RSS | Register/註冊 | Log in/登入
Site search:
Home>FFTC Document Database>Extension Bulletins>Projecting the World Food Supply and Demand Using a Long-Term Dynamic Simulator
facebook分享
Projecting the World Food Supply and Demand Using a Long-Term Dynamic Simulator
Osamu Koyama
Japan International Research Center for Agricultural Sciences (JIRCAS)
1-2 Ohwashi, Tsukuba, Ibaraki, 305-8686 Japan, 2000-11-01

Abstract

Various institutions are currently making future projections of the world food supply and demand. Many of them use medium-term econometric models, which are not necessarily suitable for reproducing long-term dynamic changes. A new simulator, which links together the method of econometric models and the concept of system dynamics, was developed for this reason. It is flexible enough to deal with unconventional issues that require longer perspectives. Although the results of the simulation show the continuation of the past trend like the other projections, the simulation results clearly indicate the magnitude of key issues in the coming century, such as feed requirements, land expansion, global trade and so on.

Introduction

Population explosion still goes on. More than 70 million people will be added to the world population every year between 1995 and 2020. Although the earth provides an enormous amount of food for 6 billion people, nearly 800 million people currently suffer from malnutrition. There is little hope to fulfill the commitment made by the world food summit in 1996 that the number of undernourished population should be halved by 2015 (FAO 1999b). The food requirement of the world in 2050 is estimated to be twice as much as now. A serious concern about the scarcity of land and water resources have emerged in tandem with various kinds of environmental problems.

The issue of world food, however, is not simply a matter of production of food. It includes issues of distribution and access. The amount of food required also varies in relation to economic conditions and other factors. In order to understand the complexity of world food problems, various large-scale economic models have been developed in many institutes, and the results of future projections have been published. However, as the world food problems become more complicated and as people's concerns move towards long-term sustainability issues, new analytical tools suitable for these problems with a longer perspective must be developed.

This paper firstly reviews past efforts to develop food supply and demand models, and compares the characteristics and the results of the existing models. It also demonstrates an attempt to overcome the limitations of existing medium-term models. Using the new model developed, several projections targeting the year 2020 are made. The projection results are discussed briefly, concentrating on the main issues. Finally, at the end of the paper, a suggestion is made with regard to the future direction of long-term analyses of the world food supply and demand.

Econometric Models for Food Supply-Demand Analysis

Quantitative analyses for the future situation of the world food supply began along with improvements in international statistics for the food sector. In the 1960s, the Food and Agriculture Organization of the United Nations (FAO) made public its first "agricultural commodity projection" based on single-equation model methodology (FAO 1967). In the projection, various types of mathematical functions were carefully selected for each country and commodity in order to reproduce patterns of past trends. This single-equation method, however, did not adjust for gaps in supply and demand for the world as a whole.

Since the world food crisis in 1973, research activities in this field have been encouraged in many institutions. The U.S. Department of Agriculture (USDA) made several attempts to build multi-equation models in which supply-demand gaps were cleared automatically by a change in prices (USDA 1985). However, the form of the equations was rather simple and the structure of the models was comparatively static, whereby a single point of the future was compared with the starting point. Thus, the results of the projections were not very persuasive. At the same time, an ambitious attempt was made by the Ministry of Agriculture, Forestry and Fisheries in Japan (MAFF). A model named World Food Model had a dynamic nature. The passage of time was considered, and the model was equipped with elaborate non-linear functions (MAFF 1974).

Following these attempts, several so-called policy simulation models were developed in the 1980s. In step with the progress of computer technology, econometric models were made capable of reproducing fairly complicated realities. Research results based on these models were used as reliable guides for reforming trade and domestic support policies during this period. The model developed by the Organization of Economic Cooperation and Development (OECD) is one of these policy simulation models.

