Causal effect estimation of seasonal forecast skill on maize acreage reallocation using instrumental-variable ensembles: a climate-analytics evaluation of advisory-value attribution

Seasonal climate forecasts are promoted as a lever for smallholder adaptation, yet the causal value of forecast skill for acreage decisions remains poorly identified because skill and uptake are jointly determined by unobserved farmer capacity and institutional access. The framework separates forecast skill into a public signal component and a farmer-specific interpretation component, defines the local average treatment effect of skill on acreage reallocation, and specifies ensemble instruments constructed from lagged oceanic and atmospheric indices. A monotonicity condition and an exclusion restriction are stated, and a bound on the bias from instrument invalidity is derived. The Kampala reanalysis serves as a calibration case for the temperature signal that enters the forecast-skill construct, not as a test of the causal model. Implications for the design of forecast-advisory evaluations are drawn.