Seasonal Heteroskedasticity in Time Series Data: Modeling, Estimation, and Testing
Thomas M. Trimbur and William R. Bell
KEY WORDS: seasonal adjustment, trend, unobserved component
ABSTRACT
Seasonal heteroskedasticity refers to regular changes in variability over the calendar year. Models for two different forms of seasonal heteroskedasticity were recently proposed by Proietti and by Bell. We examine use of likelihood ratio tests with the models to test for the presence of seasonal heteroskedasticity, and use of model comparison statistics (AIC) to compare the models and to search among alternative patterns of seasonal heteroskedasticity. We apply the models and tests to U.S. Census Bureau monthly time series of housing starts and building permits.
CITATION: Trimbur, Thomas M. and William R. Bell. Seasonal Heteroskedasticity in Time Series Data: Modeling, Estimation, and Testing. Statistical Research Division Research Report Series OR Study Series (SECTION #2008-NU). U.S. Census Bureau.
