Time Series Concepts for Conditional Distributions*

The paper asks the question – as time series analysis moves from consideration of conditional mean values and variances to unconditional distributions, do some of the familiar concepts devised for the first two moments continue to be helpful in the more general area? Most seem to generalize fairly easy, such as the concepts of breaks, seasonality, trends and regime switching. Forecasting is more difficult, as forecasts become distributions, as do forecast errors. Persistence can be defined and also common factors by using the idea of a copula. Aggregation is more difficult but causality and controllability can be defined. The study of the time series of quantiles becomes more relevant.     

Additive and Multiplicative Uncertainty Revisited: What Explains the Contrasting Biases?

above the certainty level while for additive uncertainty the price should be lower than the certainty level. This note gives an intuitive explanation for the result after first presenting a parsimonious review of the two models. We also discuss which, if either, of the two models is more realistic. Received December 14, 2001; revised version received July 16, 2002 Published online: April 30, 2003 We thank referees for helpful comments. Ciaran Driver would like to acknowledge the research facilities from ANU, Canberra for their help in writing this paper.