Demand for and Supply of Pulses and Oil Crops in Bangladesh: A Strategic Projection for These Food Item Outlooks by 2030 and 2050

Abstract

The food items, such as pulses and oil crops, are influential means of nutritional security for the people in Bangladesh. Pulses are widely called here as an alternative to meat for poor households. The study aimed to predict productivity and national demand of pulses and oil crops in Bangladesh by 2030 and 2050, minimizing the supply and demand gaps of these crops. Using the ARIMA model developed by Box and Jenkins, the current study projected Bangladesh’s pulse and oil crop demand and supply for the years 2030 and 2050. The projections showed that the total demand for pulses in 2030 will be 17.9 lakh MT and further increased to 19.5 lakh MT in 2050. The deficit in the supply of pulses will be 12.36 lakh MT in 2030 and 9.26 lakh MT in 2050 although the shortage of pulses will disappear as a result of productivity improvements and innovations. Per capita consumption of edible oil in Bangladesh is 20–22 g per day. The majority of domestically produced soybean (5% of total demand) is used in the feed industries, while edible oil from soybean depends on import. The supply of mustard oil is predicted to be in deficit by 0.30 lakh MT in 2030 and 1.68 lakh MT in 2050. Nevertheless, the estimates also warn that supply of these food items could be threatened due to climate changes. In facing future challenges, supportive government policy and substantial investment in research and extension should be given priority for technological innovation and productivity improvement. Government also needs to develop a strategic import substitution policy for higher production of these crop and storage facilities.

1. Introduction

During the period of 2000–2010, the agriculture sector in Bangladesh played a crucial role in enormously lowering poverty by 69%, and further progress in agriculture remains important as Bangladesh’s economy repeatedly evolves [1,2,3,4,5]. Since the 1980s, Bangladesh has made massive investments aligned with many changes in agricultural policies, which have greatly aided the country in achieving self-sufficiency in cereal production, in particular rice, and key drivers for both food availability and access [6,7,8].

Nonetheless, the policies are causing a substantial bias against a more diversified production. The agri-food sector in Bangladesh has substantial market prospects for productive diversification and increased value addition for the agri-food sector in Bangladesh. Bangladesh’s dietary patterns evolve due to the rapid urbanization and faster income growth. The demand for nutrient-dense foods grows; however, the average amount of cereal consumed declines, and food systems in Bangladesh are transforming rapidly [9,10,11]. Note: Lakh MT is equivalent to 0.1 million MT where MT indicates metric ton.

Bangladesh has a huge comparative advantage of producing rice, pulses, oil seeds, potato, onion, maize, vegetables, chili and garlic. Even though there is a potential of pulses in growing countries of the world, Bangladesh could not achieve self-sufficiency in pulses production. Besides, the high population growth in Bangladesh causes pulses deficiency in this country [8,12,13,14]. There is, therefore, a lot of room for crop diversification. According to the comparative advantage study that was done, Bangladesh can grow a wide range of crops either to replace imported goods or to export [9]. High-value crops become more important in Bangladesh; their shares of the food consumption basket in 2020 were 40% and 49% in rural and urban areas, respectively. Given high-income elasticities, this share is expected to further increase in the future [15,16].

Current climate change events often affect the food security of the millions of people of Bangladesh. It is now one of the most vulnerable countries for climate risks in the world. In Bangladesh, damage caused by natural catastrophes is threat to production and food security for poor households [11,12,13,17,18].

Food security is a prime concern of many developing countries. There is a growing body of literature of concern on the projection of demand for and supply of food in the universe. The world food system is facing pressure due to increased population and adverse impact of climate change that will intensify in the next decades [17,19,20,21]. Increased food production will require more land, water and energy. Many countries will suffer due to scarcity of water, and completion of resources will intensify [19,21]. Agriculture sector is dynamic, changing with demand of people, availability of technology and change of management practices. In search of the answer to the question: Could the future growth of supply of food of a country match its increased demand for food as a result of population pressure and rising income? A number of studies have been conducted in this regard to project demand for and supply of key food items in various countries and assessed gap [14,19,22,23,24,25,26,27,28].

Many emerging nations are extremely concerned about their food security. A rising number of literature focuses on the major concern about the projected global supply and demand for food. The burden on the global food system will increase over the coming decades as a result of rising population and the negative effects of climate change [21]. Increased food production will require more land, water and energy. Many countries will suffer from water shortages and resource depletion, which will intensify [19,21,29]. Agriculture sector is dynamic, evolving in response to consumer demand, technological advancements and changes of management practices. It is found to have a negative relationship between poverty rates and growth in GDP per worker from the agriculture and non-agriculture sector in Bangladesh [23]. The researchers [1,23] conducted studies in 25 countries of Asia and Africa, which found that a 1% increase of income of agriculture GDP per worker would reduce poverty by 0.39% but 0.11% for 1% growth in non-agriculture GDP per worker in Bangladesh.

