Modelling seasonality in innovation diffusion. A regressive approach

Although the literature on innovation diffusion is very wide, up to now not much attention has been paid on modelling the seasonal component often present in sales data. In this paper we develop two innovation diffusion models that take into account not only the evolutionary trend of sales but also their intra-year oscillations, that are due to seasonal effects. In particular, we treat seasonality as a deterministic component to be estimated with conventional NLS techniques jointly with the trend. The results obtained by applying these models to the life-cycle of a pharmaceutical drug show a clearly better performance in providing short-term forecasts. Moreover, this methodology proves more parsimonious with respect to the autoregressive method, based on SARMA models.