Khondoker A. Mottaleb
Social Sciences Division, International Rice Research Institute (IRRI), Los Ba ~ nos, Philippines and is currently a socioeconomist, CSISA-MI, CIMMYT, Dhaka, Bangladesh
Samarendu Mohanty
Social Sciences Division, International Rice Research Institute (IRRI), Los Ba ~ nos, Philippines
Andrew Nelson
Social Sciences Division, International Rice Research Institute (IRRI), Los Ba -nos, Philippines
Adoption, Farm households, Hybrid rice, Infrastructure, Modern variety
Socio-economic and Policy
Data sources This study relied on different data sets made available by the Bureau of Statistics (BBS) and Bangladesh Agricultural Research Council, under the Ministry of Agriculture of the government of Bangladesh. The household-level data on the absolute size of land owned by households, land under rice cultivation and the size of rice land under traditional and modern varieties and hybrid rice, family members, members in the family engaged in agriculture, sex of the household head, household heads depending on agricultural labour and informal loan information are taken from the Agriculture Census 2008 data sets made available by BBS. Sub-district-level data on population density, literacy rate, percentage of population aged 7 years and above who are not attending school but are engaged in agriculture and percentage of households that are connected to electricity have been extracted from the Population and Housing Census 2011 data sets made available by BBS. Note that the use of 2011 census data to determine the adoption of hybrid and modern rice varieties in 2008 may generate a data-matching problem. However, the Population and Housing Census was actually conducted in 2010, but the data of the census started to be released only recently. Thus, within 2 years’ difference, one cannot expect a big change in electrification, population growth and literacy rate in a developing country such as Bangladesh, where, because of shortages of electricity and natural gas, the government has almost stopped electrification. These best available data are used to control for sub-district-level influences on the adoption of different rice varieties. Sub-district-level data on the proportion of irrigated agricultural land and the total length of paved roads are collected from District Profile 2007, an online data source made available by BBS. The information on the number of government-approved seed dealers at the subdistrict level is collected from the directory of registered seed dealers, an online database made available by the Ministry of Agriculture. Data on land area at the subdistrict level that is characterised by high land and very low land are collected from Land Resources Inventory (LRI) data, an online source made available by the Bangladesh Agricultural Research Council (BARC), Ministry of Agriculture. The station-level data on total rainfall and monthly average maximum and minimum temperature in 2005 were initially collected from 33 weather stations from an online data source made available by BARC. Importantly, although the Agriculture Census 2008 covered all of Bangladesh, BBS provides access to only 5 per cent of the entire census data sets that included 531,676 households from 200 subdistricts in 26 districts in Khulna, Rajshahi, Rangpur and Sylhet divisions. Ironically, for a number of subdistricts, however, information on irrigated agricultural land and the length of paved roads was not available in the online district profile of BBS. This paper tried to supplement the information on irrigated land and the length of paved roads by collecting data from Banglapedia, an online data source. Unfortunately, even in, for a number of subdistricts, there is no information on irrigated land and the length of paved roads (Banglapedia 2013). Therefore, we had to drop those subdistricts from our sample. Our study is thus based on 384,337 sampled households from 135 subdistricts in 25 districts in Bangladesh. Characteristics of the sampled households by division The largest shares of the sampled subdistricts and households come from Rajshahi and Rangpur divisions. These two divisions are located in the northwest part of Bangladesh, characterised by relatively dry area compared with that of other divisions. As farmers need to buy seeds and inputs at the beginning of a season, later it is shown that informal loans significantly and positively affect the adoption of both hybrid and MV rice compared with traditional varieties. It shows that Rajshahi and Rangpur have the highest proportion of irrigated rice land, the two divisions where hybrid rice adoption is also relatively high. This demonstrates the importance of irrigation facilities for hybrid rice adoption in Bangladesh. In Sylhet, seasonal flash floods and waterlogging are the common abiotic stresses. As hybrid rice cultivation requires a substantial amount of investment in fertiliser and irrigation relative to inbred rice, farmers in the stress-prone areas might be discouraged from adopting a new technology that is costly but for which the return is risky. In the econometric estimation approach, land characteristics are included at the subdistrict level to examine how they affect the adoption of hybrid rice at the household level. The climate variables were derived from monthly data from 31 geo-referenced weather stations across Bangladesh. Station data were spatially interpolated to provide estimates for each subdistrict. An inverse distance weighting algorithm was used to create climate surfaces of each weather variable providing estimates on a 25-km resolution grid. These estimates were then averaged to provide climate values for each subdistrict and were then assigned to each household in its respective subdistrict. All spatial data processing and analysis were done in ArcGIS v 10.0 (Redlands, CA, USA).
Australian Journal of Agricultural and Resource Economics, 59, pp. 258–274
Journal