Seasonality refers to the cyclical and recurring fluctuation around the mean of a time-series for a given period of time, usually a calendar year. It can be deterministic (e.g. Christmas is in December every year) or stochastic (like fuel consumption during the winter which also depends on the harshness of the winter). Seasonal patterns can change markedly as consequence of evolutions in business and society. For instance, the invention of air conditioning lead to a second peak of residential electricity consumption in the U.S. aside from the increased consumption during winter.
Seasonality in regression analysis is usually taken into account by including quarterly or monthly dummies depending on the time-frequency of the data. Also specific recurring events like holidays or Christmas can be considered with dummy variables which again depends on the application at hand.
For an overview of the topic see also
- Bell, W.R. et al. (2012) "Economic Time Series Modeling and Seasonality", Taylor & Francis Group, LLC, USA
- Ghysels, E. and Osborn, D.R. (2001) "The Econometric Analysis of Seasonal Time Series", Cambridge University Press, Cambridge, UK