Climate change, driven by greenhouse gas (GHG) emissions, poses a significant threat to global food security, particularly in developing countries like Ethiopia. Ethiopia's economy is heavily reliant on agriculture, which also happens to be a major source of GHG emissions within the country. While there is an existing body of research on the microeconomic impacts of GHGs on Ethiopian agriculture, there's a notable gap in macro-level studies examining the long-term dynamic interactions between GHG emissions and agricultural productivity. This study addresses this gap by employing a VAR model to analyze the dynamic relationships over the period 1990-2022. The central research question is: Does the relationship between agricultural GHG emissions and agricultural productivity (proxied by grain yield per hectare) exhibit a positive or negative long-term trend?

Existing literature highlights the significant impact of climate change on agricultural productivity globally. Studies show that increased GHG emissions can lead to reduced crop yields and food insecurity, particularly in vulnerable regions like Africa. Research on agricultural productivity and GHG emissions in developing countries suggests that yield increases could mitigate emissions growth, but the relationship is complex and varies based on the specific factors and location. While several studies focus on micro-level impacts of GHGs in Ethiopia, macro-level analyses of long-term dynamic interactions remain scarce. This study aims to bridge this gap by examining the dynamic relationship between agricultural productivity and emissions in Ethiopia over time, incorporating the country's Climate-Resilient Green Economy (CRGE) strategy into the analysis.

This study utilizes a Vector Autoregressive (VAR) model to capture the dynamic relationships between agricultural productivity and various factors, including GHG emissions. Data spanning 1990 to 2022 were sourced from the World Bank's IBRD.IDA database. Key variables include agricultural productivity (proxied by cereal yield per hectare), fertilizer consumption, arable land size, CO2 emissions, methane emissions (AME), nitrous oxide emissions (ANOE), rural population, agricultural employment, and total land allocated for cereal production. Prior to VAR modeling, unit root tests (ADF and PP) were conducted to ensure data stationarity. The AIC and FPE information criteria were used for lag length selection. A Johansen cointegration test determined the presence of long-run relationships. A Vector Error Correction Model (VECM) was then used to analyze the short-term dynamics. Further analyses included variance decomposition and impulse response functions to study the dynamic behavior of the variables and a Granger causality test to determine the direction of causality. Finally, model stability tests (recursive residuals and Ramsey RESET) were conducted.

The unit root tests revealed that all variables were stationary at first difference, except for rural population (stationary at level). Cointegration tests (Johansen's trace and maximum eigenvalue tests) confirmed the presence of a long-run relationship between agricultural productivity and the other variables. The long-run VAR model showed a positive and significant relationship between agricultural productivity and fertilizer consumption (elasticity of 0.28), agricultural land (elasticity of 2.09), nitrous oxide emissions (elasticity of 15.92), rural population (elasticity of 5.33), and land allocated for cereal production (elasticity of 1.31). In contrast, agricultural employment (elasticity of -5.82), methane emissions (elasticity of -17.11), and CO2 emissions (elasticity of -2.75) showed a negative and significant relationship with agricultural productivity. The error correction model revealed a significant negative coefficient (-0.744) for the error correction term, indicating a rapid (74.4% per year) adjustment towards the long-run equilibrium. Variance decomposition analysis showed that agricultural productivity is most significantly influenced by its own innovation, fertilizer consumption, and agricultural land, while CO2 emissions have a more moderate influence. Granger causality tests indicated bidirectional causality between agricultural productivity and fertilizer consumption, CO2 emissions, and methane emissions, and unidirectional causality for the other variables. Model stability tests confirmed the model's validity.

The findings highlight a complex interplay between GHG emissions and agricultural productivity in Ethiopia. While certain factors like fertilizer use and nitrous oxide emissions show a positive relationship with yield in the long run, others like methane and CO2 emissions exhibit a negative association, underscoring the trade-offs involved in agricultural intensification. The rapid convergence to long-run equilibrium as shown by the error correction model suggests the Ethiopian agricultural system is relatively responsive to changes in these factors. The unidirectional/bidirectional Granger causality relationships highlight the importance of considering both short-term and long-term effects of policies aimed at improving agricultural productivity and mitigating GHG emissions. The study's results emphasize the need for integrated strategies that consider both food security and environmental sustainability in Ethiopia's agricultural sector.

This study provides valuable insights into the dynamic relationship between GHG emissions and agricultural productivity in Ethiopia. The findings emphasize the need for balanced policies that promote agricultural intensification while mitigating GHG emissions. Future research could focus on analyzing the impact of specific agricultural practices on GHG emissions, exploring the effectiveness of climate-smart agriculture approaches, and examining regional variations within Ethiopia's diverse agroecological zones. This would allow for more tailored and effective policy interventions.

The study's reliance on aggregate data may mask regional variations in the relationship between GHG emissions and agricultural productivity. The study also focuses mainly on cereal production, which may not fully capture the complexity of Ethiopian agriculture. Future studies might benefit from disaggregated data to analyze regional variations and incorporate other agricultural subsectors, like livestock. The reliance on a specific model (VAR) also limits the potential insights which could be extracted.