gets (version 0.3)

isatvar: Variance of the coefficient path

Description

Takes a 'gets' object returned by the isat function as input and returns the coefficient path of the constant (and long-run equilibrium if 'lr' is specified) together with its approximate variance and standard errors.

Usage

isatvar(x, lr=FALSE)

Arguments

x

a 'gets' object obtained with the isat function

lr

logical. If TRUE and 'x' contains autoregressive elements, then isatvar also returns the long-run equilibrium coefficient path with its variance and standard deviation. See Pretis (2015).

Value

• If lr=FALSE: A Tx4 matrix (with T = number of observations) where the first column denotes the coefficient path relative to the full sample coefficient, the second column the coefficient path of the intercept, the third the approximate variance of the coefficient path, and the fourth column the approximate standard errors of the coefficient path. If lr=TRUE: A Tx7 matrix where the first four columns are identical to the lr=FALSE case, and the additional columns denote the long-run equilibrium coefficient path, together with the approximate variance and standard errors of the long-run equilibrium coefficient path.

Details

The function computes the approximate variance and standard errors of the intercept term with structural breaks determined by isat. This permits hypothesis testing and plotting of approximate confidence intervals for the intercept in the presence of structural breaks. For dynamic autoregressive models in isat the lr argument returns the time-varying long-run equilibrium together with its approximate variance and standard errors.

References

Pretis, F. (2015): 'Testing for time-varying predictive accuracy using bias-corrected indicator saturation'. Oxford Department of Economics Discussion Paper.

See Also

isat, coef.gets, plot.gets, biascorr, isattest

Examples

##Variance in presence of a break
#nile <- as.zoo(Nile)
#isat.nile <- isat(nile, sis=TRUE, iis=FALSE, plot=FALSE, t.pval=0.005)
#var <- isatvar(isat.nile)

#plot(nile)
#lines(isat.nile$mean.fit, col="red")
#lines(isat.nile$mean.fit + 2*var$const.se, col="blue", lty=3)
#lines(isat.nile$mean.fit - 2*var$const.se, col="blue", lty=3)

##Variance when there is no break
#set.seed(1)
#x <- as.zoo(rnorm(100, 0, 1))
#isat.x <- isat(x, sis=TRUE, iis=FALSE, plot=TRUE, t.pval=0.005)
#var.x <- isatvar(isat.x)

#plot(x)
#lines(isat.x$mean.fit, col="red")
#lines(isat.x$mean.fit + 2*var.x[,2], col="blue", lty=3)
#lines(isat.x$mean.fit - 2*var.x[,2], col="blue", lty=3)

##Variance of the long-run equilibrium coefficient path

#nile <- as.zoo(Nile)
#isat.nile <- isat(nile, sis=TRUE, iis=FALSE, plot=TRUE, t.pval=0.005, ar=1:2)
#var <- isatvar(isat.nile, lr=TRUE)