| Energy research in developing countries |
|Volume 10: Energy planning: models, information systems, research, and development|
B. Lapillone, P. Criqui, and J. Girod
This paper reviews three types of developments in energy modeling. First, econometric models were enhanced to offer a more realistic representation of energy demand and the economic forces that shape it. Second, technoeconomic or energy end-use models were developed as an alternative approach to explain and forecast energy demand. Third, a number of global energy models were built to incorporate the supply and demand of resources into a multiregional framework of analysis. Some of the criticisms directed at energy models are reviewed, and suggestions are made on how modelers and their models could better serve the users of these analytic tools.
Before the oil crises of the 1970s, economic output and energy demand underwent regular and rapid growth. It was, therefore, acceptable for energy agencies and companies to work with rudimentary energy models that related energy demand to GDP, sectoral value added, and in some cases, energy prices.
After these oil crises, it became apparent that such simplified econometric models could not explain the factors affecting energy demand, assess the potential impacts of policy measures, or forecast the future growth and pattern of energy demand.
Econometric models use economic indicators (for example, GDP, sectoral output or income, and energy prices) to determine the level of energy demand, in total or by sector or form of energy. In the most common models, the variable coefficients of output and price represent the elasticities of demand with respect to output and price.
Traditional econometric models have several limitations. Price elasticities are difficult to measure, and they exhibit a wide range of values. Their role may also be overemphasized because nonprice factors (for example, conservation programs and efficiency standards) are ignored in these models. Other weaknesses for forecasting include the potential asymmetry between changes in demand caused by a decrease or an increase in prices, and the influence of anticipated and marginal prices (not just average prices) on consumer behaviour. Output or income effects may also be modeled inadequately if there is insufficient disaggregation by economic sector (to show impacts of structural change) or by income group (to show impacts of patterns of income growth).
To overcome these problems, econometric models have become more disaggregated, and they are limited only by the availability of sufficient historical data on the required indicators. Knowledge of price effects has been enhanced by considering both short- and long-term responses to price changes and by allowing elasticities to vary according to the level or rate of change of prices. Finally, more complex models have been built to account for the impacts of capital stocks on energy demand. These models include structural models in which demand is shaped by either the energy efficiency and rate of use of capital stock, or KLEM models, which use production functions to estimate substitution effects among factors of production, including energy.
Technoeconomic (End-Use) Models
Technoeconomic models provide details of the end uses of energy in each of several energy-consuming sectors. They also provide detailed calculations of useful (output) and final (input) energy needs for each end use and sector. End-use models can be used to evaluate demand management and fuel-switching possibilities because they distinguish between useful energy (for example, space heating needs) and final energy (for example, the consumption of oil, natural gas, or electricity) and account for losses due to energy conversion. This modeling approach needs detailed data on end-use coefficients (for example, useful heat energy required per unit of floor area), efficiencies of energy conversion (for example, boiler efficiency), and the stock of energy-consuming units (for example, floor area by type of building or sector).
Accounting models were the simplest and earliest forms of end-use models. Most of the model variables are exogenous (determined outside the model), and the model serves as an accounting tool. The principal advantages of the model are its realism and degree of detail. These models can simulate the impacts of alternative energy policies and programs on fuel- and sector-specific energy demand and the potential for energy conservation and interfuel substitution. End-use models can also be used to develop and analyze detailed balances of energy supply and demand. The disadvantages of these models are that the large quantity of detailed energy and technical data may be difficult and costly to obtain, energy prices are not explicitly considered, and the need for many exogenous assumptions may lead to scenarios that are internally inconsistent.
Technoeconomic models have been improved to explicitly consider price effects. A common enhancement is the incorporation of submodels of market shares that allocate total useful energy needs to competing (present or future) energy forms. These submodels determine fuel-specific shares for each end use on the basis of relative life-cycle costs, which in turn are based on equipment costs and efficiencies, energy prices, the cost of capital, tax rates, and investment incentives (for example, capital grants).
