Hi there,
I have time series data on a number of variables and am hoping to do VAR and VECM analysis. To do this I believe all variables included in the VAR have to be I(1).

I have performed a series of stationarity tests but have somewhat conflicting results. The tests that I have run are ADF, DF-GLS, Phillips Perron, KPSS where lag lengths were chosen by AIC information criterion. I've attached the table of results (with significant contradictions highlighted).

Whilst there is some consensus, I'm not particularly convinced that they are I(1). The variable p, for example, is consistent across the tests. However, other variables are not - e.g. for CC, pt and rpt KPSS test rejects stationarity at first differences, suggesting that they are not I(1).

I'm aware that the power of these unit root tests is weak and I have a relatively small sample of 95 observations. Without the KPSS test the results are relatively consistent but I'd ideally like to include this test as the null hypothesis is one of stationarity, unlike the other tests I have run.
Thanks in advance,