Maximilian Rausch
2010-Apr-14 18:34 UTC
[R] ur.df ADF Unit Root Test: what is the meaning of phi1 and phi2 test statistic?
Hello,
I am using the ur.df function from the {arca} package to run the augmented
Dickey-Fuller unit root test on several time series. However; I do not
understand the econometric interpretation of the the "phi1" and
"phi2"
test-statisitc which are output if you choose a "trend" or
"drift" model. I
looked at the source code for the function but I do not quite understand the
code (which I have included below). Any help would be much appreciate.
Thanks,
Max
if (type == "drift") {
result <- lm(z.diff ~ z.lag.1 + 1)
phi1.reg <- lm(z.diff ~ -1)
phi1 <- anova(phi1.reg, result)$F[2]
tau <- coef(summary(result))[2, 3]
teststat <- as.matrix(t(c(tau, phi1)))
colnames(teststat) <- c("tau2", "phi1")
}
if (type == "trend") {
result <- lm(z.diff ~ z.lag.1 + 1 + tt)
phi2.reg <- lm(z.diff ~ -1)
phi3.reg <- lm(z.diff ~ 1)
phi2 <- anova(phi2.reg, result)$F[2]
phi3 <- anova(phi3.reg, result)$F[2]
tau <- coef(summary(result))[2, 3]
teststat <- as.matrix(t(c(tau, phi2, phi3)))
colnames(teststat) <- c("tau3", "phi2",
"phi3")
[[alternative HTML version deleted]]
