ASSESSING LONG-TERM TRENDS IN RESOURCE CONSUMPTION The role of models and scenario analysis CONTENTS Paragraph Page I. INTRODUCTION.................................... 1 - 5 1 - 2 II. ASSESSING LONG-TERM TRENDS THROUGH MODELS ....... 6 - 41 2 -16 A. Global energy consumption perspectives .... 6 - 18 2 - 7 B. Motor vehicles and road transportation ....19 - 24 7 -10 C. World demand for metals and minerals ..... 25 - 34 10 -13 D. Agriculture and food security ........... 35 - 41 14 -16 III. METHODOLOGICAL APPROACHES TO MODELING ........ 42 - 74 16 -23 A. Structure of models ..................... 43 - 63 16 -21 B. Assumptions and perspectives of models .. 64 - 69 21 -22 C. Limits and limitations .................. 70 - 74 22 -23 IV. CONCLUSIONS.................................. 75 - 78 23 I. INTRODUCTION 1. Recently, there has been growing recognition of the need for an integrated approach to inter-locking social, economic and environmental issues. Such an approach is especially useful to inform international debate on sustainable development issues which should take account of differences among various world regions in socio-economic conditions and priorities. A typical example is global climate change. Over the past few years, multi-disciplinary teams of scientists around the world have been working in close collaboration on various aspects of climate change in order to analyze the causes, mechanisms and impacts of climate change in an integrated fashion. 2. Work by the Commission on Sustainable Development (CSD) on changing unsustainable consumption patterns has provided further momentum for the ongoing efforts at integrated assessment, in particular integrated long-term assessment. The deliberations at the CSD have raised a key question - on today's consumption trends, will our planet be cleaner and healthier in future, and will future consumption patterns satisfy needs for all while protecting the environment? In trying to answer this question, policy-makers find themselves faced with uncertainties and risks. While expert judgments may provide some insights, a more systematic and consistent effort is clearly required in order to answer the question with an adequate confidence level. 3. In this regard, recent experience in climate change research suggests that modeling and scenario analysis can be meaningful tools in helping advance systematic enquiries. Models are structured frameworks for organizing knowledge and enquiries into complex relationships and processes. For producing assessment of sustainable development trajectories, there are several advantages in using models, including i) dynamic analysis of the interactions between the components of the system under consideration, following the Pressure-State-Impact-Response (P-S-I-R) mechanism; 1/ ii) the development of projections and early warnings: models allow for extending analysis in time and space and implicitly show the dynamics of environmental processes in relation to social and economic processes and thus may lead to better informed priority setting; and iii) the translation of uncertainties into risk analysis in order to assist decision-makers in decision-making under conditions of uncertainty. Making uncertainties explicit in the model projections facilitates the communication of the results to decision-makers and helps them in identifying major gaps in knowledge and in anticipating the unexpected. 2/ 4. Scenarios are descriptive images of the future, which are not meant to be predictions of the future but are hypothetical sequences of events intended to focus attention on causal processes and decision points. Scenarios are always developed for a particular purpose in a particular context and are never value-free because they reflect the visions of the scenario authors. The primary objective of scenarios is to offer a structured basis for debate about key issues, opportunities and threats with respect to sustainable development, taking into account different regional and sectoral perspectives and interests. While scenarios can be generated by models, they may often be produced by experts using qualitative judgments. 5. As a contribution to the upcoming work on long-term trend assessment carried out in preparation for the 1997 Special Session of the General Assembly, this background report serves as an introduction to models and scenarios in trend assessment. The report is not intended to provide in-depth technical analysis. Rather, it summarizes selected results so as to give some ideas of what models can do. This is followed up by a brief discussion of the methodologies of models and of their inputs and assumptions. For the sake of clarity, a glossary is attached to explain some technical terms. 3/ II. ASSESSING LONG-TERM TRENDS THROUGH MODELS A. Global energy consumption perspectives 6. World energy demand increased at an average annual rate of 2.5 per cent between 1970 and 1993. Commercial energy consumption grew faster in developing countries than in either developed countries or economies in transition. In 1970, developing countries consumed 604 million tonnes of oil equivalent (toe); by 1993, consumption had risen to 2,317 million toes. In developed market economies, demand increased from 2,888 million toes in 1970 to 4,124 million toes in 1993. Total world consumption of commercial energy reached some 7,800 million toes in 1993. 7. Because of its essential role in our economy and its impact on the environment, energy has been the subject of many modelling efforts. For instance, under the Energy Modeling Forum convened at Stanford University, models with various time horizons were used to explore alternative energy demand and CO2 emission scenarios. Most of these models were energy-sector equilibrium models, although some were aggregate economic equilibrium models. In addition, some integrated assessment models (IAM) contained sub-models addressing energy demand. These IAMs were concentrated on - though not exclusively confined to - capturing the system impact of climate change. Below we present the scenarios generated by some selected models. 1. The WEC/IIASA perspective study 8. Building on their earlier efforts, the World Energy Council (WEC) and the International Institute for Applied Systems Analysis (IIASA) recently developed six scenarios of energy demand in 2050 and 2100, using 1990 as the base year. 4/ The analysis was based on the use of several models, and the six scenarios were grouped into three cases. Case A depicts a high-growth scenario; Case B represents a more modest but more realistic outcome; whereas Case C is "ecologically driven" and assumes unprecedented international cooperation. Case A also includes three scenarios addressing key developments in energy supply, depending on whether oil and gas, coal or renewable will be the dominant source of supply. A summary of the results of the three cases is presented below. 9. As shown in Table I.1 below, the three cases share the same assumption about world population. They also envisage economic growth, though growth rates are higher in Case A 5/. The three cases also provide for continuing progress in energy efficiency, as expressed in declines in energy intensity. While improvement rates in Cases A and B are primarily due to economic growth, Case C incorporates progress achieved through demand side management and economic instrument, including substantial increases in energy prices and taxes. 