missing data and estimation problems

[tarox]

dear all ,,,

please ..i need help to see my if steps are correct or not :(

i tried to find missing data for one index( which is gcc index) for perioed Oct 2011 - Jul 2014, by using another index (msci index),,
here are my steps and more info:

1- the gcc index have data from Dec 2002 - Sep 2011, but the msci index have data from May 2007 - Jul 2014.

2- there are missing data for msci index from Sep 2010 - Jul 2012, so i applied moving average for 5 observations ( yellow ).

3- then i used normal regression between gcc index (as dependent variable) and msci index (as independent variable) for available period (May 2007 - Sep 2011),,, here i tried several regressions (normal (y1,x1) , log (y2,x2) and half log(ly1,x1) . all of them show Autocorrelation and Serial Correlation so i tried to remove it by using AR(1) but then they show Heteroscedasticity problems (using white test),,, so i used first different ( dy1, dx1) in regression ( because variables are not stationery in level) and there were no problems on it.

4- i used the regression with first different data to find missing data on gcc , by replace each value of msci in x for each month (yellow),, is it correct ?

my questions here:

are my steps correct or not ?
can i use first different for regression?
is moving average by using 5 observations correct or not, or should i use 3 observations?

here are my files i use eviews 7 for regression and excel for data

thank you

[tarox]

please ,,, anyone could help me? :?