In order to assess the mutual impact among various economic sectors, computable general equilibrium (CGE) models began to be used. The GTAP (Global Trade Analysis Project) model is a typical model of this kind. It is widely used for agricultural trade issues (Hartel ed. 1997). Since the commodity classification is not as detailed as in the partial equilibrium models, it is not a very powerful tool for future projections. However, it is true that the analysis within a single food sector would not be very meaningful, particularly in developed countries where the agricultural sector occupies only a small part of the whole economy.

For long-term dynamic assessment, the method of system dynamics attracted people's attention when a controversial report of the Club of Rome titled "The Limits to Growth" was issued (Meadows et al. 1972). Unlike conventional econometric models, this type of model is capable of handling various dimensions of data. The usefulness of this model is realized in the case of environmental assessments, where not enough statistical data are available.

Table 1(1061) is a list of presently active models which deal with food supply and demand analyses. As every method, including simple single-equation models, has merits and demerits, the selection of the methodology should depend totally on the purpose and the target of the study.

Comparison of Existing Projection Results

Several institutes launch projection results of world food supply and demand periodically. Although the results of projections are basically the outcome of econometric models, most of them also reflect the opinions of regional or commodity experts. Therefore, the final figures in the publication are not associated directly with the models. In addition, each model uses a statistical dataset with a different classification and definitions. To compare these results, therefore, is not an easy task.

The primary factors that generate the differences in theory are the assumptions, such as GDP growth and population, given to the models. However, most models use similar assumptions in reality. The structure of the models, such as the selection of explanatory variables and the size of parameters, is also a source of variation. Actual parameters applied are known to vary substantially, regardless of the fact that most of the parameters are estimated on the basis of past data and on similar econometric methods. Difference in equation forms is an influential factor concerning the long-term dynamic projections.

In Table 2(1060), annual growth rates of various projection results for world cereal consumption are compared. The results vary in a range from 1.0% to 1.6% p.a.. Although the figures are alike, accumulated growth for 20 years would produce significant differences. Fig. 1(1324) expresses these results in a linear form. There is a tendency that more recent projections have lower growth rates, reflecting the downward revisions of population and GDP assumptions. All though the projections have similar trends in world totals, the numbers for individual countries sometimes show a completely opposite trend. As models are based on past experience, no model is competent to tell the future if any unforeseen events such as the Asian financial crisis or the collapse of the USSR should occur.

The common limitation of these econometric models is that they are designed for medium-term projections. In these models, the basic economic structure is supposed to remain unchanged for the projection period of 5 to 7 years. As the problems that require long perspective have emerged, the need for analyses over a longer period of time increases. For example, constraints of natural resources, global climate changes and determination of research priorities are important issues that require long-term projections. Structural changes drive the long-term balance of world food supply and demand. The growth rate of world cereal yield per hectare, for example, has declined to nearly half of previous decades in Asia, while per capita consumption of livestock products has increased threefold in the past 20 years in Asia. Perhaps the production structure in many developing countries in the year 2020 will be quite different from the one prevailing now.

New Approach for Dynamic Long-Term Analyses

To overcome the limitations of current medium-term models, a simple framework for a long-term simulation model was designed. It is basically a price equilibrium food supply-demand model like many other models. The uniqueness of the model is the flexible application of equations and parameters. Parameters and equations are totally synthetically given for each calculation period of each region or commodity, so that long-term structural changes can easily be reflected. An example of basic data flow of the simulator is shown in Fig. 2(1195). Commodity and regional classifications can be identified in Table 3(1129), where the production data are also shown for reference.

The following equations are used for the model. Most equations have a form of constant elasticity and constant growth rate for the convenience of flexible change of parameters.

Population assumptions are derived from the medium variant of the UN population estimate (United Nations 1996). Income growth is represented by GDP growth, and compiled from a World Bank report (World Bank 1999) and other sources. The GDP growth rates are kept constant during the projection period, as no additional information was available. Income and price elasticities are formulated, with those of individual countries taken from the FAO estimates (FAO 1993) and other literatures.