6/ As a result, there are considerable variations in primary energy demand, with the "ecologically driven" case (Case C) depicting a scenario of considerably lower demand. In terms of energy supply and technology change, Case C also differs from Cases A and B in that it assumes low availability of fossil fuels and high technology dynamics of non-fossil fuels. In addition, Case C assumes the existence of a CO2 constraint and advocates environmental taxes. Carbon emissions are consequently much lower under Case C. 10. Here it may be interesting to compare the WEC/IIASA CO2 emissions scenarios with those of the IPCC. In 1992, the Working Group II of the Intergovernmental Panel on Climate Change (IPCC) using global models developed a range of scenarios (IS92 a-f) of future CO2 and other greenhouse gas emissions based on assumptions concerning population and economic growth, land use, technological changes, energy availability and fuel mix. Unlike the scenario cases developed by the WEC/IIASA study, the IPCC scenarios were Business-as-Usual scenarios, assuming no mitigation policies and no significant technological advances. However, there were considerable variations in the exogenous assumptions of population and economic growth underlying the six scenarios. By the year 2100, CO2 emissions could range from 6 billion tonnes of carbon per year (IS92c), with low population growth, to as much as 36 billion tonnes of carbon per year (IS92e), with high population and economic growth. 7/ Table I.1: A summary of three cases in 2050 and 2100 (with 1990 as the base year) ----------------------------------------------------------------------------- Case ---------------------------------------------------- A B C High Growth Middle Course Ecologically Driven Population a/ 2050 10.1 10.1 10.1 2100 11.7 11.7 11.7 GWP b/ 2050 100 75 75 2100 300 200 220 Energy intensity improvement c/ medium low high 1990-2050 -1.0 -0.7 -1.4 1990-2100 -1.0 -0.8 -1.5 Primary energy demand /d 2050 25 20 14 2100 45 35 21 Resource availability Fossil high medium low Non-fossil high medium high Technology costs Fossil low medium high Non-fossil low medium low Technology dynamics Fossil high medium medium Non-fossil high medium high CO2 emission constraint no no yes Carbon emissions e/ 2050 9-15 10 5 2100 7-22 14 2 Environmental taxes no no yes ----------------------------------------------------------------------------- Source: WEC/IIASA. Note: a) billions; b) trillions; c) primary energy/GDP, per cent/yr; d)billion tonnes of oil equivalent; e) billion tonnes of carbon. 11. In addition to CO2 emissions, the WEC/IIASA study also calculated scenarios of sulfur deposition in Europe and South and East Asia, using the Regional Acidification INformation and Simulation model (RAINS) developed at IIASA. Under Case A, a coal-intensive scenario with no abatement, sulfur emissions in Europe would increase by some 50 per cent over the next 30 years, and sulfur deposition would exceed 16 g S/m2 per year in large areas of Central, Western and Northern Europe. 8/ In the rapidly growing economies in Asia, the coal-intensive scenario suggests that if no abatement measure is implemented, SO2 emissions in South and East Asia would triple by 2020 and deposition would exceed critical loads in most of the ecosystems in the region. In some areas, the sulfur deposition would exceed the critical loads of the ecosystems by a factor of ten, with serious implications for food crops. 9/ 12. Under the policy-intensive ecologically-driven scenario (Case C), however, sulfur emissions would be much lower, as energy demand would be lower, and the supply structure would shift to less sulfur containing fossil fuels. In Asia, for instance, increase in unabated sulfur emissions in the next two decades would be kept below a factor of two, as compared with the tripling under the coal-intensive scenario. 2. The TARGETS/IMAGE energy model 13. Integrated assessment models usually contain sub-models on energy. The TARGETS/IMAGE energy sub-model developed at the National Institute of Public Health and Environmental Protection (RIVM), the Netherlands is such an example. 10/The energy model is currently a globally aggregated system dynamics simulation model. It has five sub-models: energy demand; supply of solid, liquid and gaseous fuels and electric power generation. The model includes the following features: (a) activity-related demand for heat (in five sectors) and electricity, incorporating structural economic change; (b) autonomous and price-induced change in energy intensity (energy conservation); (c) exploration and exploitation dynamics of fossil fuels, including depletion and learning dynamics; (d) price-based substitution of biofuels which are assumed to be subject to learning as well as depletion dynamics; (e) electric power generation in thermal power plants, with a non- thermal alternative (nuclear, solar) penetrating the market based on relative costs and learning. 14. The TARGETS/IMAGE energy model has been calibrated for the world from 1900 to 1990 and for the USA from 1950 to 1990. All parameter values are supposed to be entered for the period 1990-2100. Currently, the RIVM group is in the process of analyzing a number of scenarios, including the Business-As- Usual IPCC IS92a scenario, the Reference scenario of the World Energy Council, the Sustained Growth (SG) scenario by Shell and the alternative Dematerialization (DM) Scenario by Shell and the Fossil-Free Energy Scenario developed by the Stockholm Environment Institute. Preliminary findings obtained from simulating some of the scenarios suggest that while scenarios can individually depict an internally consistent picture of global energy supply, their perspective-driven assumptions about key variables may lead to different results. For instance, the most evident difference between the IS92a and Sustained Growth scenarios is the reliance on fossil fuel (notably coal) after 2050 in IS92a and a switch to renewables in the Sustained Growth scenario. This difference is mainly caused by a number of perspective-based key assumptions with respect to renewable energy technologies and power generation technologies. 11/ 3. Renewables-intensive global energy scenario 15. In contrast to fossil-fuel based scenarios, exponents of renewable sources of energy have constructed a renewables-intensive global energy scenario, with 1985 as the base year. 12/ Using the population and GDP growth assumptions adopted in IPCC high economic growth scenario, they derived renewables-intensive scenarios of future electricity generation and direct fuel use. 16. In 1985, the world generated 9,239 TWh of electricity from commercial fuels, of which hydropower generated 1,880.1 TWh of electricity, while nuclear power generated 1,399.3 TWH of electricity. Geothermal sources produced a much smaller amount of electricity - 15.0 TWh. Expecting global electricity production to more than double by 2025 and more than triple by 2050 as against the base year 1985, the scenario producers projected that the share of renewable energy in power generation would increase from 20 per cent in 1985 (mostly hydroelectric power) to about 60 per cent in 2025, with roughly comparable contributions from hydropower, intermittent renewables (wind and direct solar power), and biomass. 13/ Figure 1: CO2 emissions from fossil fuel use (million tonnes of carbon) Comparison of developing countries, industrialized countries, former USSR and Eastern Europe, World (1985, 2025, 2050) * Figure here * 17. The use of fuels for purposes other than electricity generation would grow by less than one-third. Contributions by renewables to direct fuel use could reach nearly one-fourth by 2025 and two fifths by 2050, with most of the contributions coming from biomass-derived fuels - methanol, ethanol, hydrogen, and biogas (contribution by renewables to direct fuel use in 1985 was negligible). It was also expected that methanol and hydrogen might prove to be biofuels of choice, as they would be the most adaptable energy carriers in transportation. 