Interpreting Impulse Responses???

[Jon]

I have calculated a Vector Error Correction Model, but I am having trouble interpreting the results. When all variables are in first differences and they are in natural log form, I though the interpretation was that they were elasticities. But instead, I have impulse response values which are clarly not elasticities. When only one variable is differenced it appears that the results are elasticities. One paper appeared to multiply the differenced vars by 100 and I think left the one differenced variable vars unchanged in interpretation. Can anyone help???????