Econometric Advances in Diffusion Models


Speaker


Abstract

Although diffusion models have been researched extensively, this thesis gives new and important insights for modeling new product diffusion. In particular, in the past decade the focus has shifted from studying one diffusion series to the comparison of multiple series. Such a comparison allows one to learn from the differences and similarities across products and/or countries. Additionally, the most common observational frequency has shifted from annual to monthly or quarterly, and in some cases data is available at an even higher frequency. Both developments lead to new challenges in the modeling of diffusion processes.

In this thesis we confront these challenges. We provide new models and estimation techniques. We also give some new perspectives on known issues in diffusion modeling. In the first chapters we deal with the estimation of diffusion parameters for a single series. We start with an overview of existing estimation methods and we suggest a new method, where we study the possible bias in the diffusion parameters. We specifically consider the source of noise and observational frequency. Second, we look at the challenges arising due to using high-frequency diffusion data, that is, mixed-frequency diffusion data and seasonality. Finally, we come back to the comparison of many diffusion series, where we use a model to explain the variation in the diffusion patterns across products and countries.

 
Contact information:
Marisa van Iperen
Email