serious problem: Multicolinearity

[indiex]

Hi,

I have 53 obesrvations and 5 independent variable. VIF value I am getting is 98.34, which is too high, therefore a serious problem.
I cannot increase the number of observation as no previous data is available. So that is no more an option.

What else can be done to remove multicollinerity. I have also tried to remove collinear variable but that isn't not working.

Thanks

[startz]

Why do you think multicollinearity is a serious problem?

[indiex]

It can make confidence intervals for the parameters very wide, and significance tests might give inappropriate conclusions, and so it makes it difficult to draw sharp inference.

[startz]

True, multicollinearity makes confidence intervals very wide and makes it hard to draw sharp inference. But it does not make anything inappropriate about significance tests.

The problem is that you don't have enough information to separate out the effects of the individual variables.The only cure is more information. You've said that you can't get more data. The only other option is to provide information...not from anything observable in the data...about some of the coefficients. If you know ex ante the value of one coefficient, you may get much sharper inference about the others.