Climate and Vegetation Indicators, Maize Yield, and Basis Risk In Index Insurance Evidence from Three Zambian Provinces, 2000–2023
Year: 2026
Type: article
Abstract: This thesis examines whether annual climate and vegetation indicators provide stable information about province-level maize yield and whether the same indicators can support credible triggers for index-based insurance in three Zambian provinces. The analysis uses 66 complete province-year observations for Lusaka, Northern, and North-Western provinces over 2000–2019 and 2022–2023. Pooled ordinary least squares is compared with province fixed effects, year fixed effects, two-way fixed effects, and a province-fixed-effects specification with a linear time trend; inference is assessed using HC1 robust standard errors and Driscoll–Kraay sensitivity estimates. Basis risk is evaluated with complete confusion matrices under pooled-mean, province-mean, and province-specific quantile thresholds. The rainfall–yield correlation is modest (r = 0.174). Rainfall is positive in pooled OLS, but its coefficient is not significant at the 5% level under either HC1 or Driscoll–Kraay inference. NDVI is negative in pooled and province-fixed-effects models, but this association disappears after common time structure is controlled. No environmental coefficient remains stable across the main specifications. Trigger performance is more consistent: rainfall produces lower mismatch and false-negative rates than annual province-wide NDVI under every threshold family examined. Simple AND and OR combinations do not uniformly improve performance; they redistribute basis risk between missed losses and false payouts. These results separate two empirical questions that are often conflated—whether an indicator is associated with yield and whether it is useful as a payout trigger—and highlight the importance of measurement scale, temporal alignment, and explicit error costs in agricultural risk management.
Source: Economics and Public Policy
DOI: 10.63313/epp.9057
Institution: Zhejiang University of Science and Technology
Language: English
Subfield: Soil Science
Domain: Life Sciences
SDGs: Zero hunger, Climate action
Open Access status: hybrid
OpenAlex ID: W7223205026