Help regarding missing values....

[Michael CP]

Hi,

I have the following code...When I run it it returns the following:

And it just keeps running...

//a)

clear all

set seed 1234567
set obs 100

gen t = _n -1
tsset t

gen epsilon = rnormal()
gen yt = 0
replace yt = epsilon + L.yt if _n > 1

generate model = yt - L.yt
regress model L.yt

drop yt

//b)

//Prepare Matrix:
matrix rhot = J(10000,2,.)

//Loop:
forvalues i = 1/10000 {

gen epsilon`i' = rnormal()
gen yt`i' = 0
replace yt`i' = epsilon`i' + L.yt`i' if _n > 1

generate model`i' = yt`i' - L.yt`i'

quietly regress model`i' L.yt`i'
matrix saveresults = r(table)
scalar row=`i'
matrix rhot[row,1]= saveresults[1,1]
matrix rhot[row,2]= saveresults[3,1]

drop yt`i'
drop epsilon`i'
drop model`i'

}



//Histograms
histogram rho_regr
histogram t_regr

//Percentiles
_pctile t_regr, p(5)
return list


//c)

//Prepare Matrix:
matrix rhot2 = J(10000,4,.)

//Loop:
forvalues i = 1/10000 {

gen epsilon`i' = rnormal()
gen yt`i' = 0
replace yt`i' = 0.5 + epsilon`i' + L.yt`i' if _n > 1

generate model`i' = yt`i' - L.yt`i'

quietly mvreg model`i' = L.yt`i' t
matrix saveresults = r(table)
scalar row=`i'
matrix rhot2[row,1]= saveresults[1,3] // alpha
matrix rhot2[row,2]= saveresults[1,2] // beta
matrix rhot2[row,3]= saveresults[1,1] // rho
matrix rhot2[row,4]= saveresults[3,1] // rho's t

drop yt`i'
drop epsilon`i'
drop model`i'
}

svmat rhot2, names( col )
rename c1 alpha
rename c2 beta
rename c3 rho
rename c4 rho_t

//Histograms
histogram rho
histogram rho_t

//Percentiles
_pctile rho_t, p(5)
return list


//d)

//Prepare Matrix:
matrix rhot3 = J(10000,1,.)

//Loop:
forvalues i = 1/10000 {

gen epsilon`i' = rnormal()
gen yt`i' = 0
replace yt`i' = 0.5 + epsilon`i' + L.yt`i' if _n > 1

generate model`i' = yt`i' - L.yt`i'

quietly regress model`i' L.yt`i', noconstant
matrix saveresults = r(table)
scalar row=`i'
matrix rhot3[row,1]= saveresults[3,1] // rho's t

drop yt`i'
drop epsilon`i'
drop model`i'
}

svmat rhot3, names( col )
rename c1 rho_t

//Percentiles
_pctile rho_t, p(5)
return list

[William Lisowski]

You can suppress the messages from the replace command by using the quietly prefix on it as you did on your regress commands. Then I would suggest changing 10000 to 10 and seeing if the smaller test runs successfully before spending time waiting for 10000 iterations on each of your loops to complete.

[Michael CP]

With 10 it returns this:

clear all

.
. set seed 1234567

. set obs 100
Number of observations (_N) was 0, now 100.

.
. gen t = _n -1

. tsset t

Time variable: t, 0 to 99
Delta: 1 unit

.
. gen epsilon = rnormal()

. gen yt = 0

. replace yt = epsilon + L.yt if _n > 1
(99 real changes made)

.
. generate model = yt - L.yt
(1 missing value generated)

. regress model L.yt

Source | SS df MS Number of obs = 99
-------------+---------------------------------- F(1, 97) = 1.49
Model | 1.49484416 1 1.49484416 Prob > F = 0.2259
Residual | 97.6292336 97 1.00648694 R-squared = 0.0151
-------------+---------------------------------- Adj R-squared = 0.0049
Total | 99.1240777 98 1.01147018 Root MSE = 1.0032

------------------------------------------------------------------------------
model | Coefficient Std. err. t P>|t| [95% conf. interval]
-------------+----------------------------------------------------------------
yt |
L1. | -.0219611 .0180202 -1.22 0.226 -.0577263 .0138041
|
_cons | -.3191688 .1785639 -1.79 0.077 -.6735687 .0352311
------------------------------------------------------------------------------

.
. drop yt

.
. //b)
.
. //Prepare Matrix:
. matrix rhot = J(10,2,.)

.
. //Loop:
. forvalues i = 1/10 {
2.
. gen epsilon`i' = rnormal()
3. gen yt`i' = 0
4. quietly replace yt`i' = epsilon`i' + L.yt`i' if _n > 1
5.
. generate model`i' = yt`i' - L.yt`i'
6.
. quietly regress model`i' L.yt`i'
7. matrix saveresults = r(table)
8. scalar row=`i'
9. matrix rhot[row,1]= saveresults[1,1]
10. matrix rhot[row,2]= saveresults[3,1]
11.
. drop yt`i'
12. drop epsilon`i'
13. drop model`i'
14.
. }
(1 missing value generated)
(1 missing value generated)
(1 missing value generated)
(1 missing value generated)
(1 missing value generated)
(1 missing value generated)
(1 missing value generated)
(1 missing value generated)
(1 missing value generated)
(1 missing value generated)

.
.
.
. //Histograms
. histogram rho_regr
variable rho_regr not found
r(111);

end of do-file

r(111);

.

[Michael CP]

Don't get why my code doesn't return the histograms as well...

[Nick Cox]

#3 asks for a histogram of a variable that doesn't exist. Nowhere in the code is rho_regr created.

[William Lisowski]

Note also that you still have messages coming out of your loop. Unless you want eventually to see

(1 missing value generated)

10000 times, you will want to add quietly in front of the generate command that follows the replace command.