Addressing an Uncertain Future Using Scenario Analysis

Publication Type


Date Published


LBNL Report Number



The Office of Energy Efficiency and Renewable Energy (EERE) has had a longstanding goal of introducing uncertainty into the analysis it routinely conducts in compliance with the Government Performance and Results Act (GPRA) and for strategic management purposes. The need to introduce some treatment of uncertainty arises both because it would be good general management practice, and because intuitively many of the technologies under development by EERE have a considerable advantage in an uncertain world. For example, an expected kWh output from a wind generator in a future year, which is not exposed to volatile and unpredictable fuel prices, should be truly worth more than an equivalent kWh from an alternative fossil fuel fired technology. Indeed, analysts have attempted to measure this value by comparing the prices observed in fixed-price natural gas contracts compared to ones in which buyers are exposed to market prices (see Bolinger, Wiser, and Golove and (2004)). In addition to the routine reasons for exploring uncertainty given above, the history of energy markets appears to have exhibited infrequent, but troubling, regime shifts, i.e., historic turning points at which the center of gravity or fundamental nature of the system appears to have abruptly shifted. The purpose of this paper is to explore scenario analysis as a method for introducing uncertainty into EERE's forecasting in a manner. The two questions are how could it be done, and what is its academic basis, if any. Despite the interest in uncertainty methods, applying them poses some major hurdles because of the heavy reliance of EERE on forecasting tools that are deterministic in nature, such as the Energy Information Administration's (EIA's) National Energy Modeling System (NEMS). NEMS is the source of the influential Annual Energy Outlook whose business-as-usual (BAU) case, the Reference Case, forms the baseline for most of the U.S. energy policy discussion. NEMS is an optimizing model because: 1. it iterates to an equilibrium among modules representing the supply, demand, and energy conversion subsectors; and 2. several subsectoral models are individually solved using linear programs (LP). Consequently, it is deeply rooted in the recent past and any effort to simulate the consequences of a major regime shift as depicted in Figure 1 must come by applying an exogenously specified scenario. And, more generally, simulating futures that lie outside of our recent historic experience, even if they do not include regime switches suggest some form of scenario approach. At the same time, the statistical validity of scenarios that deviate significantly outside the ranges of historic inputs should be questioned.

Year of Publication







Research Areas: