Finally, once the priors are specified, we estimate the BVAR
model using Theil's (1971) mixed estimation technique.
In particular, their demonstration of the possibility of estimating a much larger and more richly specified DSGE model (similar to that developed by Christiano, Eichenbaum, and Evans 2005), as well as their finding of a good forecast performance of their DSGE model relative to competing vector autoregressive (VAR) and Bayesian VAR (BVAR
) models, led DSGE models to be taken more seriously by central bankers around the world.
Recently this modelling technique has been used to develop a 35 variable BVAR
for the New Zealand economy, details of which are available in Bloor and Matheson (2009).
We use the BVARs
estimated in the previous section to implement the methodology.
The log MDD value is about 27.50, which is substantially lower compared to monthly Bayesian vector autoregression (BVAR
) models of unemployment and inflation.
Sims and Zha (1998) extend this framework to BVARs
, that is, to VARs with priors.
 None of these forecasts turning points well, but neither do the BVARs
(see Wi, 2000).
Accordingly, I employ a BVAR
model described in Beauchemin (2011).
We highlight these gains in forecast accuracy in two different exercises using a medium-scale (18-variable) monthly BVAR
The root mean squared errors (RMSEs)--roughly equivalent to the typical forecasting "miss"--for the forecasts coming from the BVAR
without any wage measure are 0.7 percentage point at a four-quarter horizon and 1.0 percentage point at a twelve-quarter horizon.
The VAR and BVAR
estimations use all the nine observable series utilized in the benchmark estimation of the DSGE model.
The "Bayesian" part of BVAR
means that "extra-sample" beliefs held by the forecaster may be introduced into the forecast along with the degree of confidence that he or she attaches to these beliefs.