*! lmfreg V1.0 14/08/2012 *! Emad Abd Elmessih Shehata *! Professor (PhD Economics) *! Agricultural Research Center - Agricultural Economics Research Institute - Egypt *! Email: emadstat@hotmail.com *! WebPage: http://emadstat.110mb.com/stata.htm *! WebPage at IDEAS: http://ideas.repec.org/f/psh494.html *! WebPage at EconPapers: http://econpapers.repec.org/RAS/psh494.htm *! Sahra Khaleel A. Mickaiel *! Professor (PhD Economics) *! Cairo University - Faculty of Agriculture - Department of Economics - Egypt *! Email: sahra_atta@hotmail.com *! WebPage: http://sahraecon.110mb.com/stata.htm *! WebPage at IDEAS: http://ideas.repec.org/f/pmi520.html *! WebPage at EconPapers: http://econpapers.repec.org/RAS/pmi520.htm program define lmfreg , eclass version 11.0 syntax [varlist] [if] [in] , [NOCONStant coll] tempvar E E2 E3 E4 EE ELin ELog ELYh Eo Es Es1 Ev tempvar ht ILVal L LDE LE LEo lf LnE2 LOGvars logYh LOGyvar LOGYX_ tempvar LY LYh LYh2 R2oS R2xx Rx Si SSE Time U U2 Ue VIFI wald weit Wi tempvar Wio WS X0 XQ XQX_ Yb Yh Yh2 Yhb Yho Yt YY YYm YYv Yh_ML Ue_ML TimeN tempname b B b1 b2 Beta Bv Bv1 Bx Cov D DX E E1 EE1 Eg Ew F tempname FGFF fgT FLin FLog h HT IPhi J K L M tempname M1 M2 OM P Phi Pm Q q1 S Sig2w Sig2 Sig2n Sig2o Sig2o1 Sn Sw tempname VaL1 VaL21 VaLv1 Vec vh VM VP W Wald We Wi tempname Wi1 Wio WY X X0 xq XQ Y Yh YhLin Nn N kx DF kb SSEo llf tempname YhLog Yi Z Z0 Z1 Zo Zr mh SST1 SST2 V v1 Uew Yhw tempvar `varlist' gettoken yvar xvar : varlist qui marksample touse qui markout `touse' , strok qui cap count if `touse' local N = r(N) scalar `Nn'=`N' qui gen `X0'=1 if `touse' matrix `X0'= J(`N',1,1) qui gen `TimeN'=_n qui gen `Time'=_n if `touse' qui tsset `Time' local both : list yvar & xvar if "`both'" != "" { di di as err " {bf:{cmd:`both'} included in both LHS and RHS Variables}" di as res " LHS: `yvar'" di as res " RHS: `xvar'" exit } if "`coll'"=="" { _rmcoll `xvar' if `touse' , `noconstant' `coll' forcedrop local xvar "`r(varlist)'" } local kx : word count `xvar' tsunab xvar : `xvar' tokenize `xvar' local xvar `*' mkmat `yvar' if `touse' , matrix(`Y') if "`noconstant'"!="" { qui mkmat `xvar' if `touse' , matrix(`X') scalar `DF'=`N'-`kx' scalar `kb'=`kx' } else { qui mkmat `xvar' `X0' if `touse' , matrix(`X') scalar `DF'=`N'-`kx'-1 scalar `kb'=`kx'+1 } qui gen `Wi'=1 if `touse' qui gen `weit' = 1 if `touse' mkmat `Wi' if `touse' , matrix(`Wi') matrix `Wi'=diag(`Wi') di di _dup(78) "{bf:{err:=}}" di as txt "{bf:{err:* Ordinary Least Squares (OLS)}}" di _dup(78) "{bf:{err:=}}" matrix `B'=invsym(`X''*`X')*`X''*`Y' matrix `Ue_ML'=(`Y'-`X'*`B') matrix `Sig2'=`Ue_ML''*`Ue_ML'/`DF' matrix `Cov'=`Sig2'*inv(`X''*`X') qui svmat `Ue_ML' , name(`Ue_ML') qui rename `Ue_ML'1 `Ue_ML' matrix `Yh_ML'=`X'*`B' qui svmat `Yh_ML' , name(`Yh_ML') qui rename `Yh_ML'1 `Yh_ML' tempname SSEo Sigo r2bu r2bu_a r2raw r2raw_a R20 f fp wald waldp tempname r2v r2v_a fv fvp r2h r2h_a fh fhp SSTm SSE1 SST11 SST21 Rho matrix `SSE'=`Ue_ML''*`Ue_ML' scalar `SSEo'=`SSE'[1,1] scalar `Sig2o'=`SSEo'/`DF' scalar `Sigo'=sqrt(`Sig2o') matrix `Sig2'=`SSEo'/`DF' scalar `Sig2n'=`SSEo'/`N' qui summ `Yh_ML' if `touse' local NUM=r(Var) qui summ `yvar' if `touse' local DEN=r(Var) scalar `r2v'=`NUM'/`DEN' scalar `r2v_a'=1-((1-`r2v')*(`N'-1)/`DF') scalar `fv'=`r2v'/(1-`r2v')*(`N'-`kb')/(`kx') scalar `fvp'=Ftail(`kx', `DF', `fv') qui correlate `Yh_ML' `yvar' if `touse' scalar `r2h'=r(rho)*r(rho) scalar `r2h_a'=1-((1-`r2h')*(`N'-1)/`DF') scalar `fh'=`r2h'/(1-`r2h')*(`N'-`kb')/(`kx') scalar `fhp'=Ftail(`kx', `DF', `fh') matrix `SSE'=`Ue_ML''*`Ue_ML' local Sig=`Sigo' qui summ `yvar' if `touse' local Yb=r(mean) qui gen `YYm'=(`yvar'-`Yb')^2 if `touse' qui summ `YYm' if `touse' qui scalar `SSTm' = r(sum) qui gen `YYv'=(`yvar')^2 if `touse' qui summ `YYv' if `touse' local SSTv = r(sum) qui summ `weit' if `touse' qui gen `Wi1'=sqrt(`weit'/r(mean)) if `touse' mkmat `Wi1' if `touse' , matrix(`Wi1') matrix `P' =diag(`Wi1') qui gen `Wio'=(`Wi1') if `touse' mkmat `Wio' if `touse' , matrix(`Wio') matrix `Wio'=diag(`Wio') matrix `Pm' =`Wio' matrix `IPhi'=`P''*`P' matrix `Phi'=inv(`P''*`P') matrix `J'= J(`N',1,1) matrix `D'=(`J'*`J''*`IPhi')/`N' matrix `SSE'=`Ue_ML''*`IPhi'*`Ue_ML' matrix `SST1'=(`Y'-`D'*`Y')'*`IPhi'*(`Y'-`D'*`Y') matrix `SST2'=(`Y''*`Y') scalar `SSE1'=`SSE'[1,1] scalar `SST11'=`SST1'[1,1] scalar `SST21'=`SST2'[1,1] scalar `r2bu'=1-`SSE1'/`SST11' scalar `r2bu_a'=1-((1-`r2bu')*(`N'-1)/`DF') scalar `r2raw'=1-`SSE1'/`SST21' scalar `r2raw_a'=1-((1-`r2raw')*(`N'-1)/`DF') scalar `f'=`r2bu'/(1-`r2bu')*(`N'-`kb')/`kx' scalar `fp'= Ftail(`kx', `DF', `f') scalar `wald'=`f'*`kx' scalar `waldp'=chi2tail(`kx', abs(`wald')) scalar `llf'=-(`N'/2)*log(2*_pi*`SSEo'/`N')-(`N'/2) local Nof =`N' local Dof =`DF' matrix `B'=`B'' if "`noconstant'"!="" { matrix colnames `Cov' = `xvar' matrix rownames `Cov' = `xvar' matrix colnames `B' = `xvar' } else { matrix colnames `Cov' = `xvar' _cons matrix rownames `Cov' = `xvar' _cons matrix colnames `B' = `xvar' _cons } yxregeq `yvar' `xvar' di as txt _col(3) "Sample Size" _col(21) "=" %12.0f as res `N' ereturn post `B' `Cov' , dep(`yvar') obs(`Nof') dof(`Dof') qui test `xvar' scalar `f'=r(F) scalar `fp'= Ftail(`kx', `DF', `f') scalar `wald'=`f'*`kx' scalar `waldp'=chi2tail(`kx', abs(`wald')) di as txt _col(3) "Wald Test" _col(21) "=" %12.4f as res `wald' _col(37) "|" _col(41) as txt "P-Value > Chi2(" as res `kx' ")" _col(65) "=" %12.4f as res `waldp' di as txt _col(3) "F-Test" _col(21) "=" %12.4f as res `f' _col(37) "|" _col(41) as txt "P-Value > F(" as res `kx' " , " `DF' ")" _col(65) "=" %12.4f as res `fp' di as txt _col(2) "(Buse 1973) R2" _col(21) "=" %12.4f as res `r2bu' _col(37) "|" as txt _col(41) "Raw Moments R2" _col(65) "=" %12.4f as res `r2raw' ereturn scalar r2bu =`r2bu' ereturn scalar r2bu_a=`r2bu_a' ereturn scalar f =`f' ereturn scalar fp=`fp' ereturn scalar wald =`wald' ereturn scalar waldp=`waldp' di as txt _col(2) "(Buse 1973) R2 Adj" _col(21) "=" %12.4f as res `r2bu_a' _col(37) "|" as txt _col(41) "Raw Moments R2 Adj" _col(65) "=" %12.4f as res `r2raw_a' di as txt _col(3) "Root MSE (Sigma)" _col(21) "=" %12.4f as res `Sigo' as txt _col(37) "|" _col(41) "Log Likelihood Function" _col(65) "=" %12.4f as res `llf' di _dup(78) "-" di as txt "- {cmd:R2h}=" %7.4f as res `r2h' _col(17) as txt "{cmd:R2h Adj}=" as res %7.4f `r2h_a' as txt _col(34) "{cmd:F-Test} =" %8.2f as res `fh' _col(51) as txt "P-Value > F(" as res `kx' " , " `DF' ")" _col(72) %5.4f as res `fhp' di as txt "- {cmd:R2v}=" %7.4f as res `r2v' _col(17) as txt "{cmd:R2v Adj}=" as res %7.4f `r2v_a' as txt _col(34) "{cmd:F-Test} =" %8.2f as res `fv' _col(51) as txt "P-Value > F(" as res `kx' " , " `DF' ")" _col(72) %5.4f as res `fvp' ereturn scalar r2raw =`r2raw' ereturn scalar r2raw_a=`r2raw_a' ereturn scalar llf =`llf' ereturn scalar sig=`Sigo' ereturn scalar r2h=`r2h' ereturn scalar r2h_a=`r2h_a' ereturn scalar r2v=`r2v' ereturn scalar r2v_a=`r2v_a' ereturn scalar fh=`fh' ereturn scalar fv=`fv' ereturn scalar fhp=`fhp' ereturn scalar fvp=`fvp' ereturn scalar kb=`kb' ereturn scalar kx=`kx' ereturn scalar DF=`DF' ereturn scalar Nn=_N ereturn display matrix `b'=e(b) matrix `V'=e(V) matrix `Bx'=e(b) local N=`Nn' qui tsset `Time' tempvar E E2 Yh logYh LOGyvar LYh ELYh ELin ELog YhLin YhLog FLin FLog Time di _dup(78) "{bf:{err:=}}" di as txt "{bf:{err:*** OLS Linear vs Log-Linear Functional Form Tests}}" di _dup(78) "{bf:{err:=}}" qui gen `Time'=_n qui tsset `Time' qui regress `yvar' `xvar' if `touse' , `noconstant' tempname SSELin llflin r2lin r2log YbG SumLY llflog SSELog r2loga boxcox tempname boxcoxp bmlin bmlinp bmlog bmlogp dmlin dmlinp dmlog dmlogp r2lina scalar `SSELin'=e(rss) scalar `llflin'=-(`N'/2)*ln(2*_pi*(`SSELin'/`N'))-(`N'/2) scalar `r2lin'=e(r2) qui predict `E' if `touse' , res qui gen `E2'=`E'*`E' if `touse' qui predict `Yh' if `touse' qui correlate `Yh' `yvar' if `touse' if `r2lin'==. { scalar `r2lin'=r(rho)*r(rho) } qui gen `logYh'=ln(`Yh') if `touse' qui gen `LOGyvar' = ln(`yvar') if `touse' qui regress `LOGyvar' `xvar' if `touse' , `noconstant' scalar `r2log'=e(r2) qui predict `LYh' if `touse' qui correlate `LYh' `LOGyvar' if `touse' if `r2log'==. { scalar `r2log'=r(rho)*r(rho) } scalar `SSELog'=e(rss) qui summ `LOGyvar' if `touse' scalar `YbG'=exp(r(mean)) scalar `SumLY'=r(sum) scalar `llflog'=-(`N'/2)*ln(2*_pi*(`SSELog'/`N'))-(`N'/2)-`SumLY' di as res " {bf:(1) R-squared}" di as txt _col(7) "Linear R2" _col(36) "=" %10.4f as res `r2lin' di as txt _col(7) "Log-Log R2" _col(36) "=" %10.4f as res `r2log' di _dup(75) "-" di as res " {bf:(2) Log Likelihood Function (LLF)}" di as txt _col(7) "LLF - Linear" _col(36) "=" as res %10.4f `llflin' di as txt _col(7) "LLF - Log-Log" _col(36) "=" as res %10.4f `llflog' di _dup(75) "-" di as res " {bf:(3) Antilog R2}" qui regress `LOGyvar' `xvar' if `touse' , `noconstant' scalar `SSELog'=e(rss) qui gen `ELYh'=exp(`LYh') if `touse' qui regress `ELYh' `yvar' if `touse' , `noconstant' scalar `r2lina'=e(r2) qui regress `logYh' `LOGyvar' if `touse' , `noconstant' scalar `r2loga'=e(r2) di as txt _col(7) "Linear vs Log-Log: R2Lin" _col(36) "=" %10.4f