lag length selection and ADF @trend

[shevchenko62]

I have 4 sets of monthly time series data (stock price, M2, interest rates and CPI) and i use ADF to check for unit roots. When I use trend and intercept model, there is @trend value. How do I know whether the @trend is significant? Shall I look at coefficient or the statistic?

Furthermore, I run lag length criteria on VAR for all the variables at level to get the lag length for cointegration. Based on the LR criteria, i got a recommended lag length of 10. But when I do cointegration, shall I use 9 instead since cointegration is using difference of the variables? Or should I estimate the VAR using difference (d(stock price), d(M2), d(interest rates) and d(CPI)) and use the value recommended directly (in this case, I got 5)?