To facilitate this planning, projections of future supply and demand for food are important. Despite the fact that agricultural growth has been higher than the rate of population growth, concerns have been raised as to whether the land mass of Bangladesh is actually capable of supporting its expanding population by 2030 and 2050.

To feed an expanding population, planning for future food demand is crucial to address sustainable food security challenges. Future supply and demand estimates for food are crucial for this planning [8,11]. Notwithstanding the fact that agricultural growth has outpaced population growth, questions have been raised over whether Bangladesh’s land mass will be able to feed the country’s growing population by 2030 and 2050. The present study is an attempt to carry out such future projections with a view to assessing the likely gap between supply of and demand for cereal in Bangladesh and the implications on food security. The objectives of the study are to estimate supply and demand projections of pulses and oil crops produced in Bangladesh in 2030 and 2050; to estimate gaps in demand and supply of pulses and oil crops in 2030 and 2050; to carry out sensitivity analysis under alternative scenarios; and to suggest recommendations and policy implications.

Therefore, this study will be useful for the researchers and policymakers for the development of new crop variety, generation of new technology and actionable policy in the future.

2. Methods and Model

The Model for Forecasting Pulses and Oil Crops

The study used Box–Jenkins Autoregressive Integrated Moving Average (ARIMA) time-series methodology for modeling and forecasting [30]. It is a widely used model in the world for forecasting purposes because of its very low forecast error and reliability of estimates (Figure 1). Many studies have used the ARIMA model for projections of demand for and supply of food [14,17,18,31,32,33,34,35].

There were six steps used for the following structure of the ARIMA model:

In the first step the ARIMA time series model was used to forecast the per capita calorie intake to 2030 and 2050 [30]. The ARIMA (p, d, q) model has p and q autoregressive and moving average terms of the stationary time series of order d; it can be expressed as:

(Y_t − α₁Y_t−1 − α₂Y_t−2 − … − α_pY_t−p) = β₀ + (u_t − β₁U_t−1 − β₂U_t−2 − … − β_qU_t−q)

(1)

where Y_t is the dth difference of the original time series and U is a random noise. Differencing of the original variable gives a stationary time series t.

Step 1: The total calorie demand is the product of per capita calorie demand and population projections. We used the population projection of the Bangladesh Bureau of Statistics, Ministry of Planning of the Government of Bangladesh. The population of Bangladesh in 2030 is 188.12 million and in 2050 is 205.02 million [36].

Step 2: Estimation of the food demand. The food to calorie conversion ratio (g/kcal) is the quantity of food required to provide one kcal of nutritional supply. For rice, it is about 0.28 g/kcal. Calorie content of selected crops has been taken from the Institute of Nutrition and Food Science, University of Dhaka, Bangladesh.

Step 3: Estimation of the feed demand. The feed to calorie conversion ratio of a crop is the ratio of the total quantity of crops–products used as feed to the calorie supply from animal products. The feed demand of cereals increases with the increasing supply of animal products [37,38].

Step 4: Estimation of the total crop demand. It is the sum of the demand for food, feed, seed and other uses and waste.

Step 5: Estimation of the food supply. The ARIMA models predict the area and yield of selected crops. The procedures used in the Box–Jenkins ARIMA Method are shown in Figure 2.

To estimate ARIMA models and projections, the statistical program NCSS (version 12) was used in this study. This forecasting software identified the most parsimonious ARIMA model. Time series without trends, known as stationary time series, are used in ARIMA analyses. Using differencing, the time series data is initially made stationary. Then the order of the ARIMA called p and q is determined through generating Portmanteau test statistics. The results of the Portmanteau test indicate whether the model is suitable or not. If the model is not sufficient, the order of the model is changed and re-run; the process is repeated until a reliable model is estimated.

3. Sources of Data

Secondary time series data were rounded up from the Bangladesh Bureau of Statistics (BBS), Department of Agricultural Extension (DAE) of Bangladesh and on-line time series data from the Food and Agriculture Organization known as FAOSTAT [39,40]. Other sources, such as National Agricultural Research Institutes, Reports, Journals and relevant studies, were also used. The time series data of calorie intakes for ARIMA modeling have been taken from the Food Balance Sheets of the FAOSTAT database [39,41]. These include per capita calorie supply of cereals: rice, wheat and maize. The time series data for ARIMA modeling of the crop’s areas, production and yields have been taken from BBS and DAE for the period 1971 to 2022. Also, primary data were collected through regional workshops, Focus Group Discussions (FGDs), Key Informant Interviews (KIIs) and consultations with scientists, extension experts and farmers for the validation of models.

4. Results

4.1. Projections of Calorie Consumption Demand of Pulses and Oil Crops

Most of the estimates were found to be significant as indicated by standard errors of the estimates, t-values, and all the models were adequate as judged by Portmanteau Test statistics (

Table 1. Parameter estimates of ARIMA model of calorie intake of pulses and oil crops.

