Integrating interannual climate variability forecasts into weather-indexed crop insurance. The case of Malawi, Kenya and Tanzania

In this study we explore the potential for re-insurance schemes built on regional climatic forecasts. We focus on micro-insurance contracts indexed on precipitation in 9 villages in Kenya, Tanzania (Eastern Africa) and Malawi (Southern Africa), and analyze the precipitation patterns and payouts resulting from El Niño Southern Oscillation (ENSO). The inability to manage future climate risk represents a “poverty trap” for several African regions. Weather shocks can potentially destabilize not only household, but also entire countries. Governments in drought-prone countries, donors and relief agencies are becoming aware of the importance to develop an ex-ante risk management framework for weather risk. Joint efforts to develop innovative mechanisms to spread and pool risk such as microinsurance and microcredit are currently being designed in several developing countries. While ENSO is an important component in modulating the rainfall regime in tropical Africa, the micro-insurance experiments currently under development to address drought risk among smallholder farmers in this region do not take into account ENSO monitoring or forecasting yet. ENSO forecasts could be integrated in the contracts and reinsurance schemes could be designed at the continental scale taking advantage of the different impact of ENSO on different regions. ENSO is associated to a bipolar precipitation pattern in Southern and Eastern Africa. La Niña years (i.e. Cold ENSO Episodes) are characterized by dry climate in Eastern Africa and wet climate in Southern Africa. During El Niño (or Warm Episode) the precipitation dipole is inverted, and Eastern Africa experiences increased probability for above normal rainfall (Halpert and Ropelewski, 1992, Journal of Climate). Our study represents the first exercise in trying to include ENSO forecasts in micro weather index insurance contract design. We analyzed the contracts payouts with respect to climate variability. In particular (i) we simulated possible payouts using historical precipitation data and analyzed the differences between years with different ENSO states from 1961 to 2005; (ii) we applied Monte Carlo methods to simulate precipitation distributions in each location and calculated the mean and variance of payouts associated to different ENSO states. The results obtained from historical precipitation data indicate that more abundant rainfall reduces payouts and the risk of loan default during La Niña in southern Kenya and Malawi, during El Niño in Tanzania. The results of the Monte Carlo simulations confirm our findings. Our results suggest that re-insurance schemes could be successfully designed to exploit the anti-correlation patterns related to interannual climate variability for different regions in Africa. Moreover, the exploratory framework presented can potentially be refined applied to other regions (e.g. Central and Latin America).