2.1 Farm household model: The farm household model (FHM) has been employed for several decades to study the economic development issues of small-scale agricultural households in developing countries(Singh et al.1986). van Wijk et al. (2014) reviews the extensive literature on the application of FHM to study food security and climate change. Furthermore, Louhichi and y Paloma SG (2014) provide a review of mathematical programming FHMs designed for policy analysis in developing countries. The primary advantage of the FHM is it provides an a semi-commercial and semi-subsistence framework where farm households simultaneously make production/consumption and labor/leisure decisions. We modify a standard FHM to reflect Bangladesh rice production. We outline the structure and key assumption of our model; see AppendixAfor expanded discussion, technical details, and corresponding mathematical equations. We develop an ann-region farm house-hold model with representative farmers in each region simultaneously make consumption and production decisions: Farmers (a) derive utility from annual consumption of rice, composite non-rice agricultural good, manufacturing goods, and leisure and (b) produce two agricultural goods (rice and non-rice agricultural crops) using family labor, variable inputs(seed, fertilizer, and pesticides), and land via a constant-elasticity-of-substitution production function for both incomes through market sales and consumption (rice is the subsistence good, modeled via a constant-elasticity-of-substitution, Stone-Geary utility function). Based on relative output prices, farmers optimally allocate land between rice and other agricultural crops. The opportunity cost of rice and non-rice consumption is forgone income, which highlights the production and consumption trade-off. The manufacturing good is a composite consisting of all non-food expenses (e.g., cookware, bedding, medicine) which ensures the model captures total consumer expenditures. In addition to leisure and farm work, time can be spent working off the farm to earn additional income, which highlights the labor and leisure trade-off. However, rigidity in the non-farm wage rate causes unemployment, which breaks the separability between farmers’ consumption and production decisions (see Singhet al.1986for a detailed discussion on the separability condition).From the farm household perspective, prices for the manufacturing good and variable inputs (seed, fertilizer, and pesticides) are exogenous. However, the national prices for rice and the composite non-rice agricultural good and the land rental rates for each region are endogenously determined through market-clearing conditions: For rice, the national price adjusts until total supply (sum of domestic rice production from the regions plus rice imports from the rest of the world (ROW)) equals total demand (sum of farmers’ rice consumption from the regions plus rice consumption of the non-farm (or urban) population plus government purchases of rice). For the non-rice agricultural good, the national price adjusts until total supply (sum of domestic non-rice production from the regions)equals total demand (sum of farmers’ non-rice agricultural good consumption of the from the regions plus non-farm (or urban) composite agricultural consumption). For land, each region’s rental rates adjust until total supply in region equals total demand for rice and non-rice agricultural production. Within a given region intra-household effects are not considered; however, with endogenous prices, the model captures intra-regional effects as representative households alter their decisions. The model described above consists of an economically consistent system of equations that is numerically solved using Newton’smethod (the simulation is discussed in detail below). The above model differs from the simulation literature analyzing climate change in Bangladesh (Yu et al.2010and Thurlow et al.2012, who implement large-scale general equilibrium models to analyze the impacts of climate change in Bangladesh) in two key ways: First, the model is partial equilibrium with a primary focus on the agricultural sector, which is directly impacted by climate change. This model allows for transparency in the structure and assumptions of the model. Second, farms are structured as households rather than commercial farms. Household farms are semi-commercial and semi-subsistence that account for elements of both producer and consumer theory. Thus, farmers simultaneously make production versus consumption and work versus leisure decisions, which previous mathematical programming studies analyzing climate change in Bangladesh do not capture.Explicitly modeling these household intricacies extends the current literature by providing a new semi-commercial and semi-subsistence perspective on the impact of climate change on prices, production, consumption, and welfare of farmers. 2.2 Data and calibration: We consider four regions (coastal, central, and northern, and eastern) for the analysis; based on the Household Income and Expenditure Survey (HIES) (HIES2010), the coastal, central, northern, and eastern regions account for, respectively, 17.5%, 31.4%, 41.9%, and 9.3% of rice production. Therefore, withi=1,2,3,4, the above system contains 48 equations with70 exogenous parameters and variables to be calibrated based on the literature and data for the model to be numerically simulated.The primary data source is HIES conducted by the Bangladesh Bureau of Statistics in collaboration with the World Bank. To ensure accuracy and correct scaling of the model parameters, the HIES data is supplemented with aggregate production and trade data from FAO Stat (2017). Since the non-rice agricultural commodity (for both production qia and consumption cia) is a composite agricultural good, quantities of all agricultural goods are converted into their kilo-calories (kcal) equivalents base on data from Shaheen et al. (2013)to ensure consistency across agricultural commodities. Therefore, the production and consumption decisions of farmers involve a trade-off between the annual production of kcals of rice and kcals of non-rice composite agricultural goods. The Online AppendixBdiscusses in detail the calibration procedure for the parameters in the utility function, total available time, unemployment constraint, production functions, ROW rice supply function, and demand functions for non-farm rice consumption, government purchases of rice, and non-farm composite agricultural consumption such that the model accurately reflects the HIES and FAOSTAT data.