Crops	Model	Parameter	Standard Error	t-Value	Portmanteau Test Value	Portmanteau Test Decision
Peas	ARIMA (0,1,1)	0.670 **	0.097	6.8963	20.49	Adequate model
Other Pulses	ARIMA (0,1,1)	0.484 **	0.116	4.12	22.32	Adequate model
Mustard	ARIMA (0,1,1)	0.598 **	0.103	5.828	34.82	Adequate model
Ground nut	ARIMA (0,1,1)	−0.183	0.130	−1.408	37.85	Adequate model

Source: Authors’ estimation from ARIMA model, Note: ** p < 0.01.

). We have validated the forecast models by comparing our forecast results in 2021 with actual data of 2021 and have found that ARIMA models have very insignificant forecast errors.

4.2. Composition of Calorie Intake

Dietary diversity in the Bangladeshi diet slowly changes over the years. Cereals still account for a significant portion of caloric intake, but their share has steadily declined over the base period from 89.6% in 1990 to 80.5% in 2021 and further fell to 79.6% in 2030 and 77.5% in 2050 (Figure 3). The contribution to calorie intake followed by animal and fish products gradually increased during 1990 to 2021 and will continue to increase during 2030 to 2050. However, the absolute intake of these increased from 83 kcal/person/day to 290 kcal/person/day during 1990–2021 and will further rise to 310 kcal/person/day in 2030 and 354 kcal/person/day in 2050.

The share of pulses slightly decreased from 2.51% in 1990 to 2.06% in 2010, and afterward in the projection, consumption of pulses will slightly increase to 2.62% in 2030 and will be 2.5% in 2050. The share of oil rose from 2.33% in 1990 to 2.97% in 2010 and will further rise to 3.74% in 2030, and beyond that, it will rise to 4.35% (Figure 3 and Figure 4).

The share of potato in total calorie intake increased from 0.76% in 1990 to 2.82% in 2010, and it will further increase to 4.12% in 2030 and after that will remain stable (Figure 5). Absolute consumption of potato increased from 15 kcal/person/day in 1990 to 86 kcal/person/day in 2010 and will further rise to 101.5 kcal/person/day in 2030; after that, it will remain stable. The share of calorie intake from potato seems to be reaching a level of saturation beyond 2030.

The share of vegetable in total calorie intake increased from 0.46% in 1990 to 0.98% in 2010, and it will further increase to 1.45% in 2030; after that, it will remain stable (Figure 5).

Absolute consumption of vegetable would increase from 9 kcal/person/day in 1990 to 35.5 kcal/person/day in 2030, and after that, it would remain stable. Similar to potato, the share of calorie intake from vegetable is found to be reaching a level of saturation beyond 2030. The share of fruits in total calorie intake increased from 0.81% in 1990 to 1.42% in 2010 and would further slightly increase to 1.53% in 2030; after that, it will remain stable (Figure 5). Absolute consumption of fruits would increase from 16 kcal/person/day in 1990 to 37.5 kcal/person/day in 2030, and after that, it would remain stable. Similar to potato and vegetable, the share of calorie intake from fruits seems to be reaching a level of saturation beyond 2030.

The ARIMA forecasts show that the consumption of animal products (meat, milk, egg and fish) and non-cereal crops/products (potato, pulses, oil vegetables and fruits) followed similar increasing trend during 1990 to 2030 (Figure 3). Beyond 2030 the consumption of animal products will further increase (Figure 5).

4.3. Demand for Pulses

The predicted demand for pulses are validated with the actual demand of 2021. It appeared from primary survey that farmers produce 1.75 lakh Metric ton (MT) of lentils annually, and the country’s yearly requirement is around 6–7 lakh MT. Lentil import was 5.21 lakh MT in 2020 (BBS, 2020). The estimates showed that annual demand for lentil in Bangladesh in 2021 was 6.4 lakh MT. Also, the estimates of demand for chickpea, pea, mungbean, black gram and grass pea are consistent. The total demand for pulses in 2030 will be 17.9 lakh MT, and there will be an increasing trend over the years. In the year 2040, it will rise to 18.9 lakh MT and will further rise to 19.5 lakh MT in 2050. Crop wise estimates show that annual demand for chickpea, pea, lentil, mungbean, black gram and grass pea in 2030 in the country will be 1.4 lakh, 3.6 lakh, 7.1 lakh, 2.8 lakh, 1.3 lakh and 2.8 lakh MT, respectively. All these crops were found to have increasing trends individually over the years, and the corresponding consumption demand in 2050 would be 1.6 lakh, 4.0 lakh, 7.8 lakh, 3.1 lakh, 1.6 lakh and 3.1 lakh MT, respectively (

Table 2. Projections of demand for pulses.

