Hi again, I discovered that I had confused the significance level. It should be 5% significance in all cases, both bartlett and boot-strapped p-values (1896-1929 and 1982-2008).

Well, this already makes more sense. Apart from that, you forgot to give your exact specification; at the very least which variables you are including in the test. For example, I see that there is a column "t" in your datafile, which represents a time trend. Such a variable should of course not be treated as a standard endogenous one. If you did, then the answer is that the result is invalid.

cheers

sven