estimate.par: Estimates the parameters of a partially autoregressive fit using lagged variances
Description
Estimates the parameters of a partially autoregressive fit using lagged variances
Usage
estimate.par(X, useR = FALSE, rho.max = 1)
Arguments
X
A numeric vector or zoo vector representing the time series whose parameters
are to be estimated
useR
If TRUE, the estimation is performed using R code. If FALSE, the
estimation is performed using a faster C++ implementation. Default: FALSE.
rho.max
An artificial upper bound to be imposed on the value of rho.
Value
Returns a numeric vector containing three named components
rhoThe estimated value of rho
sigma_MThe estimated value of sigma_M
sigma_RThe estimated value of sigma_R
Details
The method of lagged variances provides an analytical formula for the parameter
estimates in terms of the variances of the lags $X[t+1] - X[t]$,
$X[t+2] - X[t]$ and $X[t+3] - X[t]$. Let
$$V[k] = var(X[t+k] - X[t]).$$
Then, the estimated parameter values are given by the following formulas:
$$rho = -(V[1] - 2 V[2] + V[3]) / (2 V[1] - V[2])$$
$$sigma_M^2 = (1/2) ((rho + 1)/(rho - 1)) (V[2] - 2 V[1])$$
$$sigma_R^2 = (1/2) (V[2] - 2 sigma_M^2)$$
References
Clegg, Matthew.
Modeling Time Series with Both Permanent and Transient Components
using the Partially Autoregressive Model.
Available at SSRN: http://ssrn.com/abstract=2556957