|Title||Yield trend estimation in the presence of farm heterogeneity and non-linear technological change|
|Author(s)||Conradt, Sarah; Bokusheva, Raushan; Finger, Robert; Kussaiynov, Talgat|
|Source||Quarterly Journal of International Agriculture 53 (2014)2. - ISSN 0049-8599 - p. 121 - 140.|
|Department(s)||Agricultural Economics and Rural Policy|
|Publication type||Refereed Article in a scientific journal|
|Keyword(s)||Detrending yield data - Kazakhstan - MM estimator - Weather information|
An adequate representation of the technological trend component of yield time series is of crucial importance for the successful design of risk management instruments. However, for many transition and developing countries, the estimation of the technological trend is complicated by the joint occurrence of three phenomena: (i) a high level of heterogeneity among different farms in a region; (ii) non-linear development of technological change; and (iii) high yield variations as a consequence of high exposure of rainfed agriculture to extreme weather events. Under these situations, the usually applied approach to detrend crop yield data using Ordinary Least Squares is known to be biased. Based on a unique data set of 47 farm yield data from northern Kazakhstan, we evaluated different alternative approaches. First, we consider the use of the MM-estimator, a robust regression technique for detrending. Second, we evaluate the effect of adding information on extreme climate events as an additional regressor. Finally, we consider combinations of the two former approaches and compare the implications of the different aggregation level on trend estimations. The results reveal the importance of using single farm yield data for detrending, because technical trends in Kazakh wheat yields are highly farm-specific. Furthermore, our analysis shows that the estimation of technological trends can be improved by incorporating weather information in the regression model if time series of crop yield data contain severe fluctuations due to occurrence of climatic extreme events. Thus, the presented analysis contributes to an improved crop yield analysis for many developing and transition countries facing similar conditions.