Global Energy Models
Concern about the depletion of nonrenewable resources and shocks in petroleum prices led to the development of global energy models. Three regionally disaggregated supply-demand models were developed by the Workshop on Alternative Energy Strategies (WAES) headed by the International Institute of Applied Systems Analysis (IIASA), the Massachusetts Institute of Technology, and the World Energy Conference (WEC). All three models considered the long-term balance of resource supply and energy demand in five to eight regions of the world. However, each had a particular focus. The WAES model focused on the substitution of coal for oil; the IIASA model, on the long-term prospects for nuclear energy; and the WEC model, on the impact that the rising energy demand from developing countries would have on oil markets. Other global models, such as the Energy Modeling Forum (EMF) and Choe's World Bank model, were designed to deal specifically with petroleum markets. Unlike the other three global models, the oil-market models addressed, albeit incompletely, the issue of energy-economy interactions. Global energy models could be enhanced if world energy data were improved and the importance of links between energy models and macroeconomic models were recognized.
Using Energy Models
There is no single modeling approach for all situations. The choice depends on objectives, the required degree of analytic detail, data availability, and time and budget constraints. Some general guidelines for modelers who want to enhance the usefulness of their models for decision-makers in the energy sector are
· Build as simple models as possible because complex models are increasingly mistrusted,
· Build conceptual and qualitative models, not just quantitative models,
· Use models not only to forecast, but to make and review decisions, and
· Fit models into a single structure to enable users to move from simple energy balances to energy analysis, forecasting, and planning.
Jean-Guy Devezeaux de Lavergne
This paper describes some of the features of developing countries that are relevant to energy modeling, provides an overview of what is meant by energy modeling, summarizes the features of different types of demand, supply, and global models, and suggests ways to improve the design and use of these models.
The energy sector plays several important roles in the economic activity of developing countries. Energy commodities are an input into the production of goods and services. Energy is a final consumption good that provides cooking, transportation, and other services. The energy sector also contributes to GDP and affects trade balances and the balance of payments. Therefore, there is great interest in modeling energy demand, energy supply, and energy-economy interactions.
Relevant Characteristics of Developing Countries
Developing countries share a number of economic and energy characteristics that affect energy modeling. Economic traits include
· Rapid population growth and low educational standards,
· High degrees of central planning,
· Regulated and constrained domestic markets,
· Significant traditional (agricultural and artisanal) sectors,
· Narrowly specialized economic structures and production,
· Constrained levels of capital and investment, and
· Weak currencies.
Energy characteristics shared by most developing countries include a lack of understanding of the concepts of energy planning, a scarcity of reliable long-term statistical data, and the importance of noncommercial energy (more than 90% in many developing countries). Despite these similarities, there are sufficient differences among developing countries to argue against the use of a single multicountry model. There are differences in the level of development, technical know-how, geographic size, and degree of openness of the economy. As well, the majority of energy-importing countries face different issues than the energy-exporting countries, and indigenous energy supplies vary among developing nations. These economic and energy features must be taken into account by energy models.
A model is a set of equations that represents the real world. Although both the data and the theory underlying the model may be imperfect, models are important tools. They can be used to plan and analyze the economic and energy impacts of policies and external events and to assess competing technologies that supply energy. Energy models are either hierarchical (open-loop) or global (closed-loop). Open-loop models use economic indicators as exogenous variables to define energy demand but do not include feedback from the energy sector to the rest of the economy. In contrast, closed-loop models consider the two-way interaction between energy and economic variables. The level of economic activity influences energy demand, which in turn affects the economy.
Hierarchical Demand Models
One-sector models use ratio analysis (energy demand per unit of output) and statistical time-series analysis to determine sectoral energy demand. This approach is simple, but this simplicity is also a limitation because the models consider neither substitutions among various energy forms, nor changes in the structure of the economy.
