var.lag.specification {MSBVAR} | R Documentation |
Estimates a series of test statistics and measures for VAR lag length selection.
var.lag.specification(y, lagmax = 20)
y |
T x m multiple time series |
lagmax |
Maximum lag order to be evaluated. Function will return lag length tests for all lag orders less than lagmax. |
Estimates a series of frequentist VAR models for 1 to lagmax and returns a sequence of Chi2 tests, AIC, BIC and Hannan-Quinn criterion values for each lag length.
Results are printed to standard output (screen or file). In addition, a list of two matrices is returned:
ldets |
Lag length, log-determinants, Chi2 tests, and p-values for each lag length, compared to the null of the next shorter lag length |
results |
Lag length, AIC, BIC, and HQ criteria for each lag length. Selection criteria should be minimized. |
Sizes of p-values are uncorrected for multiple testing. Use cautiously.
Patrick T. Brandt
Lutkepohl, Helmut 2004."Vector Autoregressive and Vector Error Correction Models", Chapter 3. In Applied Time Series Econometrics. Lutkepohl,, Helmut and Markus Krtzig eds. Cambridge: CUP.
See Also reduced.form.var
for frequentist VAR estimation,
szbvar
for Bayesian VAR estimation, and
szbsvar
for Bayesian Structural VAR estimation.
data(IsraelPalestineConflict) var.lag.specification(IsraelPalestineConflict, lagmax=12)