Abstract:
Soybean has become a strategically important crop for food security, protein supply, and industrial demand in Nigeria, yet its production is highly sensitive to climate variability that differs markedly across the country's agroecological zoThis study examines the influence of seven climate variables; temperature, rainfall, precipitation days, relative humidity, absolute humidity, a composite humidity index, and solar radiation on soybean production across the six geopolitical zones of Nigeria over the period 1990–2025, using 216 annual zone yearly observations. An integrated methodological framework combining hybrid model of panel econometrics, spatial autocorrelation analysis, and time-series forecasting was applied. The results on fixed effects estimation reveals that within zones, temperature (β=0.106,p=0.002), the humidity index(β=0.446,p<0.001), and solar radiation (β=0.129,p<0.001) exert positive and significant effects on log production, whereas rainfall has a negative and significant effect (β=-0.225,p<0.001), indicating that excessive within zone precipitation reduces output through waterlogging and disease pressure. Notably, the sign of the rainfall coefficient reverses between the pooled and fixed effects specifications, illustrating the ecological fallacy of conflating cross-sectional and temporal variation.Moran's I is consistently negative (mean -0.142), indicating spatial dispersion driven by the dominance of the Northcentral zone, and Lagrange multiplier tests favour a spatial error over a spatial lag specification. Log production is I(1) in all zones; zone-specific ARIMA(1,1,1) models used as a tractable approximation to the temporal component of a STARIMA framework produce out of sample forecasts with MAPE between 3.90% and 12.27%. A composite agroclimatic suitability index correlates strongly with production across zones (r=0.774), classifying Northcentral as highly suitable. The findings support zone specific, climate-smart agricultural policy and demonstrate the value of integrated spatial temporal methods for agricultural planning in data constrained settings.
Keywords: Soybean Production, Climate Variability, Panel Data, Fixed Effects, Spatial Autocorrelation, ARIMA, Nigeria.