In my fixed effects regression of about 30 panels in 30 years when i include year dummies by using -i.year- all my regressors lose significance. But when i check the VIF for year, it is 10. most of my regressors have very clear time trends, so there is very clear multicollinearity. should i include year dummies or not, because they mean that i do not get very clear results.