Title: A simple Markovian process with hyperbolic rates of convergence.
Authors: Alessandro Palandri - Dublin City University (Ireland) [presenting]
Abstract: A Markovian process is introduced with hyperbolic reversion rates to the fixed point. Proofs are provided regarding the uniqueness of the fixed point, its domain of attraction and the resulting uniform ergodicity. Two distinct lag-p generalizations of the baseline lag-one specification are presented alongside their properties. The proposed process is compared to standard AR(p) auto-regressions: unit-root tests, predictions, forecasts and the ability to capture various stylized facts. Furthermore, the proposed functional form is applied to discrete-time models of conditional variances resulting into a new GARCH specication as well as a new specication for the modeling of realized measures of variance.