Seeking help with a Monte Carlo script

[Henry Hu]

Hello,

I have been working on a monte carlo script (see code below) and was encountering an issue where the simulation keeps returning red "X"'s after the first run and output is all missing values. Strangely, if I run the code before "program MC_FCS, rclass", then proceed to run the code inside the monte carlo loop (from "program MC_FCS, rclass" to "end", non-inclusive) the code runs! Adding "trace" as an option also throws up no errors and as far as I can tell, the code is running just fine. This has me a bit confused as to why the monte carlo loop is not running, any suggestions/tips would be very much appreciated!

I have used an older script as a template and as far as I can tell, there seems to be no issue with the syntax, but I could be wrong...

*-------------------------------------------------------------------------------
*----- This is a monte carlo script for analyzing validity of FCS over time ----
*-------------------------------------------------------------------------------

**********************************SETTINGS**************************************
clear all
set more off, permanent
set varabbrev off

**********************************GLOBALS***************************************
global graph 0
global cutoff_poor 20
global cutoff_border 40
global scalefactor 1.1




**********************************DATA READ IN**********************************
use "C:\Users\15102\Desktop\Monte_FCS\MonteCarloCode\bl_fd_migrant_2019shock_old.dta", clear
keep hhid week cs_*

**********************************SETUP PANEL***********************************
xtset hhid week

**********************************SETUP DATA************************************
****FCS CATEGORIES
*STAPLES (weight=2)
cap egen cs_staples=rowtotal(cs_ricegrains cs_wheatflour)
*Pulses (weight=3)
cap egen cs_pulses=rowtotal(cs_pulsespice cs_tubers)
*Veg (weight=1)
cap egen cs_veg=rowtotal(cs_veg)
*Fruit (weight=1)
cap egen cs_fruit=rowtotal(cs_fruit)
*Meat and other protiens (weight=4)
cap egen cs_meat=rowtotal(cs_fishshrimp cs_poultry cs_eggs cs_redmeat)
*Milk (weight=4)
cap egen cs_dairy=rowtotal(cs_diary)
*Sugar and oil (weight=0.5*2=1)
cap egen cs_packaged=rowtotal(cs_packaged)

****setup assumed values to convert to consumption count:
foreach v in staples pulses veg fruit meat dairy packaged {
qui sum cs_`v' if cs_`v'>0
scalar temp_`v'=(r(mean)/5)*${scalefactor}
global v_`v' temp_`v'
di "price of ${v_`v'}:"${v_`v'}
}

****convert to number of days:
foreach v in staples pulses veg fruit meat dairy packaged {
gen days_`v'=cs_`v'/${v_`v'}
sum days_`v'
replace days_`v'=7 if days_`v'>=7
}

****Graphical check
if ${graph}==1 {
kdensity days_staples
kdensity days_pulses
kdensity days_veg
kdensity days_fruit
kdensity days_meat
kdensity days_dairy
kdensity days_packaged
}

**********************************MONTE CARLO***********************************
program MC_FCS, rclass



*Step1. Calculate random time to "do first survey"(wave1)
gen RNG1=uniform()
egen RNG1_max=max(RNG1) if week<=12
gen start=0
replace start=1 if RNG1==RNG1_max
sum week if start==1
gen wave1=r(mean)
sum wave1


*Step2. calculate (roughly) alternative waves 2-4
gen wave2=wave1+12
gen wave3=wave2+12
gen wave4=wave3+12
sum wave*

*Step3 calculate FCS for each week
gen FCS=2*days_staples+3*days_pulses+days_veg+days_fruit+4*days_meat+4*days_dairy+days_packaged
if ${graph}==1 {
kdensity FCS
}

*Step4 calculate "4 wave FCS"
forvalues v=1(1)4 {
gen FCS_w`v'=.
bys hhid: replace FCS_w`v'=FCS if week==wave`v'
}
forvalues v=1(1)4 {
bys hhid: egen temp`v'=max(FCS_w`v')
drop FCS_w`v'
rename temp`v' FCS_w`v'
}

*step5 calculate average FCS
bys hhid: egen avg_FCS=mean(FCS)
bys hhid: gen est_FCS=(FCS_w1+FCS_w2+FCS_w3+FCS_w4)/4

*step5 drop duplicates and calculate checking states
*preserve
*duplicates drop hhid, force
*var1: level difference
gen delta_level=est_FCS-avg_FCS
*var2: % difference
gen delta_pct=delta_level/avg_FCS
*classification
gen avg_FCS_poor=0
gen avg_FCS_borderline=0
gen avg_FCS_accept=0
replace avg_FCS_poor=1 if avg_FCS<=${cutoff_poor}
replace avg_FCS_borderline=1 if avg_FCS>${cutoff_poor} & avg_FCS<=${cutoff_border}
replace avg_FCS_accept=1 if avg_FCS>${cutoff_border}
gen est_FCS_poor=0
gen est_FCS_borderline=0
gen est_FCS_accept=0
replace est_FCS_poor=1 if est_FCS<=${cutoff_poor}
replace est_FCS_borderline=1 if est_FCS>${cutoff_poor} & est_FCS<=${cutoff_border}
replace est_FCS_accept=1 if est_FCS>${cutoff_border}
*var3: misclass
gen misclass=0
replace misclass=1 if avg_FCS_poor!=est_FCS_poor
replace misclass=1 if avg_FCS_borderline!=est_FCS_borderline
replace misclass=1 if avg_FCS_accept!=est_FCS_accept


drop RNG1 RNG1_max start wave1 wave2 wave3 wave4 FCS FCS_w1 FCS_w2 FCS_w3 FCS_w4 avg_FCS est_FCS delta_level delta_pct avg_FCS_poor avg_FCS_borderline avg_FCS_accept est_FCS_poor est_FCS_borderline est_FCS_accept misclass
*restore
end


*****The key concern here is that the list of things to be simulated changes depending on choices at the UI level (need fix)!
*preserve
simulate ///
r(avg_FCS) r(est_FCS) r(delta_level) r(delta_pct) r(misclass), trace dots reps(10): MC_FCS


gen avg_FCS=_sim_1
gen est_FCS=_sim_2
gen delta_level=_sim_3
gen delta_pct=_sim_4
gen misclass=_sim_5

sum avg_FCS est_FCS delta_level delta_pct misclass

[Henry Hu]

Kindly Ignore...I was missing "return scalar", keeping this post here untilI I can figure out how to delete this post.

[Clyde Schechter]

You can't delete a post. You can edit it for a brief time after original posting, but deletions are not allowed.

This is a forum for community learning. Others can learn from your problem and solution. That's why you should leave what you did up. In fact, my response to #2 is: thank you for closing the thread by showing the solution. Others coming here with similar problems in the future will be grateful.