18. As a direct result of the penetration by renewables into the energy systems, global CO2 emissions were expected to decline. As shown in Figure 1 above, CO2 emissions under the renewables-intensive scenario would be reduced by 12 per cent by 2025 and 26 per cent by 2050 from its base year level. In 2050, world CO2 emissions would be 4.2 billion tonnes, somewhat less than WEC/IIASA scenario (Case C). Emissions from industrialized countries, including the reforming economies, would be reduced nearly by half by 2025 and nearly two-thirds by 2050. Their share of global emissions would decline from three-fourths in 1985 to about two-fifths in 2025 and one-third in 2050 (this holds for nearly all scenarios). While the relative contributions by developing countries would rise, their per capita emissions in 2050 would still be only one-third of those by developed countries. B. Motor vehicles and road transportation 19. Analysis of the energy sector normally incorporates the transport sector, since its energy use accounts for a significant share of total energy consumption. Recent studies indicate that the transport sector as a whole (including passenger travel and freight movement by all modes of transport) accounted for about a quarter of world primary energy use in 1990. It was also responsible for 22 per cent of CO2 emissions from fossil-fuel use. Prospects for energy demand and emissions by the transport sector depend on a number of variables, notably GDP growth, income, industrial structure, vehicle fuel efficiency, lifestyles, and geographic features. 20. In road transportation, studies of historical relationships suggest that there would be faster growth in developing regions in the coming two decades, especially in middle-income and fast growing economies, although in terms of fleet size and fleet density they would continue to lag behind the high-income developed countries. 21. Such projections are reinforced by results of model runs and scenario analysis, such as the IPCC scenarios, the TARGETS/IMAGE energy-related scenario and models based on input-output analysis. Here we introduce one of the studies using input-output analysis, the World Model. 14/ 22. The World Model was based on the original version of the input-output model developed by Wassily Leontief. 15/ The model describes the economy as a system of inter-dependent activities. It aims at producing internally consistent multisectoral projections of economic trends and provides detailed quantitative assessments of both the direct and indirect effects of policy actions. In this regard, it also provides specific scenarios of the environment in the context of economic growth and technological dynamics. 23. The World Model divides the world into sixteen regions linked by commodities trade and flows of capital and aid. Each region in turn is described in terms of some fifty interacting sectors. In the sector of road transportation, the World Model analyzes relevant trends in 1980-1988 and generates projections to 2020 (see Table I.2 below). The Model anticipates a significant increase in motor vehicle densities (defined as motor vehicles per 100 people) in developing regions, ranging from 2.1 per cent per year in newly industrializing Latin America, to 5.2 per cent per year in centrally planned Asia. Because of population growth these increases will translate into large percentage increases in motor vehicle fleet. For instance, between 1988-2020, motor vehicles are expected to increase by 3.7 per cent a year in newly industrializing Latin America and 6.2 per cent a year in centrally planned Asia. In contrast, little increase is projected for developed regions in vehicle density or fleet. But in terms of the sales and fleet sizes, developed regions will continue to possess more than half of the world vehicles (55 per cent), as against 78 per cent in 1988. Table I.2: Motor vehicle fleet and density, 1988 and 2020 (millions of units and motor vehicles per 100 people) ------------------------------------------------------------------------------ 1980 a/ and 1988/1990 b/ 2020 1988-2020 c/ ----------------------------------------------------- Fleet Density Fleet Density Fleet Density ----------------------------------------------------- High-income North America (169.1)201.8 (66.3)72.2 243.2 73.2 0.6 0.0 Oceania (9.5)11.3 (40.1)40.8 22.8 56.9 2.2 1.1 High-income Western Europe (104.8)133.9 (35.8)45.1 148.8 50.2 0.3 0.3 Japan (37.9)52.4 (32.4)42.5 66.4 51.1 0.7 0.6 Medium-income Western Europe (15.4)23.0 (12.3)16.4 53.0 30.0 2.6 1.9 Eastern Europe (11.0)16.6 (9.8)14.3 25.4 20.0 1.3 1.1 Southern Africa (3.5)4.3 (10.8)10.5 10.0 13.7 2.7 0.8 Newly industrializing Latin America (21.7)29.0 (9.4)10.1 91.9 20.0 3.7 2.1 Former Soviet Union (15.5)24.7 (5.8)8.6 54.9 16.0 2.5 2.0 Major oil producers (6.0)11.9 (3.5)4.8 68.4 12.0 5.6 2.9 Newly industrializing Asia (4.4)10.8 (1.6)3.3 70.4 15.5 6.0 5.0 Low-income Latin America (3.5)4.6 (3.2)3.3 12.1 5.1 3.1 1.4 N.Africa and other Middle East (2.5)3.4 (1.4)1.5 13.3 3.0 4.3 2.2 Low-income Asia (3.4)5.0 (0.3)0.4 44.8 2.0 7.1 5.2 Sub-Saharan Africa (2.1)2.6 (1.1)1.0 7.2 1.1 3.3 0.5 Centrally planned Asia (0.9)4.3 (0.1)0.4 30.1 2.0 6.2 5.2 World (411.1)539.8 962.53 1.8 ----------------------------------------------------------------------------- Source: Institute for Economic Analysis based on various studies, as quoted in Faye Duchin et al, op.cit. Note: a) 1980 figures are provided in brackets for comparison; b) Density is understated because fleet figures are for 1988, while the population figures are for 1990; c) 1988-2020 are projections of growth rates expressed as percentage changes per year. 24. Average fuel efficiencies are expected to continue to make headway. By 2020, trucks in developed countries can run 20 miles per gallon, compared with 9.6 miles per gallon in 1987. Fuel efficiency for cars will improve to 39.8 miles per gallon as against 19.1 miles per gallon in 1987. If consumers accept changes in vehicle size and performance, transport energy intensity could decline by 60-80 per cent in the next thirty years. For the transport sector as a whole, improved vehicle efficiency might reduce greenhouse gas emissions per unit of transport activity by 20-50 per cent in 2025 relative to 1990. These projections lead to the statement that transport, in particular public transport is one area where government policies could make a big difference. 16/ C. World demand for metals and minerals 25. In addition to energy, commodities represent one sector where there has been a tradition of forecasts and projections, mainly geared to demand and supply conditions and possible paths of price fluctuations. In the 1980s, consumption of metals in industrialized countries registered further declines, whereas demand in developing countries kept growing. The picture at the global level for most major metals is that of continuing growth (see Table I.3 below). Table I.3: World primary production of metals (in 1000 tonnes) ------------------------------------------------------------------------------ 1980 1985 1990 ------------------------------------------------------------------------------ Aluminum 16,064 16,568 19,347 Crude steel 716.500 386.741 769.800 Copper 7,540 8,221 8,561 Lead 3,343 3,181 2,761 Zinc 6,147 6,466 6,685 Nickel 757 780 858 ----------------------------------------------------------------------------- Source: As compiled in Detlef van Vuuren. 26. As regards long-term projections, both economic and integrated system assessment models have been applied. For instance, the integrated assessment model TARGETS includes a Minerals sub-model which models mineral resource use on a global scale. The mineral resource is not represented as one group in the model, but is divided into metals of abundant supply and medium scarce supply, with each group modelled separately. In addition, recycling and waste material fluxes are built into the model. 