Denton disaggregation method, VAR and Causality

[hermann]

Hi everyone,
I am planning on constructing a VAR model that has 3 quarterly time series, namely GDP (which is originally annual and which I am going to disaggregate via Denton Proportional Method using some proxy variables), and 2 other series M2/GDP (which is available quarterly) and BCR (bank credit ratio, which is available quarterly).
I would have like to know your thoughts on the following matters :
1. What is the overall impact of the disaggregation procedure on the VAR model and the standard causality tests (Granger and Johansen)
2. All three series clearly exhibiting an upward trend and a seasonal trend, tested to be I(1) through ADF tests and Correlogram, what is likely to be the impact on the coefficients and the interpretation of the results, and second how would I remove the seasonal effect from the variable? (should I remove seasonality thus creating new series before implementing a VAR model, or shoud I implement a VAR model and taking implicitely into account the seasonality through dummy variables (which I am not very familiar with)?)
Thank you all very much for any insight on these matters.
Hermann