Year	Demand (lakh MT)
	Chick Pea	Pea	Lentil	Mungbean	Blackgram	Grass Pea	All Pulses
2021	1.3	3.3	6.4	2.6	1.3	2.6	16.1
2030	1.4	3.6	7.1	2.8	1.3	2.8	17.9
2035	1.5	3.8	7.3	2.9	1.5	2.9	18.4
2040	1.5	3.8	7.5	3	1.5	3	18.9
2045	1.5	3.9	7.7	3.1	1.5	3.1	19.3
2050	1.6	4	7.8	3.1	1.6	3.1	19.5

Source: Authors’ estimation from ARIMA models, Note 1 lakh = 0.1 million.

).

4.4. Demand for Oil Crops

Per capita consumption of edible oil is 20–22 g per day. Most households prefer soybean oil for home cooking purposes, but such oil is often blended with palm oil. In 2021, total edible oil consumed in the country constitutes palm oil 51%, soybean oil 43%, mustard oil 5% and ground nut 1% (Figure 6). As the food processing sector grows, domestic consumption of soybean and palm oil is also expected to grow. It is projected that demand for mustard oil will increase from 1.48 lakh in 2021 to 2.1 lakh MT in 2030. It will further rise to 2.94 lakh MT in 2040 and 3.94 lakh MT in 2050 (

Table 3. Projections of demand for edible oil.

Year	Demand for Edible Oil (Lakh MT)
	Ground Nut	Mustard	Soybean
2021	0.14	1.48	11.95
2030	0.14	2.10	13.93
2035	0.15	2.48	14.77
2040	0.14	2.94	15.59
2045	0.14	3.46	16.34
2050	0.13	3.94	17.00

Source: Authors’ estimation from ARIMA model.

). Demand for ground nut oil will be 0.14 lakh MT in 2030, and it will remain almost similar till 2045 and will slightly decline to 0.13 lakh MT in 2050. Demand for soybean oil will increase from 11.95 lakh MT in 2021 to 13.93 lakh MT in 2030, and thereafter it will gradually rise to 17 lakh MT in 2050.

Local soybean production contributes approximately 5% of total annual soybean demand in Bangladesh. Domestically produced soybean is used predominantly in the feed industry while edible oil is imported both in crude form and as beans that are crushed locally. According to USDA estimates [38], Bangladesh’s soybean imports for 2021–2022 stand at 26 lakh MT and soybean oil imports at 7 lakh MTs. There are about 80 soybean oil refineries in Bangladesh that import crude soybean oil and refine it for the domestic market.

4.5. ARIMA Models of Areas of Pulses and Oil Crops

All the models were found to be valid with significant parameters as indicated by standard errors of the estimates, t-values and Portmanteau test results (

Table 4. Parameter estimates of ARIMA models of areas of pulses and oil crops.

Crops	Model	Parameter	Standard Error	t-Value	Portmanteau Test Value	Portmanteau Test
Peas	ARIMA (1,2,0)	−0.391 **	0.131	−2.975	27.07	Adequate model
Chickpea	ARIMA (1,2,0)	−0.470 **	0.125	−3.7345	37.40	Adequate model
Lentil	ARIMA (1,2,0)	−0.486 **	0.124	−3.895	27.82	Adequate model
Mung bean	ARIMA (1,2,0)	−0.494 **	0.124	−3.981	29.92	Adequate model
Black gram	ARIMA (0,2,1)	0.975 **	0.017	54.436	26.21	Adequate model
Grass pea	ARIMA (1,2,0)	−0.426 **	0.129	−3.295	14.76	Adequate model
All Pulses	ARIMA (0,1,1)	0.918 **	0.054	16.971	14.05	Adequate model
Mustard	ARIMA (1,1,1)	−0.809 **	0.234	−3.455	29.09	Adequate model
		−0.905 **	0.1759142	−5.143
Ground nut	ARIMA (0,2,1)	0.968 **	0.015	62.115	46.40	Adequate model

Source: Authors’ estimation, Note: ** p < 0.01.

).

4.6. Projection of Areas of Pulses

During the period from 1971 to 1985, the pulses area of Bangladesh increased from 0.68 million ha to the peak of 1.53 million ha. After 1985, the pulses area gradually declined, and it was lowest in 2010 at 0.72 million ha. Thereafter it slightly increased to 0.69 million ha in 2021. The ARIMA model predicts that the pulses area will increase to 0.80 million ha in 2030 and will rise to 1.00 million ha in 2050 (Figure 7).

4.7. Projection of Areas of Groundnut and Mustard Crops

During 1971–2021, the groundnut area of Bangladesh increased from 32 thousand ha to 35 thousand ha. During the same period, the mustard area of Bangladesh increased from 217 thousand ha to 309 thousand ha; the ARIMA model predicts that all these crop areas will have increasing trends (Figure 8). The total ground nut area will increase to 42 thousand ha in 2030 and will rise to 61 thousand ha in 2050, and the mustard area will increase to 343 thousand ha in 2030 and will rise to 377 thousand ha in 2050.

