Although Myanmar has made progress in reducing malnutrition, its prevalence among young children remains high, as 26.7 percent of children age 6-59 months are moderately or severely stunted (MoHS, 2019). Furthermore, nutrient inadequacy remains widespread. The National Nutrition Centre (NNC) in the Ministry of Health and Sports (MoHS) has identified and implemented interventions for five conditions resulting from under-nutrition: protein energy malnutrition, iron deficiency anemia, vitamin B1 deficiency (also known as beriberi), vitamin A deficiency, and iodine deficiency disorder (ibid, 2019). For example, recent evidence from the Myanmar Micronutrient and Food Consumption Survey (MMFCS) finds that anemia is prevalent among children and women. The survey found that 35.6 of children aged 6-59 months and 51 percent of children 5-9 years of age are anemic, as well as 30 percent of both adolescent girls (age 10-14) and women of reproductive age (age 15 to 49) (ibid, 2019). While there are multiple factors that affect nutrition outcomes, one of the underlying causes of malnutrition is a lack of adequate food of sufficient nutritional quality (IFPRI, 2015). However, in every region of the world, the cost of protein- and micronutrient-dense foods, such as animal-source foods, fruits, and vegetables, are often considerably higher than the cost of energy-dense, staple foods such as cereals (Miller et al., 2016; Headey et al., 2018). While factors other than the cost of different foods may affect dietary choices and thus nutrition outcomes, relative food costs likely play an important role in household dietary choices, especially for poorer households. In recent years, the Government of Myanmar has made important commitments to reduce malnutrition in the country, including joining the Scaling Up Nutrition (SUN) Movement in 2013, joining the UN Zero Hunger Challenge in 2014, and bringing a number of ministries together in 2018 to create a Multi-sectoral National Plan of Action on Nutrition (MS-NPAN) (GoM, 2018). One goal of this paper is to inform the MS-NPAN by providing empirical analysis of household dietary patterns and the cost and affordability of a nutritious diet in Myanmar. This paper builds on previous empirical work on dietary patterns in Myanmar, as well as a recent approach from a reinvigorated international literature on estimating the cost of a nutritious diet (Masters et al., 2018; Dizon and Herforth, 2018). In this study, we use the Cost of a Recommended Diet (CoRD) approach demonstrated by Dizon and Herforth (2018) and developed by Herforth et al. (2018). This approach estimates the cost of consuming a nutritious recommended diet as defined by a country’s food-based dietary guidelines (FBDG). Because the CoRD method uses only a few lowest-cost food items from each recommended food group to estimate the cost of acquiring a recommended diet, it likely underestimates this cost compared to the cost if local tastes and preferences are taken into account. In order to estimate the cost of consuming a recommended diet using a set of foods that reflect these preferences, we propose a modification to the CoRD method called the Food Preferences CoRD (CoRD-FP). The CoRD-FP method estimates the cost of a recommended diet using prices from a wider range of foods that reflect current food consumption patterns and preferences, as observed in household survey data. In this paper, we apply these methods to household food expenditure survey data from Myanmar (2010 and 2015) to demonstrate the utility of these methods for evaluating economic constraints on nutrition, and to characterize those constraints in the specific and complex setting of Myanmar. Our objectives are to: a) Analyze household food consumption patterns in Myanmar relative to local and international definitions of a recommended diet; b) Use the CoRD method to estimate regional minimum costs of a recommended diet in Myanmar; c) Develop and demonstrate the CoRD-FP method to better reflect consumer preferences; d) Assess the affordability of the CoRD and CoRD-FP relative to household food expenditure; and e) Investigate the key drivers of the costs of the recommended diet using both the CoRD basket of minimum-cost foods and the CoRD-FP basket of preferred foods. Our key findings are summarized below. We find that, relative to recommended diet guidelines, a majority of households in Myanmar considerably under-consume all food groups except staples. In 2015, only 38 percent of the population lived in households that consumed the recommended quantity of proteinrich foods, 38 percent fats and oils, 16 percent vegetables, 9 percent fruits, and less than one percent consumed the recommended quantity of dairy products. These consumption patterns also vary considerably by region. For example, 47 percent of those in the Delta agro-zone consume the recommended quantity of protein, compared to 28 percent in the Hills and Mountains agro-zone. With the exception of staples, consumption of each of the other five food groups increases considerably as total household expenditure increases. For example, only 8 percent of those in the poorest total household expenditure quintile consume the recommended quantity of proteindense foods compared to 66 percent of those in the wealthiest quintile. Yet, even mean consumption per adult equivalent (AE) for households in the highest quintile falls below the recommended diet quantities for dairy, vegetables and fruit. This implies that income is not the only constraint to consuming a nutritious diet. Consumption of the recommended number of servings of protein foods, vegetables, fruit, and fats increased from 2010 to 2015. This dietary shift is consistent with the 7.2 percent per year increase in Myanmar’s GDP per capita between 2010 and 2015 (World Bank, 2019). However, even with increases in consumption of non-staple foods, many individuals lived in households that overconsumed staples relative to the recommended quantity, yet significantly under-consumed each of the other five recommended diet food groups. The results above beg the question of why so many Myanmar households tend to over-consume staples and under-consume all non-staple food groups. While factors such as food preferences and nutritional knowledge affect household dietary choices, relative food costs also play an important role in these choices, especially for poorer households. Consistent with recent research from countries throughout South and Southeast Asia (Headey et al. (2018)), we find that prices per calorie of the most micronutrient-dense foods in Myanmar are considerably higher than those of staple foods such as rice, which are calorie-dense yet relatively low in micronutrients. These results suggest that a key factor leading many Myanmar households to vi over-consume staples such as rice and under-consume more nutrient-dense foods is their inability to afford the latter. For example, the price