var_ols: VAR model (least squares estimates)

In nathanesau/StocVal: Values embedded options in DB pension plan.

Description

First step of VAR model calibration. Number of state variables is equal to length(varData[1,]) - however, currently 4 is required since the pricingKernel is currently implemented to accept a 4 component VAR. Note that the MASS and vars packages are required.

Usage

var_ols(varData)

Arguments

varData	The VAR variables to use in the model
marketData	The data containing the VAR variables and term structure historical yields.

Details

When calibrating the VAR model, the mean is subtracted and we estimate (zt-u) not zt. Also note that a VAR(1) model is estimated. For example, marketData <- readRDS("~/Dropbox/Research/StocVal/data/Canada/varinput_canada.Rda"); varData <- data.frame(onemonth=market_data$onemonth, inflation=market_data$inflation, tenyear=market_data$tenyear, stock=market_data$stock); mu <- matrix(c(mean(varData$onemonth), mean(varData$inflation), mean(varData$tenyear), mean(varData$stock)), 4, 1); varData <- data.frame(onemonth=varData$onemonth - mu[1], inflation=varData$inflation - mu[2], tenyear=varData$tenyear-mu[3], stock=varData$stock - mu[4]); var_ols.out <- var_ols(varData, p=1, type="none")

Value

A list containing Phi (the coefficients) and Sigma (the covariance matrix)

nathanesau/StocVal documentation built on May 23, 2019, 12:18 p.m.