Published in Romanian Journal of Economic Forecasting,
Requires a PDF viewer such as Xpdf
In this study, we set the problem of the probable existence of an additive and multiplicative mixed seasonality. In this context, we show by some simulation that the seasonality correction according to a pure additive or a pure multiplicative scheme leads to biased estimators of the coefficients and, consequently, of the calculation of seasonally adjusted series which is necessary for quantitative demand analysis.
The use of an analytical resolution technique allowing simultaneous estimation of the trend coefficients and the additive and multiplicative seasonal coefficients works perfectly if the series is affected by a simple linear trend. In this case, the estimation gives the theoretical seasonal coefficients.
An application study to the mobile phone market, the new product split in two markets, professional and individuals, allows evaluating the contribution of the methodology.
seasonality, demand analysis, time series
JEL Classification: C32, C12