Question: An electric utility company faces the following forecasting problem. Because of restrictions set by state regulatory agencies, it is not profitable to build another greenfield plant. Averaged over the year, it has adequate excess capacity but faces a peak-load problem. On average, peak demand is about the same during cold winter days and hot summer days, but peaks in individual years depend on the weather, which cannot be predicted in advance. If capacity is not adequate, the company must buy power on the deregulated free market at five times the normal cost. You are asked to design a forecasting model to determine whether it is better to do routine maintenance during the winter or summer months.
(a) How would you use past historical data to determine whether excess demands are more likely to occur in the winter or summer, given that weather forecasts are random?
(b) How would you determine whether the price elasticity for electric power is higher in the winter or in the summer? How would that affect the model?
(c) How would these results be modified if "global warming "were significant? Assuming that the utility had historical data on degree-days for the region it serves, what type of tests would you use to determine whether or not the trend was significant?