as res `r2lina' di as txt _col(7) "Log-Log vs Linear : R2log" _col(36) "=" %10.4f as res `r2loga' di _dup(75) "-" scalar `boxcox'=e(N)/2*abs(ln((`SSELin'/`YbG'^2)/`SSELog')) scalar `boxcoxp'=chi2tail(1, abs(`boxcox')) di as res" {bf:(4) Box-Cox Test}" _col(36) "=" %10.4f as res `boxcox' as txt _col(50) "P-Value > Chi2(1)" _col(70) %5.4f as res `boxcoxp' di as txt _col(7) "Ho: Choose Log-Log Model - Ha: Choose Linear Model" di _dup(75) "-" di as res " {bf:(5) Bera-McAleer BM Test}" qui regress `ELYh' `xvar' if `touse' , `noconstant' qui predict `ELin' if `touse' , res qui regress `logYh' `xvar' if `touse' , `noconstant' qui predict `ELog' if `touse' , res qui regress `yvar' `xvar' `ELog' if `touse' , `noconstant' qui test `ELog'=0 di as txt _col(7) "Ho: Choose Linear Model" _col(36) "=" %10.4f as res r(F) as txt _col(50) "P-Value > F(1, " e(df_r) ")" _col(70) %5.4f as res r(p) scalar `bmlin'=r(F) scalar `bmlinp'=r(p) qui regress `LOGyvar' `xvar' `ELin' if `touse' , `noconstant' qui test `ELin'=0 di as txt _col(7) "Ho: Choose Log-Log Model" _col(36) "=" %10.4f as res r(F) as txt _col(50) "P-Value > F(1, " e(df_r) ")" _col(70) %5.4f as res r(p) scalar `bmlog'=r(F) scalar `bmlogp'=r(p) di _dup(75) "-" di as res " {bf:(6) Davidson-Mackinnon PE Test}" qui regress `yvar' `xvar' if `touse' , `noconstant' qui predict `YhLin' if `touse' qui regress `LOGyvar' `xvar' if `touse' , `noconstant' qui predict `YhLog' if `touse' qui gen `FLin'=`YhLin'-exp(`YhLog') if `touse' qui gen `FLog'=`YhLog'-ln(`YhLin') if `touse' qui regress `yvar' `xvar' `FLog' if `touse' , `noconstant' qui test `FLog'=0 di as txt _col(7) "Ho: Choose Linear Model" _col(36) "=" %10.4f as res r(F) as txt _col(50) "P-Value > F(1, " e(df_r) ")" _col(70) %5.4f as res r(p) scalar `dmlin'=r(F) scalar `dmlinp'=r(p) qui regress `LOGyvar' `xvar' `FLin' if `touse' , `noconstant' qui test `FLin'=0 di as txt _col(7) "Ho: Choose Log-Log Model" _col(36) "=" %10.4f as res r(F) as txt _col(50) "P-Value > F(1, " e(df_r) ")" _col(70) %5.4f as res r(p) scalar `dmlog'=r(F) scalar `dmlogp'=r(p) di _dup(78) "-" ereturn scalar r2lin=`r2lin' ereturn scalar llflin=`llflin' ereturn scalar r2log=`r2log' ereturn scalar llflog=`llflog' ereturn scalar r2lina=`r2lina' ereturn scalar r2loga=`r2loga' ereturn scalar boxcox=`boxcox' ereturn scalar boxcoxp=`boxcoxp' ereturn scalar bmlin=`bmlin' ereturn scalar bmlinp=`bmlinp' ereturn scalar bmlog=`bmlog' ereturn scalar bmlogp=`bmlogp' ereturn scalar dmlin=`dmlin' ereturn scalar dmlinp=`dmlinp' ereturn scalar dmlog=`dmlog' ereturn scalar dmlogp=`dmlogp' qui tsset `TimeN' end program define yxregeq version 10.0 syntax varlist tempvar `varlist' gettoken yvar xvar : varlist local LEN=length("`yvar'") local LEN=`LEN'+3 di "{p 2 `LEN' 5}" as res "{bf:`yvar'}" as txt " = " " local kx : word count `xvar' local i=1 while `i'<=`kx' { local X : word `i' of `xvar' if `i'<`kx' { di " " as res " {bf:`X'}" _c di as txt " + " _c } if `i'==`kx' { di " " as res "{bf:`X'}" } local i=`i'+1 } di "{p_end}" di as txt "{hline 78}" end