[Report#:DOE/EIA-0581(2000)]
To Forecasting Home Page |
Macroeconomic assessment at EIA involves several modes of analysis. The first type of analysis, used in forecasting the Annual Energy Outlook where energy prices change, uses kernel regression and response surface techniques to mimic the response of larger macroeconomic and industrial models. This mode of analysis requires a given economic baseline and then calculates the economic impacts of changing energy prices, calculated from the chosen growth path. The economic growth cases are derived from the larger core models and can reflect either high, low, or reference case growth assumptions. Analyzing economic impacts from energy price changes uses the macroeconomic activity module (MAM) within NEMS and provides a subset of the macroeconomic variables available in the larger core models. The composition of the subset is determined by the other energy modules in NEMS, as they use various macroeconomic concepts as assumptions to their particular energy model. The other type of macroeconomic analysis requires the use of the underlying core macroeconomic models separately. There are four core models, estimated by Standard and Poor’s/DRI (DRI), that serve as the basis for EIA’s macroeconomic assessment. These include: the U.S. Quarterly Model of the Economy, Personal Computer Model of Industrial Output, an Employment Model by Industry, and a Regional Model. All of these models are linked by crucial assumptions that are provided by one of the other three models. Using the core models in a standalone fashion requires that the energy market concepts contained in the core models show NEMS-like behavior. To accomplish the goal of duplicating NEMS-like behavior, all energy market concepts in the core models show the same percentage change from base of the corresponding NEMS energy market concepts. MAM provides forecasts of economic variables to the energy modules within NEMS and forecasts the impacts on the aggregate economy of changes in energy market conditions. MAM consists of three submodules (the national, interindustry, and employment submodules) and a subroutine (the regional subroutine), which are run in sequence. The national submodule provides forecasts of a wide range of economic variables at the national level. In particular, it provides variables that drive other parts of NEMS, including interest rates, final demands for goods and services, and disposable income. The interindustry submodule calculates the industrial output needed to satisfy the final demands from the national submodule. The levels of industrial output are used by the industrial and transportation demand modules in NEMS to calculate energy consumption in those sectors. The employment submodule calculates the levels of employment necessary to produce the output by industry. Finally, the regional subroutine of MAM takes national variable values from the national and interindustry submodules and transforms them into appropriate regional values. The overall interrelationships are shown in Figure 3. Figure 3. Macroeconomic Activity Module Structure The sequence of interactions in MAM begins with the energy supply and demand modules of NEMS, which determine the reaction of energy markets. These energy market effects are passed to the national submodule, and the economy is projected to react to the altered energy markets. The altered macroeconomic final demands are in turn passed to the interindustry submodule, which calculates their effect on interindustry activity. The altered interindustry projections are then used to determine employment levels. Macroeconomic Activity Module Table Macroeconomic concepts, industry gross output, and employment levels are passed back to the other NEMS modules, and the system iterates until convergence is achieved. MAM outputs include measures of macroeconomic performance and manufacturing and nonmanufact- uring sector production activities. Depending on the concept, these measures are provided nationally or at the geographic level of the nine Census divisions (see Figure 1). National Submodule The national submodule of MAM is a kernel regression representation of the DRI U.S. Quarterly Model. The kernel regression approach relies upon databases containing simulation results of the DRI U.S. Quarterly Macroeconomic Model. The simulations represent different assumptions about the price of crude oil and about policy instruments. The databases are linked so that the input variables (i.e. tax collections and energy prices and quantities) from each simulation are associated with their respective output variables (i.e. macroeconomic concepts.) The kernel regression representation simulates the reaction to changing energy markets of the larger version of the DRI U.S. Quarterly Model. The growth potential of the economy is essentially grounded in the growth of the factors of production—labor, capital, and energy—and the aggregate productivity of these factors. The user can select three different macroeconomic growth cases before executing MAM, and MAM will estimate how the economy changes in response to changing energy markets. These growth cases are initially derived from simulations of the DRI U.S. Quarterly Model and are used as the starting point to examine energy/economic impacts. Approximately 100 economic variables from the DRI simulations are used by NEMS to define an economic growth case. Interindustry Submodule The interindustry submodule provides industrial output projections to the regional MAM subroutine and the energy modules. It also calculates interindustry impacts associated with changes in national economic activity resulting from energy market changes. A response surface version of the DRI Input-Output Model for the Personal Computer (PCIO) constitutes the core interindustry segment of this submodule, which is fully linked to other NEMS components. The interindustry submodule calculates deviations from a given baseline interindustry projection whenever macroeconomic final demands change. Because of the top-down structure of input-output modeling, the interindustry and national submodules do not iterate directly with each other. However, through their joint effect on the projections of the energy supply and demand modules, which in turn alter the macroeconomic outlook, changes in interindustry projections affect the results of the national submodule. Employment Submodule The employment submodule provides employment projections to the regional MAM subroutine and the energy modules. It also calculates employment impacts associated with changes in national industrial output resulting from energy price changes. A response surface version of the DRI Employment Model by Industry constitutes the core employment segment of the submodule, which is fully linked to other NEMS components. Regional Subroutine The initial implementation of the regional subroutine is a set of equations that share national economic and industrial output data to specific regions. The sharing equations are derived from the Regional Macroeconomic Model developed by DRI. The regional subroutine disaggregates national economic variables such as disposable income, industrial production, and consumer prices into appropriate regional values. Use of the Underlying Core Models MAM estimates macroeconomic impacts of energy market changes. In order to estimate these energy market impacts, an underlying baseline has to be provided. As the NEMS energy inputs change, MAM estimates the resulting changes to the macroeconomic variables provided in the baseline. The underlying core models are used in two ways: to provide economic baselines upon which percentage change impacts are calculated and to provide kernel regression and response surface representations which mimic the behavior of the larger scale structural models. The core models are provided by DRI and include their U.S. Quarterly Model of the Economy, PCIO Industrial Output Model, an Employment Model by Industry, and a Regional Economic Model. All four of these models are used in order to provide the macroeconomic growth cases (the high economic growth, reference case, and low economic growth). |
