A Duration Analysis Approach to Variety Change on Wheat Farms in Bihar, India
*Names in bold indicate Presenter
Since the Green Revolution in mid-1960s, India’s wheat research program has developed and nationally released on average 14.2 new varieties each year. About 1.3 of these released varieties per year have been specifically targeted to the wheat growing zones of Bihar (Mittal et al., 2015). Despite such acclaimed progress, slow rate of variety replacement by farmers has posed a major challenge in promoting new wheat varieties in Bihar. According to the recent estimates of the adoption of wheat varieties in Bihar conducted using the expert elicitation method, the average wheat varietal age in Bihar is 14 years, which is higher in orders of magnitude compared to the agriculturally developed states of northern India. As a result, in the year 2010-11, the yields in Bihar was 1948 kg per hectare compared to 4600 Kg per hectare in Punjab (Directorate of Wheat- India, 2011). In other words, despite the high rate of wheat varietal releases in India, many farmers still cultivate older varieties, which implies that the genetic gains embedded in new wheat varieties is reaching the farmers at a much slower pace or at an uneven pace across the wheat production landscape. Thus, to better develop diffusion strategies and bring the benefits of agricultural research to farmers’ fields, the government and agricultural research system needs to understand the factors that determine wheat varietal adoption decisions, and how they differ across the target population. This paper is a first of its kind in systematically investigating this aspect in Bihar, India, which is the epicenter of the second green revolution efforts in India.
The findings of this paper thus adds to the debate on how/why poorest farmers are not benefitting as much from the improved research and development of national and international agencies. The methodology also suggests how using a duration model is a substantial improvement over usual logit/probit model by including time varying variables.