Modeling and Predicting Livestock Export in Somaliland: A Comparative Hybrid Model Approach

Livestock exports are a vital component of Somaliland’s economy, significantly contributing to trade, livelihoods, and economic stability. This study uses single and hybrid time series models to investigate forecasting methods for livestock exports, specifically goats and sheep, cattle, and camels. The data utilized in this study were obtained from the Ministry of Livestock and Rural Development of Somaliland, comprising monthly livestock export records from January 2013 to December 2023. The dataset includes exports primarily from Berbera Port, the official and primary gateway for international livestock trade, along with supplementary data from the informal Asha-Ado Port. Data were analysed using the time-series methods, including SARIMA, ETS, Holt-Winter, NNAR, TBATS, and six hybrid models. Descriptive analysis reveals notable differences in export volumes and volatility across livestock categories. Model performance was assessed using RMSE, MAE, MAPE, MASE, and Theil’s U metrics. Results indicate that SARIMA(0,0,0)(2,0,0)12 outperforms other single models for goats and sheep, ETS(A,N,N) is optimal for cattle, and TBATS(0.41, {0,0}, -, { }) is best for camel exports. However, hybrid models demonstrate superior accuracy, particularly SARIMA-ARIMA for goats and sheep, SARIMA-ETS-NNAR for cattle, and SARIMA-ETS-NNAR-TBATS for camels. Despite improvements, wide confidence intervals highlight forecast uncertainty, emphasizing the need for refined modeling approaches. The study underscores the potential of hybrid modeling for livestock export forecasting and recommends integrating exogenous factors such as climate conditions, economic indicators, and policy shifts to enhance predictive accuracy. These findings offer valuable insights for policymakers and stakeholders to optimize decision-making and develop sustainable livestock trade strategies in Somaliland.