Laserfiche WebLink
<br /> <br />Administration (DOE) Midterm Forecasting system3,4 incorporate aspects <br />of several modeling approaches. Often one methodology is utilized to <br />develop input parameters for a model structured by another methodology. <br />Simulation modeling requires the creation of a mathematical descrip- <br />tion of the system being modeled. To the extent that the modeler has <br />accurately determined future relations between the exogenous and endoge- <br />nous variables of his mathematical description, the model's output will <br />forecast the future. <br />Models are frequently more valuable for the understanding they cast <br />on the interrelationships of various factors than on the absolute magnitude <br />of any results. In the area of economic modeling, an inherent flaw is the <br />inability to predict and account for major shocks to the system such as <br />wars, depressions, and technological breakthroughs.5 <br />Though not a flaw, a factor which must be considered in accepting <br />the results of a model is that the use of econometric or input-output <br />(I/O) model price projections requires that numerical relationships be <br />established between prices and factors of production, consumption, and <br />cost. An examination of the extended historical price and production <br />curves of the three primary energy commodities, presented in Figure 2-1, <br />tends to support the conclusion of others that there is no simple direct <br />relationship between primary energy commodity production and prices. <br />That is, there is no long-term price-supply elasticity. 1 ,6,7 As a recent <br />study states, "One of the vexing tasks of energy projections [is] to <br />extract from a history of price declines the basis for judging the demand <br />response to possible price increases in the future.,,8 <br />Often greater confidence is ascribed to the results produced by a <br />highly detailed mathematical model than is warranted by the joint proba- <br />bility associated with the inputs and established relationships. There is a <br />danger in blindly using such methodologies for long-term projections <br />because of this implicit modeling uncertainty, as well as the model's <br />inability to correctly predict major "shocks" to the system. <br /> <br />2-2 <br />