Multisector models overcome some of these problems. They distinguish between the traditional and modern sectors (for example, Parikh's model of India-Parikh 1976; Parikh and Srinivasan 1977), among several economic sectors (for example, Resources for the Future), and among sectors and end uses of energy (for example, the MEDEE model-Modele d'evolution de la demande d'energie). All of these approaches are useful because they are relatively simple to develop and understand, and they provide a better understanding of the energy system than one-sector models. But they are not without limitations. They ignore price effects, lack a macroeconomic forecasting framework, and do not provide sufficient treatment of inertias and time lags. One way to enhance the macroeconomic coherence of open-loop models is to use input-output (IO) tables. The World Bank's MSEDM (Minimum Standard Energy Demand Model), for example, uses 10 tables to convert fine] demand into gross output and demand for various energy forms.
Hierarchical Supply Model.
Linear programing models are the oldest optimization models. They were used initially to develop least-cost investment plans in the power sector (for example, the WASP-Wein Automatic System Planning Package-dynamic linear program). More recently, the models have been combined with IO tables to compute alternative and minimum-cost investment plans for the energy sector as a whole. Two limitations of these models are that the data used are more technological than economic and that IO tables embody fixed technical coefficients that must be (but are seldom) updated regularly.
Global models of supply and demand incorporate feedback from the energy sector (particularly price effects) into the macroeconomy. Global models of supply can be used to assess the price at which a given energy technology becomes feasible and to determine the impact of changes in energy prices on energy supply industries and on various macroeconomic indicators. Global models of demand (for example, Mukherjee's (1981) energy-economy model) combine demand functions with macroeconomic production functions to trace the effects of exogenous changes (for example, energy prices) on energy demand and economic variables (for example, capacity utilization, aggregate prices, and domestic energy prices). A feedback loop is used to start the next iteration, which determines the impact of energy prices on energy demand and the economy. The lack of an integrated energy demand-supply block, especially with regard to investment requirements, is the main limitation of these models. An extension of Mukherjee's model (SIMA, Simulation of Macroeconomic Scenarios to Assess Energy Demand) was an early attempt to build an integrated supply-demand tool.
Suggestions for Further Research
These models can be adapted for use in most developing countries. To enhance the applicability of the models, several things are needed:
· Better data collection because lack of data is the main problem in energy modeling,
· Improvements in the structure of demand models to include additional price variables, the traditional (noncommercial) sector, and emerging types of energy commodities,
· Use of a coherent set of exogenous values,
· Improvements in the models to include monetary and financial variables (for example, interest and exchange rates), and
· Expansion of the supply models to include noncommercial and renewable energy and the impact that the use of these forms of energy has on the environment.
Energy R&D has become a big business, especially since the price crises of 1973 and 1979. Therefore, it is essential, especially when national financial and human resources are scarce, to establish a logical and sensible approach to allocating funds and effort to energy R&D. This approach is facilitated in most developing countries because the impetus for R&D is the responsibility of the central government. This paper develops a rational and systematic framework, as well as relevant criteria, that can be used to help determine policies and priorities for R&D in any developing country.
The process for the development of an R&D policy for energy is straightforward and consists of five steps.
From a comprehensive list of energy R&D topics, select a subset that is most pertinent to the immediate needs, medium-term development goals, and long-term aspirations of the nation (in that order of priority). This step is guided by a thorough assessment of the country's resource endowment, its fiscal position and outlook, financial characteristics, and the technologies and options available.
Evaluate the topics in this subset using a simple formula that is designed to calculate the net expected benefits (or economic efficiency) of each topic:
NB = B(R&D) - C(R&D) + B(IMPLEM) - C(IMPLEM)
where NB is the calculated net benefit of the R&D undertaking, B(R&D) is the present discounted value (or present value, PV) (see endnote 1) of the stream of benefits accruing from the R&D undertaking, C(R&D) is the PV of the costs of the undertaking, B(IMPLEM) is the PV of the benefits derived from implementing the results of the R&D project, and C(IMPLEM) is the PV of the costs of implementation.
All costs and benefits are expressed in terms of their economic, not market, value. The "price" of a cost or benefit is its highest value when used for some purpose other than for the project in question. Care must be taken in the selection of appropriate values (often called "shadow prices") because costs are usually more readily identified than benefits and long times are involved.