17/ 27. Essential for the Minerals model is that it is not an independent model but is functionally linked up with the TARGETS model through an interactive process, following the Pressure-State-Impact-Response scheme generally used with the TARGETS model. The interactive representation scheme allows the Minerals model to show the possible inputs and outputs of the model to the rest of the TARGETS, as well as the most important parameter assumptions (see Figure 2 below) 28. The Minerals model has a strong focus on the United States, for which more historical data are available, thus making it possible to calibrate the model for this country. However, there is also a world version in which mineral resource use is analyzed in regional perspective, and the results of the United States calibration are translated into the world version, following the compilation of the world data set. 29. The world version generates projections of production and consumption of lead. The results suggest that in developed countries, lead production and consumption started to increase from the beginning of the century and surged after 1945. After 1975, production and consumption began to stabilize and even started to fall. This trend is projected to continue into 2000. In the case of developing countries, the picture is somewhat different. Up till now, lead production and consumption have been on the rise and are expected to stabilize before 2000. 30. In addition to the integrated assessment models, input-output models were also used to assess long-term trends and relationships. 18/ In its metal sector study, the World Model described above derived region-specific input coefficients of metals, taking into account such variables as future energy prices, technological changes, economic growth and materials substitution. 19/ 31. For the developed regions, the Model assumes that the trend toward lower input coefficients for materials will continue due to product substitution, improved production processes, materials-saving designs and innovative use of traditional materials. The exception is the aluminum coefficients which are assumed to increase slowly since aluminum is expected to continue to substitute for steel and copper (See Table I.4 below for detailed breakdowns). 32. For developing regions, a less uniform picture is expected to emerge. For those more advanced developing countries, the copper and steel coefficients are expected to decline. For those less industrialized countries, their materials input structures is likely to move toward the present input structures of the developed regions. In some areas, such as cable communication, they may benefit from the most advanced technology, such as optical fibers, and see their copper coefficient decline in the coming years. 33. For the formerly centrally planned economies, the Model assumes that there will be a major shift from steel-intensive heavy industry toward patterns of metal use more typical of consumer goods production. It is also assumed the technological gap with the developed market economies will narrow and there will be increasing incentives to avoid wasting materials. Steel coefficients are thus projected to decline rapidly, while the input coefficients for aluminum will increase more rapidly than in most other regions. 34. As regards virgin materials, the Model expects their share to continue to decline relatively slowly. Much of the metals is used for construction and infrastructure, and has a long life cycle. The increasing use of composite materials and alloys also reduces the potentials for recycling. The Model therefore expects the share of virgin materials in total metal consumption to decline at an annul rate of 0.5 per cent for all regions and all metals. Table I.4: Changes in gross metal input coefficients, 1990-2020 (average annual per cent change) ------------------------------------------------------------------------------ Copper Aluminum Nickel Zinc Lead Steel ------------------------------------------------------------------------------ High-income North America -1.50 0.50 -1.00 -0.50 -0.50 -1.00 Newly industrializing Latin America -1.50 0.50 -0.50 -1.00 -1.00 -2.00 Low-income Latin America -1.00 1.50 0.50 -1.00 -1.00 -2.00 High-income Western Europe -1.50 1.00 -1.00 -0.50 -0.50 -1.00 Medium-income Western Europe -1.50 0.50 -0.50 -0.50 -0.50 -2.00 Eastern Europe -1.50 1.50 -0.50 -0.50 -0.50 -2.00 Former Soviet Union -1.50 1.00 -1.00 -1.00 -1.00 -2.50 Japan -1,50 0.50 -1.00 -0.50 -0.50 -1.00 Centrally planned Asia -1.50 1.00 -0.50 -1.00 -1.00 -2.75 Newly industrializing Asia -1.50 1.00 -0.50 -1.00 -0.50 -2.00 Low-income Asia -0.50 1.50 -0.50 0.00 0.00 -1.50 Major oil producers -1.50 0.00 3.00 0.00 -0.50 -2.75 N.Africa and other Middle East -0.50 1.00 4.00 0.00 -0.50 -1.00 Sub-Saharan Africa -1.50 0.00 0.00 -1.00 0.00 -2.00 Southern Africa -1.50 0.50 -0.50 -0.50 -0.50 -1.00 Oceania -1.50 0.50 -0.50 -0.50 -0.50 -1.00 ----------------------------------------------------------------------------- Source: Institute for Economic Analysis projections, as cited in Faye Duchin et al, op. cit. D. Agriculture and food security 35. For planning and food security purposes, models have also been used in assessing prospects for future food and agricultural production and supply. The 20 year perspective study by FAO represents one such effort. 20/ 36. The overall approach under the FAO study was to start projections using Engel demand functions and exogenous assumptions of population and GDP growth, with 1988/1990 as the base year. This was followed by projections of production using provisional targets for each commodity and country. For cereals, livestocks and oilseeds sectors, a formal flex-price model (the FAO World Food Model) was used to provide starting levels for iterations and to keep track of changes in variables. The World Food Model itself is composed of country modules and world market feedbacks leading to market clearing through price adjustments. Throughout this process, specialists for specific commodities, countries and disciplines were consulted for expert judgment. The results of the perspective study may be described as a set of projections which meet conditions of accounting consistency and reflect views expressed by country and commodity specialists. 37. Under the FAO perspective study, world availability of food and agricultural products (expressed as calories/caput/day) will increase from 2700 cal/caput/day in 1988/90 to 2800 cal/caput/day in 2010 and 3000 cal/caput/day in 2025. In 2010 considerable gaps in food consumption will remain between developed (3470 cal/caput/day) and developing countries (2740 cal/caput/day). 21/ By 2025, the consumption level in developed countries is expected to remain at the 2010 level while that of developing countries will increase to 2900 cal/caput/day. However, this scenario will materialize only if the developing regions with low per caput food availability succeed in increasing their food production substantially or other regions have surplus and trade flows can be generated to match regional surpluses and deficits. Figure 3: Consumption of forest products, 1990 and 2010 Comparison of developed and developing countries ** Figure here ** 38. The FAO study projects only marginal increases in marine fish captures. As a result of overfishing in recent years and the declines in fish stocks, total marine catches are not likely to exceed 100 million tonnes in 2010, as compared with 86 million tonnes in 1989/91. Fisheries differ from other sectors of agriculture in that it is more difficult to relax the production constraints by merely investing in technology and exploration. This means that increases may have to come from aquaculture, which may become very important to sustaining growth. Assuming the growth rate of the last few years continues, the study expects world aquaculture production to increase to 20 million tonnes in 2010 from the 1990 level of 12 million tonnes. 39. For forest products, projections for consumption are based on the estimated relationships between economic and population growth and growth in demand for forest products. For developing countries, growth is projected to be approximately equal to their economic growth (5 per cent per annum), while consumption in developed countries is likely to lag behind economic growth, given the high average consumption level prevailing now. 40. Developed countries are likely to increase industrial roundwood consumption from 1,270 million m3 in 1990 to 1,900 million m3 in 2010. For developing countries, roundwood consumption is expected to increase from 380 million m3 in 1990 to 800 million m3 in 2010. Similar growth rates are projected for paper consumption, with developed countries increasing paper consumption from 196 million tonnes in 1990 to 310 million tonnes in 2010. Paper consumption in developing countries is expected to grow to 130 million tonnes in 2010, up from a low level of 42 million tonnes in 1990 (See Figure 3 below). 