Foreign Exchange Value-at-Risk with Multiple Currency Exposure: A Multivariate and Copula Generalized Autoregressive Conditional Heteroskedasticity Approach
Abstract
Large DND projects and acquisitions are exposed to more than one foreign currency at the same time which complicates managements foreign exchange risk assessments. We extend the Centre for Operational Research and Analysis (CORA) in-house Generalized Autoregressive Conditional Heteroskedasticity (GARCH) models to a full multivariate setting. Our extensions involve two models types: multivariate GARCH and copula-GARCH. We find that both models give qualitatively similar value-at-risk (VaR) estimates, and that both models provide a much improved risk assessment relative to the current practice-correcting VaR estimates on the order of 25% in cases in which multiple currency exposures are of similar size. Using the USDCAD, the EURCAD, and the GBPCAD, we demonstrate estimation techniques for each model. Finally, we show the strength of our improved models through a 100-day VaR calculation.
Document Details
- Document Type
- Technical Report
- Publication Date
- Nov 01, 2014
- Accession Number
- AD1003599
Entities
People
- David W. Maybury
Organizations
- Defence Research and Development Canada