Speaker: Professor Peter Boswijk - University of Amsterdam
Series: Economics Department Seminar Series
Economics Department Seminar
Given the well-established fact that many key macro-economic and financial variables are subject to permanent changes in unconditional volatility, in this paper we consider estimation and hypothesis testing on the cointegrating relations and adjustment coefficients in vector autoregressions with non-stationary (unconditional) volatility of a very general form, which includes single and multiple volatility breaks as special cases.
We show that the conventional results in Johansen (1996), i.e., the Gaussian maximum likelihood estimators of the cointegrating vectors and adjustment coefficients are (mixed) normal, with the associated likelihood ratio tests for linear restrictions being asymptotically chi-squared, break down under permanent volatility changes.
As a consequence, standard confidence intervals and tests of hypothesis are potentially unreliable. As a solution, we propose wild bootstrap inference methods which do not require the practitioner to specify a parametric model for volatility, nor to assume that the pattern of volatility is common to, or independent across, the vector of series under analysis. We formally establish that the wild bootstrap allows to replicate the relevant asymptotic distributions and that it has very good finite sample properties under a wide range of volatility models. An application to the term structure of interest rates in the US illustrates the difference between standard and bootstrap inferences regarding hypotheses on the cointegrating vectors and adjustment coefficients.
Speaker's details can be found here.
Share this event
When & where
5.00pm - 6.20pmWednesday 5th June 2013