We provide various estimates of temperature effect for accommodating seasonality in energy demand, particularly natural gas demand. We exploit temperature response and monthly temperature distribution to estimate the temperature effect on natural gas demand.

Both local and global smoothed temperature responses are estimated from empirical relationship between hourly temperature and hourly energy comsumption data during the sample period (1990-1996). Monthly temperature distribution estimates are obtained by kernel density estimation from temperature disperation within a month.

We integrate temperature response and monthly temperature density overall the temperatures in the sample period to estimate temperature effect on energy demand. Then, estimates of temperature effect are compared between global and local smoothing methods.