4.8. Estimated Models of Yields for Pulses and Oil Crops

All the models were found to be valid with significant parameters as indicated by standard errors of the estimates and t-values and Portmanteau test results. The projections of crop yields below assume that factors that contributed to growth in the past, such as advances in technology and high-yielding crop varieties, will continue to be developed and contribute to yield increases (

Table 5. Parameter estimates of ARIMA models of yields of pulses and oil crops.

Crops	Model	Parameter	Standard Error	t-Value	Portmanteau Test Value	Portmanteau Test Decision
Peas	ARIMA (1,1,0)	−0.487 **	0.161	−3.021	61.04	Adequate model
Chickpea	ARIMA (1,1,0)	−0.526 **	0.131	−4.012	57.37	Adequate model
Lentil	ARIMA (1,2,0)	−0.274 **	0.137	−1.999	29.55	Adequate model
Mung bean	ARIMA (1,2,0)	−0.599 **	0.117	−5.108	48.22	Adequate model
Black gram	ARIMA (1,1,0)	−0.281 **	0.165	−1.708	56.72	Adequate model
Grass pea	ARIMA (1,2,0)	−0.677 **	0.106	−6.361	57.64	Adequate model
All Pulses	ARIMA (1,2,0)	0.618 **	0.113	−5.454	53.10	Adequate model
Mustard	ARIMA (0,1,1)	0.293 **	0.135	2.156	46.12	Adequate model
Ground nut	ARIMA (0,1,1)	0.612 **	0.108	5.636	47.76	Adequate model

Source: Authors’ estimation, Note: ** p < 0.01.

).

4.9. Projections of Yields of Pulses

The average yield of all pulses will increase by 3% annually between 2021 and 2030 and will reach from 1.1 ton/ha to 1.4 ton/ha in 2030. It will further increase by 1% annually between 2030 and 2050 and will reach to 1.7 ton/ha by 2050 (Figure 9). All the selected 7 pulses crops will have increasing trends in yields during 2030 to 2050. The projected yield of pea, chick pea, lentil, Mungbean, Blackgram and grass pea will be 1.1, 1.2, 1.3, 1.5, 1.0 and 1.2 ton/ha, respectively, in 2030. Also, all these yields will further rise to 1.3, 1.3, 1.5, 2.8, 1.1 and 1.4 ton/ha, respectively, in 2050 (Figure 9).

4.10. Projections of Yields of Oil Crops

It is predicted that the average yield of ground nut will increase by 2.6% annually between 2021 and 2030 and will reach from 1.78 ton/ha to 2.1 ton/ha in 2030. It will further increase by 1% annually between 2030 and 2050 and will reach to 2.34 ton/ha by 2050 (Figure 10). The average yield of mustard will increase by 1% annually between 2021 and 2030 and will reach from 1.14 ton/ha to 1.22 ton/ha in 2030. It will further increase by 1.14% annually between 2030 and 2050 and will reach to 1.50 ton/ha by 2050 (Figure 10).

4.11. Projection of Supply of Pulses

Among the six pulses, only lentil, mungbean, grass pea and pea will have increasing trend in 2030 and will continue till 2050 (

Table 6. Projections of production (supply) of pulses.

Year	Supply of Pulses (Lakh MT)
	Chick Pea	Pea	Lentil	Mungbean	Blackgram	Grass Pea	All Pulses
2021	0.05	0.08	1.77	0.41	0.37	1.31	3.99
2030	0.05	0.17	2.28	0.76	0.33	1.94	5.54
2035	0.04	0.22	2.72	0.99	0.3	2.48	6.76
2040	0.04	0.29	2.99	1.24	0.28	2.85	7.69
2045	0.03	0.35	3.48	1.58	0.28	3.47	9.2
2050	0.02	0.44	3.76	1.89	0.26	3.87	10.24

Source: Authors’ estimation from ARIMA model.

). Lentil production will increase from 1.77 lakh MT in 2021 to 2.28 lakh MT in 2030 as yields increase from 1.1 ton/ha to 1.3 ton/ha and area from 1.14 lakh ha to 1.76 lakh ha.

Similarly, it will further rise to 3.76 lakh MT in 2050 due to advanced technology and area expansion. The growth of supply pattern for Mungbean, pea, Grass and Chickpea, including Blackgram in the projected period, was mostly similar and the advanced technologies and area expansion will be a major contributor to production of these crops (Table 6).

4.12. Projections of Surplus or Deficit of Pulses Supply

Projections of deficit of supply of pulses for the period 2030 to 2050 are presented in

Table 7. Projections of surplus and deficit of supply of pulses by 2030 and 2050.

Year	Deficit of Pulses (Lakh MT)
	Chick Pea	Pea	Lentil	Mungbean	Blackgram	Grass Pea	All Pulses
2021	−1.23	−3.20	−4.65	−2.15	−0.92	−1.26	−12.12
2030	−1.38	−3.47	−4.84	−2.09	−1.10	−0.91	−12.36
2035	−1.43	−3.53	−4.61	−1.94	−1.16	−0.46	−11.66
2040	−1.47	−3.55	−4.53	−1.76	−1.22	−0.16	−11.19
2045	−1.50	−3.58	−4.19	−1.48	−1.25	0.41	−10.06
2050	−1.53	−3.55	−4.00	−1.22	−1.29	0.77	−9.26

Source: Authors’ estimation from ARIMA model.

