- Advances in Regulatory Economics series
Edited by Michael A. Crew and Paul R. Kleindorfer
Chapter 5: Forecast Uncertainty in Dynamic Models: An Application to the Demand for Mail
* Catherine Cazals, Jean-Pierre Florens, Frank Rodriguez and Soterios Soteri 1. INTRODUCTION National postal operators use a variety of techniques to generate projections of the mail market to inform ﬁnancial and strategic planning decisions. In the UK, for example, Royal Mail uses econometric time series models to produce business projections and to provide a framework to analyse and understand the evolving nature of the demand for mail in the UK (Nankervis et al., 2002). Other national postal operators, such as Finland Post and the United States Postal Service, also possess detailed econometric models which they use extensively for projection and scenario analysis purposes. However, while the reporting and use of such models has tended to be reasonably well documented within the postal economics literature,1 relatively little quantitative analysis has been undertaken on the uncertainty surrounding volume projections. Understanding the nature and extent of mail volume uncertainty is necessary to assess appropriately business and policy-related risks. This chapter attempts to bridge this information gap. In particular, we use an econometric time series model of the demand for mail to identify potential sources of model-based projection errors and via the use of Monte Carlo simulation techniques obtain quantitative estimates of the uncertainty surrounding such projections. The results provide a number of insights into the level of uncertainty surrounding projections of mail volumes using time series econometric models. We proceed as follows. Section 2 provides an overview of the nature of forecasting errors that could arise from time series econometric models. Section 3...
You are not authenticated to view the full text of this chapter or article.