if i am not wrong then u are asking the CI of estimator of <bblatex>\hat{y}</bblatex>, right? if it is right then u need to check what kind of regression method for estimation u have used, e.g. OLSE, MME, GMME, MLE, or PMLE et al. normally if u use OLSE like what showed with R code by 肖楠 then u need to calculate the asymptotic distribution of <bblatex>\hat{y}</bblatex>, under the assumptions refered in Econometrics by Fumio Hayashi(u can check in chapter 1:http://press.princeton.edu/chapters/s6946.pdf) <bblatex>\hat{y}</bblatex> is asymptotically normal distributed <bblatex>\sqrt{n}(\hat{y}-y) ---\ N(0, \sigma^2(X'X)^{-1}) </bblatex>, now u have already have the ditribution then u can calculate the CI by yourself.
