
Variable
|
Coefficient
|
Std. Error
|
t-Stat
|
P-value.
|
Δ LnCO2
|
0.432**
|
0.160
|
2.692
|
0.014
|
Δ LnTEM
|
0.520**
|
0.257
|
2.025
|
0.057
|
Δ LnLF
|
0.636**
|
0.165
|
2.618
|
0.067
|
Δ LnFC
|
-0.059**
|
0.030
|
-1.947
|
0.067
|
Δ LnWA
|
0.803
|
0.431
|
1.859
|
0.579
|
CointEq(-1)*
|
-0.791***
|
0.104
|
-7.607
|
0.000
|
Panel B: long-run results
|
Variable
|
Coefficient
|
Std. Error
|
t-Stat
|
P-value
|
LnCO2
|
1.828***
|
0.518
|
3.524
|
0.002
|
LnTEM
|
-0.680
|
0.609
|
-1.115
|
0.279
|
LnLF
|
0.207**
|
0.144
|
2.436
|
0.068
|
LnFC
|
-0.320**
|
0.141
|
-2.259
|
0.036
|
LnWA
|
3.211***
|
1.093
|
2.937
|
0.008
|
C
|
-6.660
|
4.034
|
-1.650
|
0.116
|
Panel C: diagnostics tests results
|
|
|
|
|
R-squared (0.979)
|
M-D var (4.398)
|
|
|
|
Adjusted R-squared (0.967)
|
S-D-D var
|
|
|
|
S.E. of regression (0.034)
|
A.I.C (-3.632)
|
|
|
|
Sum squared resid (0.021)
|
S.C (-3.114)
|
|
|
|
Log likelihood (63.673)
|
H.Q.C (-3.470)
|
|
|
|
F-statistic (84.269)***
|
D-W stat (2.610)
|
|
|
|
Prob(F-statistic) (0.000)
|
|
|
|
|
Note: ***Statistical significance at the 1% level. **Statistical significance at the 5% level. *Statistical significance at the 10% level.
Sources Calculated by the Authors
The NARDL analysis
We attempted to study both short- and long-term connections following the verification of long-term connections using the ARDL approach. The findings from the long- and short-term dynamics are summarized in Table 8.
Short-run relationships were also discovered in this empirical study. The annual CO2 emission value in Table 8 (panal A) is (0.432), indicating an inverse relationship with crop productivity, implying that an increase in yearly CO2 emissions reduces crop yield in the short run. Based on ARDL, Rehman et al., (2022b) published a study showing the impact of crop production, agricultural land use, and fertilizer usage on carbon dioxide (CO2) emissions in Nepal 1965–2018. On the other hand, it has been shown that a decrease in the amount of land used for crop production will result in a higher amount of carbon dioxide (CO2) being emitted in the long run as well as in the short run.
Furthermore, the yearly average temperature has a positive and considerable short-run value. The justification could be that temperature has a substantial influence on crop productivity, implying that higher temperatures will impair the physiological processes required for crop growth and development, and that crop yields will most likely decline from current levels. Besides (Li et al., 2019), investigated time series data from 1981 to 2016 in the United States and discovered that high temperatures in warm states could mitigate such negative impacts and make excessive rainfall helpful to crops by meeting their water needs and reducing heat stress. We only noticed a sliver of a bigger yield decline trend.
Furthermore, the variable labor force has a statistically significant positive short-run coefficient (0.520) with a probability value of 0. (0.0579). As a result, labor and crop production have a short-run co-integration. The findings of this study are supported by (Warsame et al., 2021; Pickson et al., 2020) studies according to (Sui et al., 2022). Table 8 expected consequences (panal A) also show that fertilizer use has a negative but substantial impact on crop output at a 5% significance level. By increasing fertilizer by 1%, it is expected that crop productivity will increase by 0.05%. Adding fertilizer to the soil is considered one of the most important inputs to increase crop production output, since it improves the soil fertility and, in turn, increases the productivity of plants. As a matter of fact, the majority of farmers in Pakistan's rural areas use a combination of chemical and natural fertilizers with the aim of improving soil fertility. A significant role has been played by chemical fertilizers in improving crop production in Pakistan over the last few decades. In addition to this empirical conclusion, there are a number of studies on fertilizer consumption to boost agricultural production and sector development in order to feed the world's rapidly growing population. (Wakeel et al., 2022; Ramzan et al., 2022). To achieve long-term agricultural output, agricultural factor productivity should be considered in a context in which the antecedents of each factor are analyzed in order to understand their effects on the efficiency and effectiveness of the use of inputs.If current management methods continue, N losses may increase as a result of the excessive use of fertilizers to produce more food. Similarly, the current findings are comparable to those of (Huang et al., 2022), where the coefficient of water availability is non-significant (0.803) in the short run, indicating a negative association between crop productivity and water availability. The argument could be that Pakistan's infrastructure is inadequate, and hence the short-run relationship between water availability and crop productivity is insignificant.
