Agriculture significantly contributes to Ethiopia's GDP, employment, and export income. Pulse crops, including grass pea (*Lathyrus sativus*), are crucial dietary components. Grass pea cultivation in Ethiopia is notable, covering about 1% of total cultivated land or 9% of pulse crop land, with regional variations. While globally significant (estimated 1.5 million ha annually), grass pea overconsumption causes lathyrism, a debilitating neurological disorder. Lathyrism prevalence is high in the Amhara region, linked to food shortages from drought. Despite its health risks, grass pea's tolerance to harsh conditions makes it popular among farmers. This study focuses on understanding the factors influencing land allocation to grass pea production, despite its health consequences, to inform policies aimed at reducing dependence on the crop and minimizing lathyrism.

Farmers' land allocation decisions are influenced by socioeconomic factors (marketing, policies), crop physiology (drought resistance), and access to inputs, extension, and credit services. Studies have shown that household size, age, dependency, education, income, landholding size, credit access, and extension services influence land use decisions. However, limited empirical research in Ethiopia focuses on the specific determinants of grass pea land allocation. This study aims to fill this gap by investigating these determinants and their influence on the intensity of land allocated to grass pea production in the Amhara region.

The study was conducted in Delanta and Dawnt districts of the Amhara region, Ethiopia, using a mixed-methods approach. Quantitative data were collected through a semi-structured questionnaire survey of 384 farm households selected using a multi-stage sampling technique targeting kebeles (lower administrative units) within the dega agro-climatic zone known for grass pea production. Qualitative data were gathered through six focus group discussions (FGDs) and 13 key informant interviews (KIIs) with various stakeholders. The Heckman two-stage selection model was used to analyze quantitative data, addressing the selectivity bias inherent in the land allocation decision (a two-stage process: first deciding whether to grow grass pea, then determining land allocation). The probit model was used to estimate the selection equation, and an OLS regression was used to estimate the outcome equation. Qualitative data were analyzed thematically to complement quantitative findings. Household wealth was estimated based on asset ownership and food security.

Farmers allocated approximately 26% of their farmland to grass pea production. The Heckman model revealed that household landholding size (positive and significant at p<0.1), age of the household head (positive and significant at p<0.05), and the head's primary school attendance (positive and significant at p<0.1) had positive effects on land allocated to grass pea. Access to farm extension services had a significant negative effect (p<0.1) on land allocation, likely due to increased awareness of health risks and crop diversification. Household wealth was positively associated with the decision to grow grass pea (p<0.01), but not the intensity of land use. Distance to market centers negatively influenced the decision to grow grass pea (p<0.01). Household size was positively associated with the probability of growing grass pea (p<0.01). The descriptive analysis revealed that many households grow multiple crops (wheat, teff, lentils, chickpea, etc.) due to cultural practices and risk mitigation strategies. Access to farmland through sharecropping arrangements influenced land allocation to grass pea, particularly for households with limited labor and resources.

The findings highlight the complex interplay of socioeconomic and environmental factors driving land allocation decisions for grass pea. The positive influence of landholding size and age aligns with existing literature, while the negative effect of extension services underscores the importance of awareness campaigns on lathyrism prevention. The positive association of primary education with grass pea cultivation suggests a potential need for more nuanced educational interventions to target the specific market-oriented aspects of grass pea while also emphasizing health risks. The results suggest that policies promoting diversified crop production and market access could reduce grass pea dependence. Improving access to credit and promoting alternative high-yielding and drought-resistant crops could provide farmers with better options, thereby mitigating the risks associated with grass pea production.

This study demonstrates that socioeconomic factors significantly influence land allocation for grass pea production in the Amhara region. Policies to improve market access, education, credit availability, extension services, and family planning are crucial for reducing grass pea dependence and mitigating lathyrism. Future research could explore the dynamics of sharecropping arrangements in relation to grass pea production and investigate the long-term impacts of interventions aimed at promoting crop diversification and alternative livelihood strategies.

The study's limitations include its cross-sectional nature, limiting causal inferences, and its focus on two districts, potentially limiting generalizability. Future research should incorporate longitudinal data and expand geographically to enhance the findings' robustness and relevance to other regions.