cointegration with autocorrelation problem

[upani]

Dear All,

I am looking for a cointegration relationship between Spot and Future Price of commodites. The problem i am facing follows:

1. After estimating by Engle-Grranger Method, i found that the residuals are stationary at their level I (o), which is required to fulfill the cointegration test. But the autocorrelation problem arises, as DW statistics is signficantly low 0.50-0.88 for various commodities. My question is shall i go ahead with the results or not.

2. When i use Johansens Method i found at least one cointegrtion relation. But i am confused with lag selection criteria. I use VAR to select the lagselection criteria. But there is autocorrelation problem with the lags it is providing for AIC. Whether i should take first difference of the price level to estimate the VAR, then how to use the same lag selection criteria, when i am using price series in levels to estimate the cointegration by johansen method.

Looking forward for your help

With sincere regards,
Upananda