41. The growing demand for forest products, combined with the expected expansion of agricultural land into forest areas, is expected to accelerate deforestation in developing countries. Driven by worsening incidence of rural poverty, landless farmers may claim land from forest areas at a higher rate than warranted by the required growth in production. In addition, deforestation is projected to continue, aggravating land degradation. Of the 1.2 billion hectare of degraded land worldwide, deforestation and overgrazing are estimated to account for two-thirds. III. METHODOLOGICAL APPROACHES TO MODELING 42. The preceding section presented scenarios of selected consumption trends generated by different models. In this section we take a look at the models themselves, and examine their differences in methodological approaches by focusing on their structures and input assumptions. A. Structure of models 43. The last two decades have witnessed much progress in modeling, especially in the macroeconomic, environmental, energy, agriculture and transportation sectors. One way of comparing these models is to divide them into two large categories - "top-down" and "bottom-up" models. 1. "Top-down" models 44. "Top-down" models are econometric models of the aggregate economy. They usually contain rather aggregate sectoral demand functions based on economic indices of prices and elasticities. They aim at capturing the overall macroeconomic impact of policy variables. Their key structural features include endogenization of behavioral relationships and incorporation of large economic components (such as investment, trade, consumption and income distribution). 22/ But their description of specific sectors, such as energy or agriculture, is limited. Consequently, while "top-down" models have strong orientations toward economy-wide effects, they are considered weak in exploring technological options and potentials at the sectoral level. 45. In the IIASA/WEC energy perspective study, an energy-economy interactions model called 11R was used to check for consistency between macroeconomic development and energy demand. 11R is a modified version of Global 2100 model originally developed by Manne and Richels in 1992 and subsequently widely used in energy studies throughout the world. 23/ However, the "top-down" models have been criticized for failing to describe adequately the underlying determinants of sectoral demand dynamics. 2. "Bottom-up" models 46. Traditionally "bottom-up" models have been used to study the dynamics of a given sector, such as agriculture or energy. They adopt a more disaggregated approach to demand and supply and provide for technological changes. They contain more detailed and precise descriptions of parameters, such as end use patterns and technological alternatives, but no treatment of feedbacks between the parameters and underlying economic variables. Because of this lack of economic behavioral feedbacks, "bottom-up" models are considered best suited for simulation analyses to explore, for instance, the efficiency effects of the introduction of a given set of new technologies. 47. "Bottom-up" models can more easily be prescriptive because they examine the dynamics of introducing alternative technologies or practices, including "best practice" or "state-of-the-art". Models that explore the system-wide introduction of renewable sources of energy normally fall under this approach. In contrast, a "bottom-up' model that adopts a descriptive approach will explore technological alternatives and individual practices that result from actual decisions. They tend to be less optimistic about normative policy stances. 48. In the WEC/IIASA study, a "bottom-up" model called MESSAGE III was used to generate detailed estimates of energy demand and supply. MESSAGE III is a dynamic linear optimization model, calculating cost-minimal supply structures under the constraints of resource availability, given technologies and demand for energy. The WEC/IIASA study used both the "top-down" and "bottom-up" models in order to capture the benefits arising from different model structures and perspectives. 49. Another "bottom-up" model that has been applied for policy purposes is the ESCAPE model, developed for the European Union through a collaborative effort involving experts from Britain and the Netherlands. 24/ The ESCAPE (Evaluation of Strategies to address limate change by Adaption to and Preventing Emissions) model consists of suite of models (modules) which enables scenarios of greenhouse gas emission to be constructed and their impact on global and regional climate and sea level and sectors of the European economy to be assessed. The default time step is five years and the projection extends from 1900 to 2100. 50. The ESCAPE model comprises four inter-linked modules, including an emissions module, two integrated climate modules and a climate change impacts module, all developed by RIVM in the Netherlands and institutes from UK. The results of the model calculations highlighted three important characteristics of the global climate change problem: past emissions of greenhouse gases and the inertia of the global development path have committed the world to future warming irrespective of current and near-future policy interventions; the efficacy of a climate policy implemented solely within the EC on altering the course of future climate change is very small; and the impacts of climate change on the economy and environment of the EU differ markedly between northern and southern Europe. 3. Integrated Assessment Models (IAM) 51. Despite the progress in econometric and sectoral modelling, the search for a comprehensive grasp of the natural and human systems gave rise to integrated assessment models (IAM). The practice of separating human and natural systems as embodied in traditional economic and sectoral modelling has proved inadequate to capture the dynamics of global changes. 25/ 52. IAMs provide a framework for co-ordinating research in different disciplines. The exercise may generate unique insights about key policy questions that are hard to come by through aspect-compartment oriented research. The integration process itself allows researchers to coordinate assumptions from different fields and introduce feedback mechanisms unavailable in individual discipline-based assessments. This can be achieved by linking relevant component modules in a computer program through formal mathematical representation. The TARGETS/IMAGE models presented above are examples of such models. In the IIASA/WEC study, integrated assessment models were used to help determine SO2 and NOx emissions and land use changes ( See Appendix 1 for a summary characterization of IAMs). 53. IAMs are recognized to have several advantages. Among others, they provide a comprehensive setting necessary for systematic assessment of long- run dynamics. For instance, integrated assessment helps place the phenomenon of climate change in the broader perspective of global change, which include all human interventions and responses. Secondly, IAMs help identify gaps in knowledge and information in individual disciplines and reset priorities for decision making accordingly. Thirdly, IAMs help identify and clarify sources of uncertainties, and transform them into risk analysis to assist in decision- making under uncertainty. 26/ 54. As a young discipline, IAMs have their own disadvantages. Because IAMs are intended as a means of capturing the entire cause-effect chain of a system under consideration, they are prone to an accumulation of uncertainties, affecting the reliability of results. Furthermore, though linked up, sub- models of IAMs have yet to be fully integrated to reflect full systemic interactions among the various components of human and natural systems. 