Seasonal climate forecasts are now being routinely released in Zimbabwe and elsewhere, with the expectation that this information will improve crop and resource management. Most studies focus on the household as the unit of analysis, with interventions designed to benefit production risk management at the household level. Here we investigate the implications in aggregate of a widespread response to climate forecast information using the case of Zimbabwe in the 1997/1998 El Nino event and the following year's La Nina, assuming that changes in observed area planted in those two seasons can be used as a guide to potential responses to forecast information. Data from the Zimbabwe National Early Warning crop statistics database and household level surveys were used in the analysis. In the 1997/1998 El Nino year, when the official forecast for a poor rainy season was broadly disseminated, decreases in area planted were observed, but in the following year when La Nina conditions and traditional indicators portended higher than average rainfall, area planted per household rose, particularly in the driest zone. Applying observed changes in area planted and crop mix to yields over the preceding 15 seasons, we show that the impact of a forecast of drought conditions could potentially decrease production below that which would result from behavior without a forecast, but production could potentially increase in years when the forecast is for greater than average rainfall. Since production increases in favorable years would be greater in magnitude than the potential decreases in poor rainfall years, long-term mean production could increase in the presence of forecasts. However, production volatility is also shown to increase. We suggest that, if forecast information is widely disseminated and adopted in the future, appropriate market or policy interventions may need to accompany the information to optimize societal benefit of climate forecasts. (C) 2002 Elsevier Science Ltd. All rights reserved.
619VZTimes Cited:12Cited References Count:27