M.A. Samad Azad
University of Tasmania, Australia
Sanzidur Rahman
University of Plymouth, UK
Sample selection framework, Stochastic production frontier, Hybrid rice, Adoption, Production efficiency
Six districts of Bangladesh
Variety and Species
Adoption of technology, Efficiency
Study Area A survey was conducted to collect primary data on inputs and outputs of hybrid and inbred rice production. Using structural questionnaire rice farmers were purposively selected and interviewed at six districts of Bangladesh, where both the Bangladesh Rice Research Institute (BRRI) released hybrid rice and the imported hybrid rice varieties were primarily grown in the Aman (July–December) and Boro (January–June) seasons. The study areas cover both the North-west and South-west part of Bangladesh including Jessore, Barisal, Pabna and Magura. The sampling technique yielded a total number of 336 and 180 sample farmers for the year of 2004 and 2005, respectively. The selection of the study areas and the sample size for each area were purposively determined by the geographic location of hybrid rice producers and the intensity of BRRI released hybrid rice cultivation. Data and Variables Data on input and output quantities and their prices was obtained at the rice plot level. Some key information on the socio-economic characteristics of the sample farmers was also collected. To operationalize the adoption and efficiency models two sets of variables are needed: one set for the probit variety selection equation model, and the other for the stochastic production frontier model. In the probit equation, the criterion of the variety selection is treated as the dependent variable, which is a simply a binary variable that takes the value of 1 if a plot is planted with hybrid rice varieties, and 0 otherwise. On the contrary, the dependent variables considered in the probit equation includes farmers’ socio-economic characteristics (i.e., age, education, number of working family members, share of owned land, subsistence pressure), gross returns from rice and relative prices of production inputs (P’i : fertilizers, labour, and pesticides) normalized by the price of output (Py : price of rice). In addition, pesticide and organic manure users, and the location of rice farms were included in the model as dummy variables. The justification for inclusion of these variables into probit model is as follows. Farmers’ age is considered as one of the explanatory variables in the model as older and experienced farmers may have better access to information than younger peers, which may assist them in rice variety selection (Ransom et al. 2003). This statement contrasts with other findings (e.g., Van Dusen 2000; Uaiene et al. 2009) as they claim that younger households may be more flexible and hence willing to adopt new technologies than older households. As such, we have included farmers’ age to test its independent influence on variety selection decision. As an explanatory variable, farmers’ education is commonly used in many previous adoption studies (e.g. Uaiene et al. 2009; Rahman 2008; Wadud and White 2000). Educated farmers usually have better access to information as well as the greater capacity to understand the technical aspects of new technology that may influence rice variety selection. The education level of the farmer is, therefore, included in the model to test its influence on variety selection decision. The number of working member in a farm household may influence the adoption of new production technology. The studies conducted by Mariano, et al. (2012), Abdulai et al. (2008) and De Souza Filho et al. (1999) confirm that farm households with a higher number of working family members likely to adopt new technologies than smaller households. The impact of tenancy on modern technology adoption is varied (Hossain et al. 1990). Therefore, the share of own land is incorporated in the probit equation to examine its independent influence on the decision regarding the selection of rice variety. The subsistence pressure variable is included in the model to account for its influence on hybrid rice selection. Gross return variable is included in the model as farmers’ decision to rice variety selection can be driven by farm income. A number of studies (e.g., Ransom et al. 2003; Rahman et al. 2009; Rahman, 2011; Rahman and Chima 2014) confirm that gross return is one of the important determinants in rice varieties. In the stochastic production frontier model, eight input variables such as land, labour, seed, irrigation, chemical fertilizers, mechanical power, pesticides and organic manure are included. The selection of input variables for the frontier model is based on the existing literature of production efficiency which offers similar justifications. For example, the amount of cultivated land is considered as one of the explanatory variables in the production frontier model as many studies found that productivity of rice farm increase with area of land devoted to production (e.g., Baten and Hossain 2014; Rahman et al. 2012; Alam et al. 2011). There have been mixed findings related to influence of labour on technical efficiency of rice farms. For example, Fani et al. (2016) observe that technical efficiency of rice farms can be improved with the increase of labour use. But this claim is inconsistent with the finding of Alam et al. (2011). Thus the present study considers labour as an explanatory variable of the frontier model to further examine its effects on production efficiency of hybrid rice. Access to irrigation facilities is an important prerequisite for growing modern and/or high yielding rice varieties (Rahman 2008). Due to its greater influence on farm efficiency it is considered as an explanatory variable of a stochastics frontier model in many earlier studies (e.g., Rahman et al. 2012; Alam et al. 2011; Asadullah and Rahman 2009). The study attempts to examine the impact of fertilizer use on technical efficiency in hybrid rice production, as some previous studies show that fertilizer does have an impact on the efficiency of rice farms (e.g., Anik 2012; Miah et al. 2010). Other input variables such as seed and pesticides are also incorporated in the frontier model as these variables have significant impacts on the production efficiency (Baten and Hossain 2014; Rahman et al. 2012). The structural model satisfies the identification criterion as the input variables in the probit variety selection equation and the stochastic production frontier differ (Maddala 1983). To assess individual rice farm’s performance of all the input and output variables used in the stochastic production frontier were measured on a per farm basis.
The Journal of Developing Areas, Volume 51 No. 1 Winter 2017
Journal