ABSTRACT
This chapter presents popularity of unrestricted Vector autoregressions (VARs) and explains variants of VARs, such as Bayesian or structural VARs, by their extensive adoption in macroeconomic forecasting and finance. For our City of London VAR, it considers both the Akaike information criterion and the Schwarz information criterion. The chapter summarises the output the multivariate tests and contains complete presentation both of the univariate and multivariate tests along with guidance how to perform the tests in EViews. An alternative method to study the effects of shocks to the variables in the system and therefore their dynamic relationship is variance decomposition. Forecasts for each variable are obtained by the VAR and the mean squared errors are calculated. The VAR framework can be used as a means of running Granger causality tests to establish whether movements in one variable precede movements in another.
