Testing for Causality in Variance Using Multivariate GARCH Models
Year: 2008
Type: article
Abstract: Tests of causality in variance in multiple time series\nhave been proposed recently, based on residuals of estimated\nunivariate models. Although such tests are applied frequently\nlittle is known about their power properties. In this paper we\nshow that a convenient alternative to residual based testing is to\nspecify a multivariate volatility model, such as multivariate\nGARCH (or BEKK), and construct a Wald test on noncausality in\nvariance. We compare both approaches to testing causality in\nvariance in terms of asymptotic and finite sample properties. The\nWald test is shown to have superior power properties under a\nsequence of local alternatives. Furthermore, we show by simulation\nthat the Wald test is quite robust to misspecification of the\norder of the BEKK model, but that empirical power decreases\nsubstantially when asymmetries in volatility are ignored.
DOI: 10.2307/27715168
Language: French
Subfield: Finance
Domain: Social Sciences
Open Access status: green
OpenAlex ID: W3147749544