Furthermore, the parameter estimates are asymptotically
unbiased and normally distributed. In contrast, the OLS
estimator of (1) while super-consistent is not asymptoti-
cally unbiased or normally distributed and its finite sample
bias may be large and persistent. Therefore pass-through
estimates obtained from the Engle–Granger procedure
may be far from their true value in finite samples. The bias
in the long-run coefficient estimates will in turn carry
through to the estimate of the error correction term used
in the analysis of the short-run dynamics of the monetary
transmission mechanism. Similarly, estimates of the long-
run parameters obtained via the Johansen method may
not be robust in situations where the equilibrium error is
persistent.
đang được dịch, vui lòng đợi..