PNNL

Abstract

Technology is the major driving force of productivity gains and economic growth. Historical studies throughout the last decade attributed about half of economic growth to technological change and the other half to the combined effect of all other driving forces, such as the larger and better-qualified labor force and accumulated stock of capital. V. Peterka (1977) was one of the first to break new ground exploring the mathematical methods for forecasting market shares of competing technologies. Having information about the historical market shares of competing technologies, Peterka described a unique set of algebraic derivations in the MLEST Model. Since Peterka, a large number of studies described the theoretical basis for technology diffusion, however very few provided a real world examples or verified the applicability to the market diffusion theory for energy applications. Objective: The objective of this study is to provide an example of projecting market shares of competing technologies using maximum likelihood estimation (MLE). The application applies to the emergence of the electronic ballast for fluorescent lighting applications. In this example we model the historical competition between the existing technology (magnetic ballasts) and the emerging technology (electronic ballasts). The factors surrounding electronic ballasts as replacement for the magnetic ballasts provide a rich example of competing technologies. The lessons from this example could be used to inform forecasting of many other similar technologies penetrating the market in the U.S. energy sector (e.g. fuel cells, digital information and communication technologies ICTs ...etc). The example will forecast the market shares of magnetic and electronic ballast technologies to 2020 and discuss the energy savings and other benefits. This type of modeling and analysis can help inform the rule-making process for any potential future DOE standards for electronic ballasts. Method: Kennedy, Peter (1993) listed several desired properties for using MLE when compared to other statistical techniques such as the Ordinary Least Square (OLS) in forecasting including that MLE estimates are unbiased, consistent, and asymptotically efficient. The model introduces a new technology at certain point of time using some initially small (