Increase of yield per hectare is calculated from the historical trend. As constant growth rates sometimes give unrealistic numbers, some yield levels in 2020 are thought to be too high. In this case, some adjustments were made. Changes in the world average yield of cereals are shown in Fig. 3(1065), together with population growth. It is noted that assumptions of yield follow a very similar trend to those of population.

Food demand FOi,j,t = FOi,j,t-1 (Yj,t/Yj,t-1)bi,j,t (Pi,t/Pi,t-1)Ci,j,t (Nj,t/Nj,t-1)
Where, Y: per capita GDP, P: equilibrium price index, N: population,
b: income elasticity, c: price elasticity, i: commodity, j: region.
Feed demand EFi,j,t = FEi,j,t-1 (Pi,t/Pi,t-1)Ci,j,t (Lj,t/Lj,t-1)
Where, L: production of livestock products.
Other demand ODi,j,t = ai,j,t OCi,j,t-1
Where, a: trend factor in the form of annual growth rate.
Harvested area HAi,j,t = ai,j,t HAi,j,t-1 (Pi,t-1/Pi,t-2)ci,j,t
Yield per ha YHi,j,t = ai,j,t YHi,j,t-1
Production Oi,j,t = HAi,j,t YHi,j,t
Net trade Ti,j,t = Oi,j,t - FOi,j,t -FEi,j,t - ODi,j,t
Market clearing ?Ti,t = 0

Implication of the Simulation Results

Using this new simulator, three projections were made. One is the so-called baseline projection, and the other two are scenario studies. The baseline results indicate the continuation of recent trends into the future. Consumption of livestock products will continue to increase rapidly, especially in developing regions. As shown in Fig. 4(1060), total per capita consumption of livestock in 2020 is projected to be 2.2 times larger than the base period (1993/97 average) in East and Southeast Asia, excluding Japan, although high income elasticity in the region was assumed to drop gradually in the future. This rapid growth corresponds to an average annual increase rate of 3.6% for the projection period, which is less than the assumption of GDP growth (5.8% p.a.) for the region.

Reflecting the increase in livestock production, feed requirements also expand dramatically. They will more than double in East and Southeast Asia during the next 20 years, and a large amount of cereals and oil meals will be imported from outside the region, according to the projection. The projected net trade situation for each region is shown in Fig. 5(1124). One of the controversial indicators in this field is the feed-livestock coefficient. Although feed efficiency was kept nearly constant in the baseline scenario, it will be very influential if the balance between improvements in feeding technology and the prevalence of intensive commercial production changes in the future.

Expansion of the harvested area is also a matter of concern. The production of cereals and oil crops in 2020 will require 70 million ha more land than the land currently used. The increase in land requirements comes mainly from the oil crop sector as in past years ( Fig. 6(1147)). However, a large-scale harvested area is essential for future food production in Sub-Saharan Africa, where the increase in yield per hectare is expected to be smaller than in the other regions. Harvested area will be increased, partly by the exploitation of new arable land, and partly by raising crop intensity.

Starting from the results of the baseline projection, two scenario studies can be made. One is the lower GDP scenario, and the other is the lower yield scenario. The GDP assumption used for the baseline scenario was rather optimistic, partly because it becomes gradually higher on a per capita basis due to the gradual decline in population growth, and partly because it is originally higher compared with recent economic performances. Thus, it is not an unrealistic assumption that the rate would be halved. Lowering yield assumption might be an exaggerated scenario, as the baseline assumptions of yield increase were adjusted below the past trend for many regions. However, it is perhaps worth doing this simulation considering the people's concern for global environmental degradation, which may force a further decline in yield growth.

Results of simulations in terms of the world price index for cereals are shown in Fig. 7(1511). As seen in the chart, a slower yield increase would raise world prices substantially. On the other hand, lower economic growth has a negative impact on prices. As far as the amount of consumption is concerned, both scenarios have a negative impact, especially in developing regions where the income effect is generally large ( Fig. 8(1116)). Thus, it is suggested that a higher increase in yields can reduce the world food price, but a declining world food price does not necessarily mean an increase in consumption. However, it is noted that yield increase and income are closely related in many developing regions.

What we learn from this simulation can be summarized as follows. Firstly, yield increase of crops is the least known and most influential factor in the food supply and demand balance in the distant future. Thirty years ago, not many crop scientists could foresee that the current level of yields would be realized. New technology, including biotechnology, has huge potential for yield increases. But at the same time, environmental constraints may limit even the current level of harvest. In addition, technical innovation is strongly correlated to the level of investment, and thus the level of relative prices. Although agricultural investment has not been very active, due to the relatively low prices of agricultural products, a higher price of food in the future would surely attract investment to this area.

Secondly, the future situation of food supply and demand will be determined largely by growth in income. Population growth, which has been the major factor for the increase in demand, will be moderated in the future. The average growth rate of the world population between 2015 and 2020 is estimated at 1.0% p.a.. Shifts in dietary patterns, which include the diversification of traditional diets in developing regions, and health and environmental concerns in developed countries, will be a major component of projected demand from now on. This also means that future projections must consider the situation of food processing.

Thirdly, it should be noted that trade among the regions will play a greater role in the future. Since the model does not show the amount of gross trade, and no policy variable is considered in the model, not much can be said about trade. However, the net trade among the regions is projected to increase rapidly. As the basic factors such as land availability or population increase vary widely over the regions, regional gaps would surely enlarge without global trade. Trade, however, sometimes has a negative effect on local production. The balance between the global division of labor and effective utilization of local resources will become a more and more difficult task for policy makers.

Conclusion

This paper emphasizes the importance of dynamic structural changes in the long-term food supply and demand analysis. Many current medium-term models are not capable of reproducing such changes in their projections. A new approach, which links the methods of econometric models and the concept of system dynamics, could be a powerful tool in this regard. Although the results of the simulation show the continuation of past trends, the magnitude of emerging key issues in the coming century, such as feed requirements, land expansion, global trade and so on, are clearly indicated by the results.

In consideration of the additional food requirement in the future, gains in agricultural productivity must continue to be pursued. The market will further promote relocation and specialization of production. The effectiveness of medium-term econometric models for policy simulations is undeniable. However, additional factors must be considered concerning long-term policies. The future is an unprecedented world where the limitations of the earth must be recognized. Issues of sustainability become a priority goal for the food supply and demand sector. The dynamic simulator developed in this paper also has the potential to deal with such issues. By adding new functions, the simulator can reproduce the long-term dynamics in world food sector more realistically, and show us the way to a sustainable future.