. There will be deficits of supply of all pulses in 2030 amounting to 12.88 lakh MT. Deficit of lentil, Mungbean and grass pea will decrease over the years due to major contribution of increased productivity. As a result, the total shortage of all pulses will gradually decline over the years. There will be a deficit of all pulses with the amount of 9 lakh MT in 2050.

4.13. Projection of Supply of Ground Nut and Mustard

The ground nut production will increase from 0.58 lakh MT in 2021 to 0.89 lakh MT in 2030 as a result of yield increase from 1.8 ton/ha to 2.1 ton/ha and area expansion from 32.4 thousand ha to 42.3 thousand ha (

Table 8. Projections of production (supply) of ground nut and mustard.

Year	Supply of Grains (Lakh MT)
	Ground Nut	Mustard
2021	0.58	3.57
2030	0.89	4.40
2035	1.02	4.71
2040	1.15	5.02
2045	1.29	5.35
2050	1.43	5.65

Source: Authors’ estimation from ARIMA model.

). Similarly, mustard production will increase from 3.57 lakh MT in 2021 to 4.40 lakh MT in 2030 as a result of yield increase from 1.16 ton/ha to 1.2 ton/ha and area expansion from 309 thousand ha to 343 thousand in 2030. Ground nut production will gradually further rise to 1.43 lakh MT in 2050 due to area expansion to 61 thousand ha and yield expansion to 2.34 ton/ha. Also, mustard production will gradually further rise to 5.65 lakh MT in 2050 due to area expansion to 377 thousand ha and yield expansion to 1.5 ton/ha.

4.14. Projections of Surplus and Deficit of Ground Nut and Mustard Supply

We have converted production of whole ground nut and mustard into oil. There will be a surplus of ground nut oil of 0.30 lakh MT in 2030 and 0.58 lakh MT in 2050. There will be a deficit of mustard oil of 0.30 lakh MT in 2030 and 1.68 lakh MT in 2050 (

Table 9. Projections of surplus and deficit of supply of ground nut oil and mustard oil (lakh MT).

Year	Demand for Oil		Supply of Oil		Surplus (+)/Deficit(−)
	Ground Nut	Mustard	Ground Nut	Mustard	Ground Nut Oil	Mustard Oil
2021	0.14	1.48	0.29	1.43	0.15	−0.05
2030	0.14	2.06	0.45	1.76	0.30	−0.30
2035	0.15	2.48	0.51	1.88	0.36	−0.60
2040	0.14	2.97	0.57	2.01	0.43	−0.96
2045	0.14	3.46	0.64	2.14	0.50	−1.32
2050	0.13	3.94	0.71	2.26	0.58	−1.68

Source: Authors’ estimation from ARIMA model.

).

5. Sensitivity Analysis of Demand for and Supply of Pulses and Oil Crops

Alternative scenarios have been considered for sensitivity of pulses and oil crops demand and supply.

Demand side: projections: (i) Business as usual (BAU); (ii) Scenario 1: Crop demand decrease by 5%; (iii) Scenario 2: Crop demand increase by 5%.

Supply side: (i) BAU; (ii) Scenario 1: Supply short fall by 8% due to shocks (climate change, disease and pest, market vulnerability, other reason, etc.); (iii) Scenario 2: Supply short fall by 10% due to shocks (climate change, disease and pest, market vulnerability, other reason, etc.).

5.1. Sensitivity Analysis Results: Demand Side

The total demand for pulses is predicted to be 17.01 lakh MT in 2030 and 18.53 lakh MT in 2050 under alternative scenario 1. Projections further indicate that total deficit of pulses will be 11.47 lakh MT in 2030 and will decline to 8.29 lakh MT in 2050 due to an increase of supply 5.54 lakh ton in 2030 to 10.24 lakh ton in 2050 (

Table 10. Projections of total demand and net surplus of pulses under scenario 1.

Year	Quantity (Lakh MT)
	Business as Usual Scenario Demand	Alternative Scenario 1 Demand	Business as Usual Scenario Supply	Business as Usual Scenario Deficit	Alternative Scenario 1 Deficit
2021	16.1	15.30	3.99	−12.11	−11.31
2030	17.9	17.01	5.54	−12.36	−11.47
2035	18.4	17.48	6.76	−11.64	−10.72
2040	18.9	17.96	7.69	−11.21	−10.27
2045	19.3	18.34	9.2	−10.1	−9.14
2050	19.5	18.53	10.24	−9.26	−8.29

Source: Authors’ estimation.

