Data This study uses the Household Income and Expenditure Survey (HIES) data from Bangladesh collected in 2000, 2005, and 2010 by the Bangladesh Bureau of Statistics (BBS). The BBS uses a two-stage stratified random sampling to ensure greater precision in its data collection process. In the first stage, the BBS selects primary sampling units (PSUs) consisting of specific geographical areas; whereas, in the second stage, it randomly selects 20 households from each PSU that represent rural and urban areas. For example, in the 2000 HIES survey, 7,440 households were randomly selected from seven divisions, 64 districts, and 303 sub-districts. In the HIES 2005, 10,080 households were randomly selected from seven divisions, 64 districts, and 355 sub-districts. Finally, in the HIES 2010, 12,240 households were randomly selected from seven divisions, 64 districts, and 384 sub-districts. Since information on food prices was missing for 28 households, this study uses information from 29,732 households, of which 9,203 were urban, and 20,529 were rural. The HIES data on household-level consumption is quite detailed. For example, a household’s consumption of food items, both in quantity and expenditure, was divided into 17 major food categories; households were queried on their consumption habits during a two-week period. Finally, the HIES data also collected information on rice consumption, by the grain type. These include fine-grain, medium-grain, and coarse-grain types of rice. The three grain types will be used as the dependent variables in the multivariate probit (MP) model. 4. Descriptive findings and model specification In Bangladesh, short and bold coarse-grain rice is considered as a poor quality rice and receives a lower market price for merchants and farmers. In contrast, long, slender grain rice is considered a fine-grain type of rice and receives a higher market price. In this study, we use the visual shape classification of grain type: coarse-grain, medium-grain, and fine-grain rice. It presents basic information on the sampled households and their rice consumption patterns based on whether or not the sampled respondents consumed fine-, medium-, and coarse-grain rice during the sampled years 2000, 2005, and 2010. Only 37 households reported no rice consumption during the sampled years. Out of the 29,760 households, about 8 percent of them (2,295) consumed fine-grain rice during the reference period, 39 percent (11,634) consumed medium-grain rice, and 68 percent (20,380) consumed coarse-grain rice. On average, the per capita daily rice expenditure, in Bangladesh Taka (BDT), was about BDT5, and the daily per capita consumption of rice was about 432 grams, of which about 8 grams were fine-grain rice, 137 grams was medium-grain rice, and 287 grams was coarse-grain rice. Note that the rice consumption information is presented on a per capita daily consumption basis. This means, for example, in total a sampled respondent consumed 2.92 kg of fine-grain rice yearly, and on average a sampled household with 4.81 family members, consumes 14.05 kg of fine-grain rice in a year. It shows that on average, 88 percent of the sampled households were headed by a male older than 45 years with about four years of schooling and 70 percent of them are likely to reside in rural areas. A female headed about 12 percent of the sampled households. On average, spouses had three years of education, and the average family size was about five members. It shows that, on average, sampled households spent BDT14 per capita daily for all food items. Based on this information, we equally divided all sampled households into four income quartiles consisting of 7,440 households (25 percent of the total sample) in each group, where the households located in the first income quartile (Q1) were the poorest 25 percent of households and the households located in the fourth quartile (Q4) were the wealthiest 25 percent. It shows that relatively highly educated household heads and spouses were more likely to consume fine-grain rice. They spent more on all food items compared to households consuming coarse-grain rice. Furthermore, more than 60 percent of the households (4,464) of the total 7,440 households in the wealthiest quartile (Q4) consumed fine-grain rice, while only 5 percent of the households (372) of the total 7,440 households in the poorest income quartile (Q1), consumed fine-grain rice. In total, 50 percent of the households residing in the urban areas consumed fine-grain rice over the sampled years, while only 24 percent of them consumed coarse-grain rice. Based on income quartiles, as income increased, the consumption of fine-grain rice increased and the consumption of coarse-grain rice decreased with the increase in income. The pairwise correlation between the daily total expenditure on food and the consumption of fine-, medium-, and coarse-grain rice also support the findings which are 0.18, 0.27, and −0.17 respectively, and all are significant at the 1 percent level (po0.00). The findings show that the educational attainment of the head and spouse, income, and urbanization (place of residence) positively affect consumption of fine-grain rice in Bangladesh. Finally, it shows that the sampled households are overall consuming less rice, but progressively consuming more fine-grain rice. This may indicate that, with the increase in income, households tend to consume less rice, but consume more finer-grain rice, replacing coarse-grain rice.