55. IAMs are a rapidly evolving field. Growing research interest and increasing applications in policy context will undoubtedly strengthen them. Yet it needs to be borne in mind that findings of IAMs depend a lot on the findings of sector/discipline-oriented studies. The quality of integrated assessments will be enhanced when gaps and weaknesses in individual disciplines are remedied. This will allow a more precise representation of variables underpinned by the knowledge in corresponding domains. 4. Global input-output model 56. In input-output analysis, an input-output table is used to provide a systematic picture of the flow of goods and services among the producing and consuming sectors of the economy. It also registers the flows of goods and services out of a given region and the flow of goods and services into the receiving region. 57. The input-output model, pioneered by Wassily Leontief, is considered well suited for analysis of inter-dependent relationships, such as inter-industry activities. The World Model described above in the transport and minerals sections was based on a slightly modified version of the original input-output model. Although an economic model, the input-output approach differs from the mainstream general equilibrium models. Eschewing the equilibrium perspective, the World Model seeks to represent the activities of a real economy and arrive at quantitative results. 27/ 58. The input structure of the producing sector is expressed in terms of a set of technical coefficients specifying the amount of goods and services required to produce a unit of outputs. There is also a separate set of capital coefficients describing the required capital stock. The inputs of primary resources, such as land, water and minerals, are also depicted and analyzed along with the production and consumption of ordinary goods and services. 28/ 59. A key feature of the input-output analysis is considered to lie in its empirical content of the data (although from an operational perspective this can become a disadvantage, as they are date-intensive, requiring a lot of effort in data collection). Unlike general equilibrium models, in which all major relationships are represented by mathematical equations, the detailed description of technological choices embodied in technical coefficients brings the model closer to the real economy. In addition, the model allows greater openness to multidisciplinary collaboration than equilibrium models. 5. Econometric model plus expert judgment 60. In the FAO perspective study of world agriculture, use of models is combined with expert judgment to produce projections, which is often the case in scenario development and analysis. A significant part of the research effort is initially devoted to assembling a consistent set of historical and base year data, covering each individual country and commodity. This makes it possible to construct an overall quantitative framework for demand-supply analysis, based on supply utilization accounts (SUA). 29/ Subsequently, SUAs for the year 2010 are drawn up, by commodity and country. For the cereals, livestock and oilcrop commodities, the projections were derived from a multi-commodity, multi-country flex-price model (World Food Model). The model contains demand and supply equations for each country. Trade flows provide the dynamics between domestic and world markets. The world market in turn is cleared through price adjustment. 61. The initial projections thus generated by the model are subjected to inspection by specialists. Adjustments are then made by fine-tuning the model's parameters and coefficients, mainly those related to the supply side, (production and trade parameters). This process of iterative computations underpinned by expert opinions continues till the final round when the world demand, production and trade balances are fully adjusted. 62. The end result of this mixed methodological approach is therefore supported by detailed country/commodity quantitative analyses. The world picture can be decomposed back into constituent single country or commodity statements. This feature is regarded as a strong point of the study since most global studies are carried out at the level of regions and major countries, with the major products grouped into a few commodity aggregates. 30/ 63. However, the heavy dependence on great detail and expert input is also considered a major weakness of the mixed approach. Projections based on specialist input suffer from the lack of uniformity and standardization of assumptions and criteria. As a result, they cannot be reproduced; nor can they be used to estimate alternative scenarios by varying certain assumptions. 31/ B. Assumptions and perspectives of models 64. It is clear from the preceding discussions that model assumptions of dynamic variables and parameters play an essential role in determining the output of the model runs and in exploring alternative scenarios. To a large degree, formulation of such assumptions represents one key channel of infusing model builders' knowledge, judgments and values into the model. 65. Assumptions affect the outcomes of models both implicitly and explicitly. In selecting model types and determining the structure of the model, model builders' assumptions of the large picture are automatically embedded in the mathematical structure of the model. For instance, economists tend to prefer "top-down" models which endogenize behavioral relationships and allow for an economy-wide evaluation of changes in policy variables. Engineers are likely to build " bottom-up" models which enable them to make more detailed statements about a specific sector and explore the dynamics and potentials of technological options. 66. In the energy sector, for instance, the "top-down" approach embodies the "economic paradigm", which tends to be less optimistic about the potentials for energy efficiency gains to be found in the best available technologies. Economists argue that there may be reasons why consumers have not adopted the optimal technologies (cost or price factors). Engineers, on the other hand, seem to rely more on the "engineering paradigm", pointing to evidence that shows the energy efficiency gap - the gap between the energy efficiency of equipment actually chosen by consumers and the efficiency of the best available technology. They are therefore more optimistic about the technological dynamics. 67. When model preferences are decided on, assumptions about key variables and inputs may affect model outcomes more explicitly. For instance, while all models include population and economic growth assumptions, and have similar, if not identical values, they may disagree about many other variables. In the energy sector, varying assumptions of fossil fuel supply, price and income elasticities of demand, fuel substitutions and technological dynamics may well lead to divergent and opposing outcomes. The ongoing debate about the autonomous efficiency improvement is a case in point. 32/ In the food and agricultural sector, continuing increases in output may depend on availability of land and high-yield varieties (both assumptions are however surrounded with uncertainty). 33/ 68. The increasingly complex problems models are expected to tackle inevitably mean that there may be more possible outcomes and hence more scenarios to be simulated. Assumptions are accordingly varied in order to match the complexity. In addition, under normative scenarios, assumptions can be policy intensive. For instance, under the ecologically-driven scenarios, the final outcome may hinge on policy measures in energy investment (in favor of renewables), and on whether energy and carbon taxes are levied or not. 69. That scenarios can be perspective-driven accentuates the issue of values and perspectives of model builders. To what extent are the model structure, assumptions and outcome affected by the values of model builders? Most probably, there is no uniform answer to this question. But a number of criteria can be used to judge the soundness of scenarios. These may include comprehensiveness (whether environmental, economic, social and cultural dynamics are covered at sufficiently detailed levels); diversity (whether different interests and opinions are represented); and methodological soundness (whether transparent, consistent, and reproducible). 