The relationship between carbon dioxide and agriculture in Ghana: a comparison of VECM and ARDL model

In this paper, the relationship between carbon dioxide and agriculture in Ghana was investigated by comparing a Vector Error Correction Model (VECM) and Autoregressive Distributed Lag (ARDL) Model. Ten study variables spanning from 1961 to 2012 were employed from the Food Agricultural Organization. Results from the study show that carbon dioxide emissions affect the percentage annual change of agricultural area, coarse grain production, cocoa bean production, fruit production, vegetable production, and the total livestock per hectare of the agricultural area. The vector error correction model and the autoregressive distributed lag model show evidence of a causal relationship between carbon dioxide emissions and agriculture; however, the relationship decreases periodically which may die over-time. All the endogenous variables except total primary vegetable production lead to carbon dioxide emissions, which may be due to poor agricultural practices to meet the growing food demand in Ghana. The autoregressive distributed lag bounds test shows evidence of a long-run equilibrium relationship between the percentage annual change of agricultural area, cocoa bean production, total livestock per hectare of agricultural area, total pulses production, total primary vegetable production, and carbon dioxide emissions. It is important to end hunger and ensure people have access to safe and nutritious food, especially the poor, orphans, pregnant women, and children under-5 years in order to reduce maternal and infant mortalities. Nevertheless, it is also important that the Government of Ghana institutes agricultural policies that focus on promoting a sustainable agriculture using environmental friendly agricultural practices. The study recommends an integration of climate change measures into Ghana’s national strategies, policies and planning in order to strengthen the country’s effort to achieving a sustainable environment.

to meet the rapidly growing 125 population poses threat to agricultural and environmental sus-126 tainability. Agriculture has been identified as one of the main 127 sources of greenhouse gas emissions (GHG) (Burney et al. 128 2010) due to unsustainable agricultural practices in order to 129 boost productivity, which leads to food security. 130 Agriculture is one of the major drivers in Ghana's growing Agriculture). With regards to industrial crop production, co-145 coa production occupies the largest area of about 1,600, 146 000 ha, followed by oil palm production of about 360, 147 000 ha, tomato production of about 50,000 ha, seed cotton 148 production of about 20,000 ha, other vegetables account for 149 20,000 ha, pineapple accounts for 328,000 ha, and other (co-150 conut, banana, kola, rubber, tobacco, etc.). Production account 151 for 2,000,000 ha, summing up to 4,060,000 ha (Ministry of 152 Food and Agriculture).

153
In the same vein, livestock production has been increasing 154 from 1999 to 2010. Cattle production rose from 1,288,000 155 heads to 1,454,000 heads, sheep production rose from 2,658, 156 000 heads to 3,779,000 heads, goat production rose from 2, 157 931,000 heads to 4,855,000 heads, pig production rose from 158 332,000 heads to 536,000 heads, and poultry production rose 159 from 18,810,000 birds to 43,320,000 birds, respectively. preciably due to increasing food production to meet a growing 164 Ghanaian population to achieve food security. To the best of 165 our knowledge, the causal nexus between food production and 166 carbon dioxide emissions are yet to be assessed.

167
The remainder of the study is sectioned into Literature  The final publication is available at Springer via http://dx.doi.org/10.1007/s11356-016-6252-x software. Their study concluded that the cultivation of saffron 196 emits 2325.5 kg CO 2 eq. ha −1 greenhouse gas emissions.   Therefore, a multidisciplinary approach that tackles the core is-   The empirical specifications for the model can be quantified as: 313 314

328
The corresponding VEC model can be expressed as: 329 330 332 333

335
Where y t = α 0 + α 1 x t is the long-run cointegrating relation 336 existing between two variables of interest and, λ y and λ x are 337 the error correction parameters measuring the reaction of y and 338 x towards the deviations from long-run equilibrium.

429
Perron-Fisher Chi-square methods (Choi 2001;Hadri 2000;430 Maddala and Wu 1999) assumes a common and individual 431 unit root process as a null hypothesis of a unit root.

432
However, we fail to accept the null hypothesis at level based 433 on 5 % p value. The model proposed is stationary at level, but 434 non-stationary at first differences ( (HQ) select 4 as the optimal lag as indicated by "*" in Table 3.

446
Results and discussion

447
In this section, empirical evidence and discussions of the ma-448 jor findings are outlined.

449
Cointegration test and vector error correction model 450 Using the optimal lag selected by information criteria tests, 451 Johansen's method of cointegration is estimated.  Another evidence from the analysis shows that CO 2 emis-480 sions Granger cause CHAL, COAGRAPROD, COCOBE, 481 FRUPROD, LIVEHEC, and VEGPROD (see Table 6).

ARDL cointegration test, long-run and model selection 483
In this section, the study estimates the ARDL model With the existence of cointegration, the next step is 504 to select the optimal model for the long-run equilibrium 505 relationship estimation. In Fig. 2, the Akaike informa-506 tion criterion model selection is given. Akaike informa-507 tion criterion was used to select the best model with the 508 specification: ARDL (4,4,3,4,4,4,4,4,4 In the same vein, ARDL was subjected to several diagnos-533 tic tests as presented in Table 9. Results from the test shows 534 that the null hypothesis cannot be rejected at the 5 % signifi-    Table 5 Long-run multivariate causality of the error correction model   hanced international cooperation will help reduce the use t10:1