). In scenario 2, both overall pulses demand and deficit slightly increase (

Table 11. Projections of total demand and net surplus of pulses under scenario 2.

Year	Quantity (Lakh MT)
	Business as Usual Scenario Demand	Alternative Scenario 2 Demand	Business as Usual Scenario Supply	Business as Usual Scenario Deficit	Alternative Scenario 2 Deficit
2021	16.1	16.91	3.99	−12.11	−12.92
2030	17.9	18.80	5.54	−12.36	−13.26
2035	18.4	19.32	6.76	−11.64	−12.56
2040	18.9	19.85	7.69	−11.21	−12.16
2045	19.3	20.27	9.2	−10.1	−11.07
2050	19.5	20.48	10.24	−9.26	−10.24

Source: Authors’ estimation.

).

Similar results are also found for oil demand (mustard and ground nut), which showed that, under scenario 1, total demand for oil will marginally decline to 2.09 lakh MT in 2030 and 3.87 lakh MT in 2050 compared to the existing BAU case (

Table 12. Projections of total demand and net surplus of oil crops under scenario 1.

Year	Quantity (Lakh MT)
	Business as Usual Scenario Demand	Alternative Scenario 1 Demand	Business as Usual Scenario Supply	Business as Usual Scenario Surplus	Alternative Scenario 1 Surplus
2021	1.62	1.54	3.99	2.37	2.45
2030	2.2	2.09	5.54	3.34	3.45
2035	2.63	2.50	6.76	4.13	4.26
2040	3.11	2.95	7.69	4.58	4.74
2045	3.6	3.42	9.2	5.6	5.78
2050	4.07	3.87	10.24	6.17	6.37

Source: Authors’ estimation.

). As a result, the overall surplus of oil will slightly rise to 3.45 lakh MT in 2030 and will decline to 6.37 lakh MT in 2050. In scenario 2 case, there will be a modest increase in national oil demand and a slight drop in the surplus (

Table 13. Projections of total demand and net surplus of oil crops under scenario 2.

Year	Quantity (Lakh MT)
	Business as Usual Scenario Demand	Alternative Scenario 2 Demand	Existing Scenario Supply	Business as Usual Scenario Surplus/Deficit	Surplus Under Alternative Scenario 2
2021	1.62	1.70	3.99	2.37	2.29
2030	2.2	2.31	5.54	3.34	3.23
2035	2.63	2.76	6.76	4.13	4.00
2040	3.11	3.27	7.69	4.58	4.42
2045	3.6	3.78	9.2	5.6	5.42
2050	4.07	4.27	10.24	6.17	5.97

Source: Authors’ estimation.

).

5.2. Sensitivity Analysis of Supply of Pulses and Oil Crops

According to projections, total supply of pulses will decline to 5.10 lakh MT in 2030 and 9.42 lakh MT in 2050 under scenario 1 (

Table 14. Sensitivity analysis of projections of total supply and net surplus/deficits of pulses under alternative scenarios.

Year	Quantity (Lakh MT)
	Business as Usual Scenario			Alternative Scenario 1		Alternative Scenario 2
	Demand	Supply	Surplus	Supply	Surplus (+)/Deficit (−)	Supply	Surplus (+)/Deficit (−)
2021	16.1	3.99	−12.11	3.67	−12.43	3.51	−12.59
2030	17.9	5.54	−12.36	5.10	−12.80	4.88	−13.02
2035	18.4	6.76	−11.64	6.22	−12.18	5.95	−12.45
2040	18.9	7.69	−11.21	7.07	−11.83	6.77	−12.13
2045	19.3	9.2	−10.1	8.46	−10.84	8.10	−11.20
2050	19.5	10.24	−9.26	9.42	−10.08	9.01	−10.49

Source: Authors’ estimation.

). The findings appear to be plausible in that country’s pulses supply will face unpredictable shocks from climate change, pest and diseases and market fluctuations, etc. in the coming decades. The results of scenario 2 show that pulses supply will decline further slightly to 4.88 lakh MT in 2030 and will have a deficit of 13.02 lakh MT while total supply will be 9.01 lakh MT in 2050 with a deficit of 10.49 lakh MT. In the long-term dissemination and adoption of new pulses technology, which, however, could generate a surplus of pulses supply.

Likewise, total oil supply (mustard and ground nut) will marginally decrease from BAU levels to 5.10 lakh MT in 2030 and 9.42 lakh MT in 2050 under scenario 1 (

Table 15. Sensitivity analysis of projections of total supply and net surplus/deficits of oil crops under alternative scenarios.

Year	Quantity (Lakh MT)
	Business as Usual Scenario			Alternative Scenario 1		Alternative Scenario 2
	Demand	Supply	Surplus	Supply	Surplus (+)/Deficit (−)	Supply	Surplus (+)/Deficit (−)
2021	1.62	3.99	2.37	3.67	2.05	3.51	1.89
2030	2.2	5.54	3.34	5.10	2.90	4.88	2.68
2035	2.63	6.76	4.13	6.22	3.59	5.95	3.32
2040	3.11	7.69	4.58	7.07	3.96	6.77	3.66
2045	3.6	9.2	5.6	8.46	4.86	8.10	4.50
2050	4.07	10.24	6.17	9.42	5.35	9.01	4.94

Source: Authors’ estimation.

