MULTIPLE REGRESSION
Posted: Wed Jul 01, 2009 10:38 am
I have a series of 31 portfolio returns and I have tried to estimate the model using the OLS programme below. I created a group for the dependent variables as well as the independent variables.
'matrix to hold results
matrix(deps.@count,regs.@count)coefs
matrix(deps.@count,regs.@count)tstats
matrix c(i)
'declare equation object to be used in loop
equation e1
'now do estimations
for !i = 1 to deps.@count
e1.ls(n)deps(!i) c regs
rowplace(coefs,@transpose(e1.@coefs),!i)
rowplace(tstats,@transpose(e1.@tstats),!i)
next
However I want to add the p-values of the regression coefficients, correleogram(Q stat and Qstat sq at lag 7 and 21 only) Jarque Bera statistics with skewness and kurtosis, LM for serial on correlation, Durbin Watson, ARCH test in addition to the usual regression parameters such as Adjusted R-squared, F stat etc and transfer them into a matrix.
I have tried to do this but I'm not getting good results. I'm quite new to eviews programming and would appreciate suggestions on how to improve this model.
Many thanks in advance.
'matrix to hold results
matrix(deps.@count,regs.@count)coefs
matrix(deps.@count,regs.@count)tstats
matrix c(i)
'declare equation object to be used in loop
equation e1
'now do estimations
for !i = 1 to deps.@count
e1.ls(n)deps(!i) c regs
rowplace(coefs,@transpose(e1.@coefs),!i)
rowplace(tstats,@transpose(e1.@tstats),!i)
next
However I want to add the p-values of the regression coefficients, correleogram(Q stat and Qstat sq at lag 7 and 21 only) Jarque Bera statistics with skewness and kurtosis, LM for serial on correlation, Durbin Watson, ARCH test in addition to the usual regression parameters such as Adjusted R-squared, F stat etc and transfer them into a matrix.
I have tried to do this but I'm not getting good results. I'm quite new to eviews programming and would appreciate suggestions on how to improve this model.
Many thanks in advance.