34/ C. Limits and limitations 70. As noted in the introduction of this background report, the need for an integrated approach to the inter-locking issues of the environment, economy and society has pushed researchers toward a more structured investigation into the vastly complex process of dynamic interactions between the human and natural systems. The emergence of integrated assessment models represents an attempt to improve and strengthen the societal response. Increasingly, modeling and scenarios analysis will become the knowledge and information basis for policy making, as already illustrated by the role of the RAINS model in controlling transboundary air pollution in Europe. 71. However, there are limits to the use of models, recognized by every model builder. In many areas, our state-of-the-art knowledge may simply be inadequate to allow us to project the future course of events with a sufficiently high level of confidence. The fact there are six different business-as-usual scenarios of carbon emissions (IPCC IS92a-f variants) is evidence of the uncertainties associated with the current state of our knowledge. However, climate change negotiations making use of scientific enquiries, including modeling exercises, do demonstrate that models can assist in policy making by transforming uncertainties into risk management. 72. Past experiences also suggest that sectoral and econometric models have their respective limitations, as briefly mentioned above. The tendency to move toward the middle ground by builders of both "top-down" and "bottom-up" models (hybrid models) indicate that progress is being made. The rise of integrated assessment models will certainly accelerate this movement. 73. Recent emphasis on changing consumers' behavior also brings into focus the need for models to incorporate such variables as consumption values and lifestyles, which are being increasingly recognized as underlying factors of consumer behavior of the same weight as that assumed by income and price changes. The difficulties in expressing these variables in equations amenable to estimation represent a technical as well as a scientific challenge to model builders. 74. Furthermore, some global models have been criticized for treating the developing economies as if they will embark on the same growth path as that followed by developed economies. While economic and regional groupings may help alleviate that concern, the assumptions about uniform economic structures and identical patterns of behavioral relationships could result in outcomes wide of the mark. IV. CONCLUSIONS 75. This summary of selected models and scenarios and the brief discussion of methodological approaches show that models are not truth machines that give definite answers, but rather tools for organizing knowledge and structuring analytical processes, which allow us to explore possible future options, and gain a better insight into complex problems. 76. Scenarios are descriptive and not prescriptive images of the future, meant to project what may happen under several alternative development pathways and to evaluate the social, economic and environmental costs and benefits of different strategies. 77. The development of a consistent set of scenarios and a coordinated data management system would help integrate different subsystems in a coherent fashion. 78. Use of integrated models in the development of long-term projections allows for analysis and evaluation of systemic dynamics and trade-offs between social, economic, and ecological developments, which may lead to better informed priority setting. The growing interdependence of the environment and socio-economic development, coupled with increasing maturing of IAMs, point to a potential role of IAMs in helping to map out the future course of sustainable development. GLOSSARY Autonomous energy efficiency improvement (AEEI) AEEI refers to improvement in end use energy efficiency not caused by or related to increases in fuel prices. They normally arise form technological development, improved processes and demand-side management. Calibrate Matching various relationships in the model to available historical data, and bringing the individual relations together into a model and demonstrating that the model reproduces the historical behavior of certain variables. Endogenous A variable that is calculated internally by the model is endogenous. Once the model is initialized, no further information or step is required and the model is programmed to calculate automatically. Exogenous Opposite of the endogenous. Information or variable that is calculated outside the model are fed by the model builders into the model to do the calculations. General equilibrium model In contrast to partial equilibrium model, which examines one market or sector, holding other parts of the economy constant, general equilibrium model (where demand always equals supply), is more sophisticated in conception and mathematical formulation, regards the economy as a whole system, and requires the simultaneous determination of all prices and quantities of all goods and service in the economic system. Parameter and variable A quantity which remains constant in a given context is a parameter, and which changes is a variable. For instance, in the equation Y = a + bX where X and Y are variables and a and b are constants and the parameters of the equation. Simulation A method by which a range of alternative scenarios are generated based on varying assumptions about future situations in order to answer "what if" type questions. In practice, this is usually achieved through altering the values of exogenous variables and parameters. It is considered well adapted for assessing the likely impact of policy alternatives. Appendix 1. Summary Characterization of IAMs (By Jan Rotmans and H.Dowlatabadi) Notes 1/ P-S-I-R refers to the driving forces, the changing states of the environment and socio-economic system, the systemic impacts, and the societal response to the changes. 2/ In the process of communication, models are often presented as reality, while in fact they are analytical tools, albeit sophisticated ones. Indeed, models usually have a large degree of uncertainty, and their results should not be interpreted as definite answers but rather as explorative indications of possible options. For a more detailed discussion, see Jan Rotmans, Models for Sustainable Development, a discussion paper presented at an expert/policy- maker meeting organized by the Division for Sustainable Development in February 1996, New York. 3/ The selection of the models is carried out among the most recent studies done with models, partly depending on materials available, and partly on the range of issues covered in the trends section of the consumption report. It does not in any way constitute approval or otherwise by the Department for Policy Coordination and Sustainable Development of the models introduced in the report. 4/ World Energy Council and International Institute for Applied Systems Analysis, Global Energy perspectives to 2050 and Beyond, 1995 report. 5/ For the period of 1990-2020, Case A envisions an annual growth rate of 2.7% for the world economy. Cases B and C share the same assumption - 2.2 %/yr. For the period of 2020-2050, Case A projects an annual growth rate of 2.6 %, while Cases B and C suggest an annual growth rate of 2.0 and 2.1 %, respectively. 