ols_mv: Multivariate ordinary least squares for VARs

In nielsaka/zeitreihe: Simulate, Estimate, Select, and Forecast Multiple Time Series Processes

Description

Perform multivariate OLS on a set of observables. Creation of lags for the VAR will be handled by the function.

Usage

ols_mv(Y, p, const = TRUE)

Arguments

Y	A (K x N+p) matrix carrying the data for estimation. There are N observations for each of the K variables with p pre-sample values.
p	An integer scalar. The lag length of the VAR(p) system.
const	A boolean scalar, indicating wether a constant should be included. Defaults to TRUE.

Details

This routine applies equation-wise OLS to a VAR. It is equivalent to running OLS with p lags for each of the variables and collecting the coefficients and residuals in a matrix.

Value

A list with four elements:

• BETA.hat is a (K x [K * p + 1]) or (K x [K * p]) named matrix containing the coefficient estimates. Its dimension depends on wether a constant is included. Each column contains the coefficents for a particular regressor, each row corresponds to a single equation. If Y did not carry names, the variables will be named y1 and counting. The names of the lagged regressors are derived by appending the variable name with a .l1 for lag one etc. If a constant is included, its coefficients are in the first column.

• SIGMA.hat is a (K x K) named matrix. It is the covariance matrix of the residuals. The columns and rows are named after the variables in Y.

• U.hat is a (K x N) matrix of residuals. Its rows are named after the variables, too.

• std.err is a matrix of the same dimension and naming scheme as BETA.hat. It carries the standard errors of the coefficient estimates.

nielsaka/zeitreihe documentation built on March 17, 2020, 8:38 p.m.