Hi, i'm trying to calculate Pesaran (2007) Cross-section dependent IPS unit root test. I'm using a panel of 3 series, 3 countries and 38 obs (n*T=456).
The code i'm using is the following:
library(plm)
Data <- log(read_xlsx("DATA.xlsx", range = "B1:M39"))
Panel <- pdata.frame(Panel, index = c("Id", "Time"), drop.index = TRUE)
CIPS_gdpp <- cipstest(Panel$BCDS, lags = 1, type = c("drift"),
model = c(("cmg")), truncated = FALSE)
And I have the following error:
Error in approx(nintv, cvals[nintl:ninth, tintl, i], n = max(nintv) - :
need at least two non-NA values to interpolate
Called from: approx(nintv, cvals[nintl:ninth, tintl, i], n = max(nintv) -
min(nintv))
¿What is your opinion? is this a problem with "n" the sample size? I run the same test in stata with XTCIP without problem