). The results seem to be plausible due to the fact that the country would face unpredictable shocks of climate change, pest and diseases outbreak and unstable market, etc. in scenarios 1 and 2. The results of scenario 2 indicate that oil supply will decline further slightly, which would lead to a consequent of deficit.

6. Discussions

The present study estimated the demand for pulses and oil crops by 2030 and 2050. The ARIMA model gave consistent estimates when validating the actual result in 2021 with the projected result. It was found that the consumption demand for cereals decreasing over the years while demand for pulses and oil crop have increasing trend over the years. This is consistent with findings of other studies that the cereal consumption demand would decrease by 2030 and demand for non-cereals (pulses, oil, vegetables, livestock products and fish) will increase. It was revealed from the study of others that the world cereal equivalent (CE) food demand is projected to be around 10,094 million tons in 2030 and 14,886 million tons in 2050, while its production is projected to be 10,120 million tons in 2030 and 15,970 million tons in 2050, having a marginal surplus. India and China are capturing a large share of global food demand. The developing country will demand more animal origin foods due to income growth in the future. The growth rate of world demand for cereals will decline till 2050 [21].

We projected the supply of pulses and oil by 2030 and 2050 through estimating the projection of crop area and yield. It was found that there will be an increasing trend in the production of a few pulses crops and oil crops due to change in yield. This is consistent with the validated result of actual yield of 2021 with the projected yield. Bangladesh already developed some new high-yielding pulses crops and mustard oil crops, and farmers are cultivating these. Sensitivity analysis was carried out on crop supply projection (described above) and confirms that the ARIMA model gave consistent results. The ARIMA model could be used for projections of demand for and supply of food nationally, regionally or globally.

7. Conclusions

The ARIMA model gave plausible results when compared with the actual and BAU result in 2021. We compared our projections with those of other authors and found that our projections were quite plausible from a realistic point of view.

Cereals is a major part of the calorie intake in Bangladesh, but their share in total calorie supply has decreased over the years due to diversification of food consumption over the years. It revealed that cereal consumption share is decreasing in Bangladesh as well as in countries with similar consumption and economic growth patterns in Asia [23,31,42].

The projections showed that the total demand for pulses in 2030 will be 17.9 lakh MT, and further increasing trend continues over the years. In the year 2050, it will increase to 19.5 lakh MT. It was found that supply of pulses will have a deficit amounting to 12.88 lakh MT in 2030 and 9 lakh MT in 2050. The shortage of pulses will gradually decline due to productivity enhancement and innovation over the years. Supply of ground nut oil will have a surplus amounting to 0.30 lakh MT in 2030 and 0.58 lakh MT in 2050, whereas the supply of mustard oil will have a deficit with the amount 0.30 lakh MT in 2030 and 1.68 lakh MT in 2050.

The sensitivity analysis confirmed that the ARIMA models gave consistent results with the increased demand for pulses and oil crops. The projected surplus could be deficit. On the other hand, with a decrease in supply (as a result of decrease in yield) due to unpredictable shocks like negative impacts of climate change, soil degradation, increased soil salinity, scarcity of irrigation water, etc., the deficit of pulses and oil crops would further increase in the coming decades. In tackling future challenges due to climate change and market distortions, agricultural transformation is needed. Climate resilient technologies of pulses and oil crops must be made available and disseminated for rapid adoption in the long term.

Research Limitations: The study used time series data for modeling. The national long term time series data available from the Bureau of Statistics may have inconsistencies in some cases. Inconsistent time series data in some cases created problems for model validation. Treatments: Done using statistical tools to clean data and generate missing information. Also, used research results and data from the National Agricultural Research Institutes for cross-checking. Also, we tried alternative models.

Future research directions and implications for policymakers: The present study pinpointed the future research directions and implications for policymakers. The scientists should generate climate-resilient high-yielding pulses and oil crops and management practices so that deficits of these crops could be meeting up. For the policymakers, it generated a clear message that supportive actionable policy should be designed to enable the researcher with increased investments to develop new high-yielding and resilient pulses and oil crops varieties. Such supportive government policy and substantial investment in research and extension will build a long-term development pathway toward promoting climate resilient technology and innovation for enhancing pulses and oil crop productivity to reduce deficits of these food items during the 2030s and 2050s. The government should prepare a long-term plan for technology development and transformation of Bangladesh agriculture in the 2030s and 2050s, giving a priority to pulses and oil crops. The Ministry of Agriculture of Bangladesh already started working on this. This policy could be aligned with the 8 Five-Year plan of government [8] and Bangladesh Perspective Plan 2041 [11].