Furthermore, the results of our study demonstrate that carbon dioxide (CO2) emissions have a positive and significant impact on agricultural output in Pakistan as measured by the long-run correlation coefficient and the probability value estimated. At a 1% level of significance, a 1% increase in CO2 emissions will eventually result in a 1.828% increase in crop yield. In The long-run coefficient of CO2 emissions, Pickson (2020), and colleagues came to the same conclusion as we did. In the long run, it has been discovered that temperature has no substantial impact on crop yield. It is possible that a minor increase in temperature will have an insignificant impact on agricultural output in Pakistan over the long-term. In contrast to tropical places where the temperature has already reached a threshold level, Pakistan is positioned next to the tropics, where a minor rise in temperature caused by climate fluctuation may not have a significant influence on agricultural productivity. The findings of the study are corroborated by a number of earlier empirical investigations (Alleyne and Jones. 2022; Rit, 2022). Our findings are analogous to those of (Ogundari and Onyeaghala 2021), who used time series data from 1981 to 2010 to assess the effects of climate change in 35 countries. According to the findings, the temperature has no major impact on agricultural productivity over the long run. When it comes to labor force, the coefficient is positive and statistically significant at the 10% level, implying that as the labor force grows, so does agricultural output, as shown in the graph. Such findings are consistent with the findings of Abbas (2022), who found that the labor force has a productive effect on the agricultural cereal food crops in Pakistan. This means that an increase in the labor force would result in an increase in crop production of 0.207% for every 1% increase in the labor force. More importantly (as indicated in Table 8 (panal B), the predicted long-term fertilizer consumption coefficient is statistically significant but negative (-0.320), with a probability value of (0.036). According to the findings of the study, the usage of fertilizer and the production of major food crops have a positive and negative relationship that is statistically significant at the % confidence level. The increase in crop production output will be around 10% higher if fertilizer use is increased by only 1%, which is comparable to an increase in crop production output of approximately 10%. When the proper amount of fertilizer is applied, this coefficient states that the crop output yields will increase. Earlier research conducted by (Ali et al., 2022), found that essential inputs in agriculture, such as improved seeds and fertilizer, can boost crop productivity and revenue in Pakistan. The current study's findings are validated by earlier research. As indicated by the predicted long-run coefficient of water availability in Table 8 (panal B), the probability value of water availability is also positive (3.211), as indicated by the probability value of water availability in Table 8. (0.008). As a result, the long-term relationship between water availability and crop productivity is beneficial to crop yields. The findings of this study are consistent with those of (Lamptey 2022; Khan and Rehman 2022).
Table 9
Results of the Johansen co-integration test outcomes
H-No. of CE(s)
|
Eigenvalue
|
Trace Statistic
|
0.05 Critical Value
|
Prob.**
|
Rank Test (Trace)
|
r ≤ 0
r ≤ 1
r ≤ 2
r ≤ 3
r ≤ 4
r ≤ 5
|
0.887
0.769
0.528
0.373
0.287
0.000
|
140.877
81.772
42.109
21.801
9.157
0.011
|
95.753
69.818
47.856
29.797
15.494
3.841
|
0.000a
0.004a
0.155
0.309
0.351
0.916
|
Rank Test (Maximum Eigenvalue)
|
r ≤ 0
r ≤ 1
r ≤ 2
r ≤ 3
r ≤ 4
r ≤ 5
|
0.887
0.769
0.528
0.373
0.287
0.000
|
59.104
39.663
20.307
12.644
9.146
0.011
|
40.077
33.876
27.584
21.131
14.264
3.841
|
0.000a
0.009a
0.320
0.485
0.274
0.916
|
Note: a indicates rejection of the hypothesis at the 0.05 significance level.