6/ For OECD countries, Case C projects energy intensity declines by 2.0 % per year between 1990-2050, as against 1.2 and 1.1 % under Cases A and B. For economies in transition, all three cases depict a higher rate of improvement compared with either OECD or developing countries. Within developing countries, the dynamics of energy intensity improvements vary in accordance with growth prospects and per capita GDP. 7/ Subsequent studies which had a better grasp of regional implications basically confirmed the scenarios developed in 1992. In more recent studies, the IPCC Working Groups used carbon cycle models to derive scenarios of accumulated CO2 emissions and to explore various stabilization paths. 8/ The Second Sulfur Protocol of the European Convention on Transboundary Air Pollution (UN ECE, 1994) provides for measures to lower maximum excess deposition of sulfur to below 3 g S/m2 per year. The level projected in the RAINS scenario is therefore a very high deposition level. 9/ Critical loads are defined as the maximum deposition levels at which ecosystems can function sustainably. The values of critical loads vary in accordance with the ecosystems affected. The development of the critical load concept in recent years has helped define the configuration of the debate on sustainable development, highlighting the growing prominence of the sink function factor in decision-making. 10/ TARGETS/IMAGE stands for Tool to Assess Regional and lobal Environmental and Health Targets for Sustainability and Integrated Modeling to Assess the Greenhouse Effect, respectively. For a more detailed discussion, see Rob Swart, Marcel Berk, and Bert de Vries, Long-term Scenarios for Global Sustainable Development: A Basis for Structured Debate, paper presented at the United Nations University Conference on the Sustainable Future of the Global System, Tokyo, 16-18 October 1995. 11/ For a detailed description of the key characteristic of the world energy scenarios simulated by the RIVM using the TARGETS/IMAGE energy model, see the Appendix 1 of the paper presented by Swart et al at the UNU conference. 12/ Thomas B. Johansson, Henry Kelly, Amulya K.N. Reddy and Robert H. Williams, " Renewable Fuels and Electricity for A Growing World Economy: Defining and Achieving the Potential", in Thomas B. Johansson et al (eds), Renewable Energy: Sources for Fuels and Electricity, Island Press, Washington D.C., 1993. 13/ It was assumed that the share of natural gas in power generation would nearly double by 2025 from its 12 percent share in 1985. The dissemination of advanced gas-fired gas turbine power generation would add to the flexibility to power generation systems so that electrical output could be adjusted quickly in response to changes in the output of intermittent power generation. 14/ The World Model is an input-output model of the world economy, a version of which is used in scenario analysis by Faye Duchin and Glen-Marie Lange. See Faye Duchin and Glenn-Marie Lange, The Future of the Environment: Ecological Economics and Technological Change, Oxford University Press, 1994. 15/ Professor of Economics at New York University who was awarded the Nobel Prize in Economic Science in 1973 for his pioneering work in analysis of the interdependencies within an economic system applying the input-output technique. 16/ Experiences in several cities show that with appropriate measures in place, urban transport could be greatly improved, reducing car use, congestion and improving air quality. 17/ See Detlef van Vuuren, Modelling Metal Resource Use: Documentation, Calibration and Extension of the TARGETS Minerals Model, RIVM/Global Dynamics and Sustainable Development, 1995. 18/ See, for instance, Wassily Leontief et al, The Future of Nonfuel Minerals in the U.S and World Economy: Input-Output Projections, 1980-2030, Lexington Books, 1983. 19/ Coefficients for past years were derived from accounting data by separately estimating the numerator (metal consumption by a given sector in a particular region) and the denominator (corresponding sectoral output). The metal input coefficients were calculated for the United States, for which consumption data were available for 1980 and 1987 and the coefficients for other regions were derived from them. 20/ See Nikos Alexandratos, World Agriculture: Towards 2010, published by FAO and John Wiley & Sons, 1995. 21/ The 1988/90 availabilities for developed and developing countries are 3400 calories/caput/day and 2470 calories/caput/day, respectively. 22/ "Top-down" model may be further divided into neo-keynesian macroeconomic models and computable general equilibrium models. The former incorporates econometrically-estimated sets of equations that keep track of the short- and medium-terms dynamics of economic aggregates and related economic activities. Such models simulate aggregate potential output as a function of aggregate inputs of capital, labour and materials. The general equilibrium models focus on long-term analysis. They rely on the resource allocation principle and market clearing mechanism for all goods by equating prices with marginal costs. Capital accumulation and exogenous growth of factors of production provide the dynamics of the models. Instead of using econometric estimations, the models are benchmarked to a base year to guarantee the consistency of the parameters. 23/ Global 2100 is one of the models used in the Energy Modeling Forum(EMF). All EMF models are considered "top-down" models. See Manne, A, and R. Richels, Buying Greenhouse Insurance; The Economic Costs of CO2 Emission Limits, MIT Press, Cambridge, USA, 1992. 24/ For a detailed introduction of the model, see Jan Rotmans, Mike Hulme and Thomas E Downing, Climate change implications for Europe: An application of the ESCAPE Model, Global Environmental Change, 1994 4, pp. 97-124. 25/ For a comprehensive treatment of past examples of IAMs as well as an overview of the current integrated assessment modelling activities, see Jan Rotmans and H. Dowlatabadi, Integrated Assessment of Climate Change: Evaluation of Methods and Strategies, 1995. 26/ Ibid. 27/ See Faye Duchin and et al, op.cit., pp. 5-7. 28/ The inclusion of natural inputs into the model means that environmental assets are considered, which is a step forward. However, the feature of fixed coefficients also makes it difficult to capture the dynamics, and works against long-term projections. 29/ The SUA is an accounting identity showing the sources and uses of agricultural commodities in homogeneous physical units. It is expressed as Food (direct)+Industrial non-food uses+Feed+Seed+Waste=Total domestic use=Production+(Imports-Exports)-(Opening stocks-Closing stocks). 30/ See Nilkos Alexandratos, op.cit., pp. 407-419. 31/ It is envisaged that future efforts at improving the methodology of the perspective study will aim at introducing some advantages of formal models, such as explicit statements of the assumed behavioral relationships, empirical verification, replication of results and derivation of alternative scenarios in a consistent manner. The strong points of the present approach will however be preserved. These include the details of analysis regarding individual countries, commodities and supply constraints, as well as the opportunity for using multidisciplinary input and expert judgments. 32/ It is often argued that with easy options in both demand and supply side management being exhausted, potentials for autonomous efficiency improvement may have been overestimated, and improvement rates are likely to decline. Furthermore, without substantial investment, developing countries are unlikely to achieve improvement rates as envisaged in some models. 33/ All these assumptions are also made (mostly implicitly) in assessments that do not make use of models (they are locked in the mental models of experts). The advantage of using models is that the consistency of these assumptions can be checked and validated and made more explicit. 34/ See Rob Swart et al, op.cit, pp. 3-6.
This document has been posted online by the United Nations Department of Economic and Social Affairs (DESA). Reproduction and dissemination of the document - in electronic and/or printed format - is encouraged, provided acknowledgement is made of the role of the United Nations in making it available.
Date last posted: 3 December 1999 10:27:35