Abstract
The authors develop a dynamic hedonic regression approach to measuring the evolution of a comparative brand premium (pairwise price difference between two products that are identical in all respects apart from the brand). In contrast to existing approaches, the proposed Bayesian estimation method exploits the premia's intertemporal dependence structure, resulting in a higher level of accuracy of the estimated time paths of the brand premia. In addition, the authors present a novel, but straightforward way to construct confidence bands that cover the entire time series of brand premia with high probability. The authors apply their approach to a large, detailed data set on laser printers, which was gathered on a monthly basis over a 4-year period.
Acknowledgements
The authors are indebted to Jan König, Andy Zuchandke, Sebastian Biermann, Fabienne Décieux, and Kerstin Weber for excellent research assistance, and to Dr Brian Bloch for his comprehensive editing of the manuscript. Two anonymous referees provided many invaluable comments. Any remaining errors and omissions, of course, are our own.