Sources Calculated by the Authors
Johansen Co-integration Test
Using the Johansen and Juselius (1990), co-integration approach, this study investigated the long-run linkage between the studied variables. Based on the robustness estimated outcomes of this approach, Table 8 shows that there is a long-term co-integration relationship between crop production, carbon dioxide (CO2) emissions, average temperature, agricultural labor, fertilizer use, and water availability. It means that the long-term benefits are quite effective and durable. As a result, the results of the co-integration test utilizing the Johansen–Juselius technique guarantee that the variables under consideration are linked throughout the long run.
Table 10
Result of pairwise Granger Causality Tests
Dependent
|
Independent
|
F-Statistic
|
Prob.
|
LUCO2
|
≠ >
|
LUCP
|
6.557
|
0.016*
|
LUCP
|
≠ >
|
LUCO2
|
1.611
|
0.215
|
LUTEM
|
≠ >
|
LUCP
|
0.371
|
0.547
|
LUCP
|
≠ >
|
LUTEM
|
8.221
|
0.008**
|
LULF
|
≠ >
|
LUCP
|
6.633
|
0.016*
|
LUCP
|
≠ >
|
LULF
|
5.057
|
0.033*
|
LUFE
|
≠ >
|
LUCP
|
2.874
|
0.102
|
LUCP
|
≠ >
|
LUFE
|
9.361
|
0.005**
|
LUWA
|
≠ >
|
LUCP
|
2.043
|
0.164
|
LUCP
|
≠ >
|
LUWA
|
0.026
|
0.871
|
LUTEM
|
≠ >
|
LUCO2
|
0.918
|
0.346
|
LUCO2
|
≠ >
|
LUTEM
|
5.861
|
0.022*
|
LULF
|
≠ >
|
LUCO2
|
4.573
|
0.042*
|
LUCO2
|
≠ >
|
LULF
|
0.008
|
0.929
|
LUFE
|
≠ >
|
LUCO2
|
0.473
|
0.497
|
LUCO2
|
≠ >
|
LUFE
|
2.105
|
0.158
|
LUWA
|
≠ >
|
LUCO2
|
0.009
|
0.924
|
LUCO2
|
≠ >
|
LUWA
|
0.729
|
0.400
|
LULF
|
≠ >
|
LUTEM
|
9.018
|
0.005**
|
LUTEM
|
≠ >
|
LULF
|
0.355
|
0.555
|
LUFE
|
≠ >
|
LUTEM
|
5.982
|
0.021*
|
LUTEM
|
≠ >
|
LUFE
|
0.015
|
0.901
|
LUWA
|
≠ >
|
LUTEM
|
0.551
|
0.464
|
LUTEM
|
≠ >
|
LUWA
|
2.323
|
0.139
|
LUFE
|
≠ >
|
LULF
|
1.442
|
0.240
|
LULF
|
≠ >
|
LUFE
|
3.838
|
0.060*
|
LUWA
|
≠ >
|
LULF
|
2.493
|
0.126
|
LULF
|
≠ >
|
LUWA
|
0.326
|
0.572
|
LUWA
|
≠ >
|
LUFE
|
0.401
|
0.531
|
LUFE
|
≠ >
|
LUWA
|
0.176
|
0.677
|
*, **, and *** represents significance level at 1%, 5%, and 10% respectively |
Sources: Calculated by the Authors |
Granger Causality Test Outcome
For the purpose of this study, the paired Granger causality tests were utilized to determine whether or not the variables were causally associated through direct causation between them. It was found that CO2 was strongly related to crop production, annual average temperature was correlated with crop production, labor force was correlated with crop production, fertilizer consumption was correlated with crop production, and water availability was correlated with crop production in the study. A table illustrating the findings of the paired Granger causality test can be found in Table 10. The null hypothesis that CO2 does not act to cause crop production declines is ruled out when it is evaluated on a 10% level. According to the findings of the study, there is a one-way causal relationship between LnCO2 and LnCP. Similarly, the null hypothesis that agricultural output consumption has no effect on temperature rise is rejected at a significance level of 5%. According to the results of this study, there is a causal relationship between LnCP and LnTEM that is one-way. The null hypothesis that labor force does not affect agricultural production reductions is also rejected at a significance level of 5%. The study demonstrates a two-way causal link between LnLF ↔ LnCP. Additionally, it has been found that at a level of significance of 10%, the null hypothesis that crop production consumption does not result in increased fertilizer use is rejected.