Rank each R&D option according to its net expected benefit to the nation (economic efficiency). Because of the uncertainties inherent in each option, it is prudent to test the results of the efficiency calculation by altering an assumption on which a cost(s) or benefit(s) is based and then redoing the calculation. The most robust option over the range of scenarios should rank first.
Consider, in advance, the constraints to implementation of the R&D program. These can include any combination of
· Lack of financial, physical, or human (skills and training) resources,
· Institutional barriers (for example, a lack of specialized knowledge on the part of decision-makers or their advisors, an absence of coordination among the various levels of government or between the relevant government agents and the private sector, and inadequate links between the nation's policymakers and foreign or international sources of funds or expertise), and
· Policy constraints, engendered by a weak or nonexistent national energy plan or planning process.
Set in motion processes that can overcome these obstacles. Increase local contributions where potentially available (for example, from power utilities and oil, gas, and coal extraction companies, especially if these are government owned). Consider taxes on energy imports and allocate the proceeds to a fund for energy R&D. Appraise domestic and foreign financial institutions of the potential for short-term investment gains from preferential loans for R&D purposes. Investigate the availability of export assistance, grants, and loans from countries that possess some of the technology that is necessary to undertake the R&D or to implement the technology that is likely to be produced. Human resources must be protected by offering attractive salaries or other perquisites. As well, skilled personnel require access to information and equipment to carry out their work. Providing these inputs may be more efficient in the long run than purchasing the skills and resources of foreign consultants.
A lack of understanding of the relevant energy issues can be overcome with education campaigns aimed at both decision-makers and members of the energy-consuming public. Interactions of the national government with foreign governments, international aid institutions, and multinational enterprises and organizations must be established, strengthened, and stabilized.
The coordination of policy and planning must begin with a conscious effort to bring together the various levels of government that are involved in energy planning and related areas (the environment, financial institutions, and fiscal, monetary, and budget policy).
1. Calculated as follows:
where B(R&D) represents the benefits of the R&D activity in year t, T is the overall time horizon in number of years, and r is the selected discount rate.
This paper sets out the parameters of a general approach to establishing a country-specific tool for energy policy and planning. This is a large undertaking that requires a substantial commitment of national resources. But, unlike specialized consultants who provide similar information on a "one-off" basis, the resulting policy and planning tool will contribute to the long-term effectiveness of government planning institutions and, hopefully, to the efficiency of the economy.
Policy analysts and energy planners are increasingly aware that the real problem when developing information systems for energy planning is not the modeling system. Rather, the challenge is to find the most appropriate resolution, system structure, and configuration for the problem and to choose the most appropriate objective function, given the data available on energy supply and demand.
A number of "off the shelf" evaluation models have been developed since the oil crises highlighted the need for comprehensive energy planning. Although these models have helped industrialized nations make decisions on their energy policies, repeated efforts to fit the special energy problems of developing nations to these models have been futile. Variability in energy systems, energy decision functions, and available information effectively limit the application of generalized systems to data management and core computer packages (for example, regression analysis, linear programing, and matrix inversion).
Consequently, the best approach when examining the situation in any given developing country is, in effect, to build a tool for energy information, policy, and planning from the ground up. The three key components of this tool are management, information, and modeling.
The most efficient and productive way to set up an information system is to involve future users of the tool in every stage of its construction. This includes the critically important, but admittedly tedious, task of gathering microlevel information (consumer surveys and fieldwork), organizing the data into a useable (computer compatible) form, and defining the model that will analyze the data and produce report-ready output.
The energy planning tool must be managed according to the specific circumstances of the country to ensure that it is used to its maximum advantage. First, the size of the energy delivery system(s) determines, to a large extent, the degree of risk involved in committing resources to any project. Small systems take on more risk than large systems when a single, large project is added. Integrated systems face smaller risks when any kind of new supply is added. Second, the information produced by the planning tool must be in a form that meshes with the decision-making structure. If local energy planning is established, planning and policy information in the national context may be of little use. Third, the possibility of fuel substitution and conservation must be taken into account (with a sufficiently high degree of resolution) to ensure that the information provided by planners is supplied in the appropriate context.