per calorie of chicken and pork are 24 and 8 times higher, respectively, than the price per calorie of rice, while the average for a number of fish and seafood items is 18 times higher. Likewise, other perishable foods like eggs, fresh milk, and certain fruits and vegetables have high prices per calorie relative to rice. Next, we estimate the CoRD for Myanmar, develop and estimate a modification to this method – the CoRD-FP – and also estimate the cost of meeting caloric needs based on the lowest cost staple food (CoCA). We find that the CoRD and CoRD-FP are 2.5 and 3.7 times more expensive, respectively, than the CoCA, at the national level. We also find that the CoRD-FP is 47 percent more expensive than the CoRD. This implies that meeting the recommended diet using foods that reflect observed food preferences (i.e. the CoRD-FP) costs more than doing so using a relatively small number of minimum-cost foods (CoRD). The CoRD-FP thus captures a “preference premium” – the additional cost of acquiring a recommended diet based on a set of foods that reflect preferences within each food group. Differences in the cost of the protein and vegetable food groups explain nearly all of the gap between the total cost of the CoRD-FP and the CoRD, at the national level. For example, the recommended diet quantity of protein foods costs 3.5 times more for the CoRD-FP than the CoRD, and accounts for about three-quarters of the preference premium. The reason for this is the cost of the CoRD-FP protein food group is based on a combination of meat (chicken, pork, and/or beef), fish, eggs, and legumes. By contrast, the CoRD is based almost entirely on legumes, which are considerably less expensive per serving than animal-source foods. In addition, the recommended diet quantity of vegetables costs 46 percent more for the CoRD-FP than the CoRD and accounts for about a fifth of the preference premium. Half of the population lives in a household that cannot afford the CoRD-FP relative to actual household food expenditure, and about one quarter cannot afford the CoRD. However, the affordability of CoRD and CoRD-FP improved compared to 2010, when 70 percent of the population lived in a household that could not afford the CoRD-FP and 32 percent could not afford the CoRD. This improvement is consistent with a 24 percent decline in the poverty headcount from 42 to 32 percent over the same time period (MOPF and World Bank 2017b). For households that cannot afford the estimated cost of the diet, the average deficiency in food expenditure relative to the CoRD or the CoRD-FP is 6 and 16 percent, respectively. There are three main policy implications from these results. First, our results suggest that Myanmar’s food security and agricultural policies should focus on diversification of farm enterprises through improvements in farm-level productivity and reductions in the marketing costs of protein- and micronutrient-dense foods, such as animal-source foods, vegetables and fruits. A focus on diversification will increase farm incomes and increase the availability and affordability of nutritious foods. For many decades, food security and agricultural policies in Myanmar have primarily focused on increasing national production levels of rice (Robertson et al., 2018). For example, in recent years up to an estimated 85 percent of the annual budget for the agricultural sector in Myanmar has focused on rice production (GoM, 2018). In addition, the Myanmar Agricultural Development Bank (MADB) provides larger loans for the production of rice relative to other crops (Robertson et al., 2018), although current rice-based vii farming systems generate significantly less income for smallholders compared to most other production systems, such as those based on beans, pulses, oilseeds, aquaculture, and a wide range of other smallholder cash crops (GoM, 2018). Likewise, in some contexts, modifications to land use legislation could facilitate farm diversification. For example, there is a need to reduce administrative and legal barriers to enable smallholders to convert paddy land into permanent high value enterprises like aquaculture or floriculture. Second, different regions have different levels of agro-ecological and market access potential for production of protein- and/or micronutrient-rich foods. This implies that region-specific strategies are needed to overcome supply side (availability and cost) and demand side (household incomes, particularly for poorer households) constraints to increased household consumption of protein and micronutrients. For example, the CoRD-FP is highest in the Hills and Mountains, followed by the Delta. However, strategies to reduce supply and demand-side constraints to improving the quality of diets in these two areas are likely to be quite different. For example, the types of high value agricultural enterprises that can generate additional income are quite different in hilly and mountainous states compared to the Delta. Finally, though the relatively high cost of many micronutrient-dense foods is a key constraint to consuming a nutritious diet, dietary preferences also play an important role. This point is clearly illustrated by consumption choices of households in the highest expenditure quintile. Though 88 percent of households in the highest expenditure quintile have household food expenditure levels sufficient to afford the CoRD-FP, only 19 and 36 percent of these households consume the recommended diet quantities of vegetables and fruits, respectively. This highlights the need for nutrition education to encourage increased consumption of nutrientdense foods. We find that consumption of the CoRD-FP food basket in Myanmar meets the average macronutrient needs for adult men and women and the requirements for most micronutrients. Most notably, the CoRD-FP food basket meets average nutrient requirements for protein, iron, and vitamins A and B1, nutrients which the government of Myanmar has identified as problem areas requiring targeted intervention (MoHS, 2019). This indicates that efforts to encourage the population of Myanmar to consume a recommended diet could significantly reduce the prevalence of health conditions resulting from insufficient nutrient intake, such as anemia in women and children. Thus, creation of national FBDG containing a recommended diet specific to Myanmar could be a powerful tool for increasing public awareness of ways to overcome known dietary shortfalls. Both greater use of the FBDG as well as development of a recommended diet specific to Myanmar (which includes recommended consumption quantities for various food groups) could help improve the effectiveness of nutrition policy. Efforts to promote consumption of a more nutritious diet would need to address both the economic constraints to eating more nutrient-dense yet relatively expensive protein foods, fruits and vegetables, as well as nutrition education and promotion of healthy diets.