References

  • FAO. 1967. Agricultural commodities-Projections for 1975 and 1985. Vol. 1. Rome, Italy.
  • FAO. 1993. The World Food Model - Model Specifications ESC/M/93/1. Rome, Italy.
  • FAO. 1999a. FAOSTAT 98, FAO Statistical Database. FAO, Rome, Italy.
  • FAO. 1999b. The State of Food Insecurity in the World, 1999. FAO, Rome, Italy.
  • FAO. 2000. Agriculture: Towards 2015/30, Technical Interim Report. FAO, Rome, Italy.
  • FAPRI. 2000. FAPRI 2000 World Agricultural Outlook. Staff Report 2-00, Food and Agricultural Policy Research Institute, Iowa State Univ. and Univ. of Missouri-Columbia, Ames Iowa, U.S.A.
  • Hertel, T.W., (Ed.). 1997. Global Trade Analysis - Modeling and Applications. Cambridge Univ. Press, New York, U.S.A.
  • Islam N. (Ed.). 1995. Population and Food in the Early Twenty-First Century. International Food Policy Research Institute, Washington D.C., U.S.A.
  • Koyama, O. 1997a. Projections of World Food Supply and Demand for 2020. Farming Japan 31-3: 19-27.
  • Koyama, O. 1997b. Projecting Food Balances in China using the World Food Model. Paper presented at the post conference "China's Food Economy in the 21st Century" of AAEA Annual Meeting 1997 in Toronto.
  • MAFF. 1974. Supply/Demand Perspectives by World Food Supply/Demand Model. Nourintoukeikyoukai, Tokyo, Japan. (In Japanese).
  • MAFF. 1995. Result of Projections by the World Food Supply Demand Model - A Trial Run. MAFF, Tokyo, Japan. (In Japanese).
  • Meadows, D.H., D.L. Meadows, J. Randers and W.W. Behrens III. 1972. The Limits to Growth. Universe Books, New York, U.S.A.
  • Meadows, D.H., D.L. Meadows and J. Randers. 1992. Beyond the Growth. Chelsea Green Publishing Company, Vermont, U.S.A.
  • Mitchell, D.O., M.D. Ingco and R.C. Duncan. 1997. The World Food Outlook. Cambridge University Press, New York, U.S.A.
  • OECD. 2000. OECD Agricultural Outlook 2000-2005, OECD, Paris, France.
  • Oga, K. and K. Yanagishima. 1996. International Food and Agricultural Policy Simulation Model - User guide. JIRCAS Working Report No. 1, JIRCAS, Tsukuba, Japan.
  • Pinstrup-Andersen, P., R. Pandya-Lorch and M. Rosegrant. 1999. World Food Prospects: Critical Issues for the Early Twenty-first Century. International Food Policy Research Institute, Washington D.C., U.S.A.
  • Rosegrant, M.W., M. Agcaoili-Sombilla and N.D. Perez. 1995. Global Food Projections to 2020 - Implications for Investment. International Food Policy Research Institute, Washington D.C., U.S.A.
  • United Nations. 1996. World Population Prospects - the 1996 revision. United Nations, New York, U.S.A.
  • USDA. 1978. Alternative Future for World Food in 1985. Foreign Agricultural Economic Report No. 146, Economics, Statistics and Cooperatives Service, U.S. Department of Agriculture, Washington D.C., U.S.A.
  • USDA. 1998. International Agricultural Baseline Projections to 2007. Agricultural Economic Report No. 767, Economic Research Service, U.S. Department of Agriculture, Washington D.C., U.S.A.
  • World Bank. 1999. Global Economic Prospects and the Developing Countries 1998/99. The World Bank, Washington D.C., U.S.A.

Index of Images

Figure 1 Various Projection Results for World Cereal Consumptiion.<BR>

Figure 1 Various Projection Results for World Cereal Consumptiion.

Source:Seethelistofreferencesattheendofthepaper

Figure 2 Structure of Global Food Balance Simulator (Cereals)

Figure 2 Structure of Global Food Balance Simulator (Cereals)

Figure 3 Assumptions for Growth Rates of Population and Cereal Yield.

Figure 3 Assumptions for Growth Rates of Population and Cereal Yield.

Source:PopulationassumptionsaretheUNestimates(UN1996)

Figure 4 Per Capita Consumption of Livestock Products, Actual and Projected

Figure 4 Per Capita Consumption of Livestock Products, Actual and Projected

Figure 5 Projection of Net Export of Cereals

Figure 5 Projection of Net Export of Cereals

Figure 6 Projection of Crop Harvested Area, World Total

Figure 6 Projection of Crop Harvested Area, World Total

Figure 7 Simulation Results of World Cereal Price

Figure 7 Simulation Results of World Cereal Price

Figure 8 Per Capita Cereal Consumption under Different Scenarios

Figure 8 Per Capita Cereal Consumption under Different Scenarios

Table 1 A List of World Food Supply and Demand Models

Table 1 A List of World Food Supply and Demand Models

Table 2 Various Projection Results and past Trends for World Cereal Consumption

Table 2 Various Projection Results and past Trends for World Cereal Consumption

Table 3 Commodity and Regional Coverage and Productionin 1993/97 (Average, Million Tons)

Table 3 Commodity and Regional Coverage and Productionin 1993/97 (Average, Million Tons)

Download the PDF. of this document(665), 236,000 bytes (230 KB).