The Information Set
The lack of information for energy planning in developing countries is a major handicap. Energy systems rely largely on traditional fuels, and their supply and demand vary from location to location across the country. The energy market can, in any given country, be fragmented into several regions, which makes aggregation to the national level difficult and potentially meaningless for policy and planning purposes. Therefore, it is necessary to undertake regional analyses of energy balances (consumption and production) and to orient policymaking accordingly. Because great detail is involved in this approach, much effort, in terms of both person hours and financial resources, is required.
Data on energy consumption can only be acquired by ground-level surveys. This is because, depending on the nation in question, up to 95% of primary energy consumption is of fuelwood or other traditional energy sources. The "market" for these energy resources is almost always local, and it must be estimated from local indicators of supply and demand. For this survey to be representative, a large number of households and fuel suppliers must be included.
Data on energy production (including imports) are often equally difficult to obtain, but advances in technology have improved their reliability and coverage. For example, satellite maps that depict the location of biomass resources (on which so much of the developing world relies) have been available since the early 1970s. The information on these maps must be locally calibrated, which involves ground-level verification, often over vast expanses of territory. The wealth of information made available is well worth the effort.
Once the data are collected, they must be organized and used as soon as possible. If too much time passes, the risk increases that when the information is finally analyzed it will yield results that are out of date when compared with the current, constantly evolving local, regional, national, and global energy markets.
Supply -The general rule when selecting from among various supply options is that the net present value (NPV) of benefits accruing from the decision option is positive (see endnote 1). There are, in addition, two evaluation approaches that are more financial (rather than economic) in perspective. The "payback period" (PBT) assesses the relative profitability of a project, which is defined as the number of years required to recover the investment costs of a project from its cash flow. In NPV terms, this is the number of years required to achieve a net present value of zero (see endnote 2). This approach is only useful if calculated paybacks are short (2 years or less indicates a good project) or very long (20 years or more indicates a poor investment). Uncertainty over intermediate times means that a more sophisticated tool must be used.
When the service produced by the investment is a fuel or energy flow, the "levelized cost" of the energy produced (or consumption avoided) can be computed. This is the "price" that must be charged to recover the total cash flow within the lifetime of the project. The levelized cost is the product of the capital recovery factor (CRF) and the present value of total costs (PVC) (see endnote 3).
Most project evaluations are based on the "internal rate of return" (IRR), which, like payback period, is calculated using the concept of discounting. The IRR is the discount rate that, when applied to a stream of benefits and costs reflected in the cash flow of a project, produces a NPV of zero (see endnote 4). Projects are generally acceptable when the IRR is greater than the accounting rate of interest (banking rate). The greater the rate of return, the more acceptable the project because a greater NPV of benefits will accrue from the investment.
Demand-Attempts to model energy decision-making by households have generally used econometric techniques. These methods are useful, but they can oversimplify actual consumer decision-making, especially in a dynamic context. A more appropriate approach might be to treat consumer decision-making as a simple accounting problem, in which consumers base their energy decisions on the capital cost of equipment (including subsidies and other policy incentives), the relative out-of-pocket costs of fuel alternatives, and an appropriate discount rate. Aggregation of the demand functions for a given population yields a target for the purpose of planning energy supply.
These classical approaches to project evaluation may appear simple. However, in practice they require a comprehensive understanding of the environmental, technical, economic, and financial factors that can affect the outcome of a decision and its long-term consequences.
1. Calculated as follows:
where B is benefits in year t, C is costs in year t, T is the overall time horizon in number of years, and r is the selected discount rate.
2. Calculated as follows:
where B is the benefit of the project in year t, C is the cost in that year, PBT is the payback time in years, and r is the selected discount rate.
CRF= r /[ (1 - (1 + r)-I ], where I is the lifetime of the project, and
where TC is the total cost in year t, T is the overall time horizon in years, and r is the selected discount rate.
4. Calculated as follows:
where B. C, T. t, and r are as defined in endnote 1, and IRR is the internal rate of return.