Different source of basis risk: - Model (index and payout function) - Spatial (distance to station) - Idiosyncratic basis risk (individual / plot specificities). Weather index-based insurance: building on two ex ante evaluation in the sudano-sahelian zone A. Leblois (École Polytechnique), P. Quirion (CIRED & CNRS), B. Sultan (Locéan, IRD), [email protected] 4 – Additional results -Simple indices does not have much lower performance in both studies. - Simulated sowing dates perform well for millet - Sowing date need to be observed in the case of cotton (institutional issues: delay in input delivery) - Considering micro-fertilised millet plots the incentive to use insurance but does not insurance gains). - Sodecoton’s inter-annual implicit price insurance • We look for an optimal weather-index insurance policy for cotton farmers. • Cotton growers pay a premium every year and receive an indemnity if weather index < defined threshold = strike (S). • We assume an insurer’s charging rate (10% of all indemnifications), & a transaction cost (1% of average yield). • Insurance parameters (S, λ, M) optimized in order to maximize farmers’ certain equivalent income (CEI), with constant relative risk aversion utility function including initial wealth (W): Evaluation of risk aversion of cotton producers using lotery games: half of the sample (N=64) have a relative risk averion > 1. tested relative risk aversions: [1, 2 , 3]. 1 – Data Yield: Rainfall: – Methods: insurance contract optimization Matching yield &high density daily rainfall data. Indemnification: function of 3 insurance policy parameters: strike (S), max. indemnity (Μ) and a slope-related parameter (λ) S Μ λ index – High basis risk at different scales Sector level cotton yields (33 sectors, 1977-2010, N=849), obtained from Sodecoton (the Cameroonian cotton company). Rainfall (small circles) Weather (large circles) stations, > one per sector different sources: Sodecoton, IRD, GHCN (NOAA). Normalised Diff. Vegetation Index (NDVI) was also considered. Niger Northern Cameroon Plot specific yields of 30 farmers per villages surveyed (10 villages, 2004-2011, N=1780). Each farmer has one plot with traditional growing techniques (low or no fertiliser use) and one plot with micro- fertilisation. Rainfall station in each of the 10 villages surveyed. Plots < 3 kilometers from the nearest rainfall station. Cameroon: certain equivalent income gain of weather insurance always < 50% lower than the same insurance using the yield observation. Unique calibration of index insurance parameters for heterogeneous zones subsidization of the dryest one & taxation of the most humid one. ≠ calibration of index insurance parameters for each rainfall zones: Indices considered: Cumulative rainfall over the simulated and observed rainy season and for ≠ critical growing phases. Duration of the rainy season in days. NDVI (satellite data) in Cameroon. Very low CEI gains: ≈1.5% for millet in Niger lower than 1% in Cameroon Niger: dry area (<500mm/year) and millet yields largely depends on rainfall but: High intra-village variation of yields