Constructing Demand Response Models for Electronic Power

Economic models should be based on real data if possible, and one of the most extensive data sources for energy consumption is the US government's Residential Energy Consumption Survey (RECS). The survey results indicate what terms are most important, and they provide much of the data necessary to fit parameters of a demand function, but they neglect seasonal variations in prices and heating and cooling requirements. With some difficulty, weather information and seasonal pricing variations from other sources can be merged with RECS data. A further complication is the need for monthly data and for cooling and heating degree data relative to various base temperatures. We deal with these issues, explore various demand functions, and use nonlinear least squares to fit their parameters to the data.