There is a one-way causal relationship between LnCP → LnLF, and the null hypothesis that labor force does not affect fertilizer consumption is rejected at the 5% significance level. Evidence supports a one-way causal relationship between LnCP → LnFE. According to the results of this study, the null hypothesis, that the average annual rainfall of major food crops alone does not influence the yield, is supported. There is a significant difference between the null hypothesis that there is no effect of CO2 consumption on temperature at a 10% significance level. Based on the evidence, there is a one-way causal relationship between LnCO2 → LnTEM. The null hypothesis that labor force doesn't cause CO2 is rejected at a 10% level of significance.
The evidence shows that the one-way causality running from LnLF → LnCO2 is rejected at a 10% significance level, indicating that labor force is not responsible for temperature. It is supported by the data that there is a one-way causal relationship between LnLF and LnTEM. Additionally, at a 10% significance level, the null hypothesis of fertilizer consumption not having any effect on temperature is rejected. Evidence supports a one-way causal relationship between LnFE and LnTEM.
Table 11
Outcomes of robust analysis
FMOLS
|
CCR
|
Robustness least squares
|
Variable
|
Coefficient
|
t-Statistic
|
Variable
|
Coefficient
|
t-Statistic
|
Variable
|
Coefficient
|
z-Statistic
|
LnCO2
|
0.692
|
3.002***
|
LNCO2
|
0.932
|
2.149***
|
LnCO2
|
0.589
|
2.752****
|
LnTEM
|
-1.347
|
-4.928***
|
LNTEM
|
-0.370
|
-4.823***
|
LnTEM
|
-0.300
|
-4.553***
|
LnLF
|
0.307
|
2.795*
|
LNLF
|
0.308
|
2.662*
|
LnLF
|
0.354
|
2.190*
|
LnFC
|
-0.018
|
-0.348
|
LNFC
|
-0.089
|
-0.759
|
LnFC
|
-0.001
|
-0.019
|
LnWA
|
0.795
|
1.457
|
LNWA
|
1.258
|
1.374
|
LnWA
|
0.732
|
0.934
|
C
|
-3.416
|
-1.227
|
C
|
-4.282
|
-1.331
|
C
|
-4.422
|
-1.086
|
R2 0.947
Adj R2 0.936
|
R2 0.943
Adj R2 0.930
|
R2 0.804
Adj R2 0.764
|
*** =1%, ** =5% and * =10% significance level |
Sources: Calculated by the Authors |
Robust Analysis
This work makes use of FMOLS, CCR, and resilient least squares regression methods to further validate long-run findings. The ARDL procedure's actual co-integrating and reliability for long-run estimates are evaluated using the FMOLS, CCR, and resilient least squares techniques. Only the temperature and fertilizer use are shown in Table 11 as adversely negative and insignificant factors based on the estimated results from both approaches. Furthermore, there is a positive and statistically significant correlation between the agricultural labor force and carbon dioxide (CO2) emissions.
For similar reasons, we provided the results of long run estimates of FMOLS, DOLS, and robust least squares for each of the three research models separately in Table 11. Overall, the findings indicate that temperature has a negative but statistically significant relationship with crop productivity in Pakistan. Furthermore, CO2 emissions as well as the labor force in Pakistan are positively associated with crop productivity. CO2 is statistically significant at the 1% level of significance. Its positive coefficients for CO2 emissions and labor force imply that a 1% and a 10% increase in crop productivity decrease crop productivity (improve environmental sustainability) by 0.692% (FMOLS), 0.932% (CCR), and 0.589% (FMOLS), respectively, if crop productivity increases by 1% and 10%. (Robustness). Similar to the agricultural labor force, the agricultural labor force is statistically significant at the ten percent level of significance. Positive coefficient values imply a percentage increase in crop output, with 0.307% (FMOLS), 0.308% (DOLS), and 0.354% (DOLS) implying a percentage increase in crop production (robustness).This finding indicates that urbanization boosts crop yields by increasing the size of the agricultural labor force.