Loop with a synthetic control method

[Luciana Pando]

Hi everyone,

I have a dataset at the state level with 39 states and I want to do a synthetic control regression (as shown) only for the states 4,5,21,34 by using the command levelsof ..., local() in a loop . Note that at the end of the regression I have to save the results in dataset "ps" that contains the number of the state. In this example if I run the regression for the state 4 I will save it in the dataset ps4 but I have to save also ps5, ps21 and ps34 in a loop.

Many thanks for your help!

tsset state year
synth cigsale beer  lnincome(1980&1985) age15to24  retprice  cigsale(1988) cigsale(1980) cigsale(1975), trunit(3) trperiod(1989) fig keep(ps4, replace)

state year cigsale lnincome beer retprice age15to24
29 1970 123.9        . . 39.3  .1831579
32 1970  99.8        . . 39.9  .1780438
10 1970 134.6        . . 30.6 .17651588
21 1970 189.5        . . 38.9  .1615542
14 1970 115.9        . . 34.3  .1851852
27 1970 108.4        . . 38.4 .17545916
22 1970 265.7        . . 31.4  .1707317
25 1970  93.8        . . 37.3   .184466
 2 1970 100.3        . . 36.7 .16900676
36 1970 124.3        . . 28.8 .18942162
 9 1970 124.8        . . 41.4  .1669667
31 1970  92.7        . . 38.5 .17867868
34 1970  65.5        . . 34.6 .20207743
 7 1970 109.9        . . 34.3  .1874455
17 1970  93.4        . . 36.2 .18313035
 4 1970 124.8        . . 29.4 .19095023
16 1970 104.3        . . 39.1  .1747241
 3 1970   123        . . 38.8 .17815833
33 1970 106.4        . . 40.4 .18314135
13 1970 155.8        . . 28.3 .18131015
15 1970 128.5        . .   38  .1690141
24 1970 172.4        . . 27.3  .1935484
19 1970 111.2        . .   34  .1757925
35 1970 122.6        . . 37.7  .1797753
11 1970 108.5        . . 37.7 .16884956
 5 1970   120        . . 42.2 .16292876
12 1970   114        . . 34.2 .18052468
 6 1970   155        . .   39 .17335767
38 1970 106.4        . . 38.5   .174287
 8 1970 102.4        . . 33.8  .1781206
23 1970    90        . . 39.7 .18485743
37 1970 114.5        . . 33.7 .17259175
28 1970 107.3        . . 38.4   .163376
30 1970 103.6        . . 32.5 .20030876
26 1970 121.6        . . 36.6  .1732195
20 1970 108.1        . . 33.9 .17373738
18 1970 121.3        . .   36   .167593
 1 1970  89.8        . . 39.6  .1788618
39 1970 132.2        . . 34.1  .1746988
 9 1971 125.6        . . 41.4  .1689976
22 1971   278        . . 34.1  .1723339
31 1971  96.7        . . 38.5 .18049243
13 1971 163.5        . . 30.1  .1822996
39 1971 131.7        . . 34.4 .17722893
19 1971 115.6        . . 34.7  .1771459
17 1971 105.4        . . 37.5 .18437305
12 1971 102.8        . . 38.9 .18155004
 6 1971 161.1        . . 41.3  .1758872
34 1971  67.7        . . 36.6 .20206134
27 1971 115.4        . . 39.8  .1765248
15 1971 133.2        . . 38.8  .1703349
 7 1971 115.7        . . 35.8 .18786626
36 1971 128.4        . . 30.2  .1898735
18 1971 127.6        . . 36.8 .16925956
 1 1971  95.4        . . 42.7 .17992784
25 1971  98.5        . . 38.9 .18638696
24 1971 187.6        . . 29.4  .1936767
 8 1971 108.5        . . 33.6 .17609245
23 1971  92.6        . . 41.7  .1860954
28 1971 106.3        . . 44.7  .1650846
 2 1971 104.1        . . 38.8 .16995385
32 1971 106.3        . . 41.6 .17881927
20 1971 108.6        . . 34.7 .17521714
33 1971 108.9        . .   42 .18430856
26 1971 124.6        . . 38.1  .1745399
38 1971 105.4        . . 40.2 .17634407
11 1971 108.4        . . 38.5   .170839
37 1971 111.5        . . 41.6 .17312744
 3 1971   121        . . 39.7 .17929636
 4 1971 125.5        . . 31.1  .1916476
30 1971   115        . . 34.3  .2004893
21 1971 190.5        . .   44 .16377378
10 1971 139.3        . . 32.2 .17797175
16 1971 116.4        . . 40.1  .1767316
14 1971 119.8        . . 39.3  .1867808
29 1971 123.2        . . 40.2  .1838495
35 1971 124.4        . . 39.5  .1813672
 5 1971 117.6        . . 45.5  .1646539
 7 1972   117  9.63889 . 40.9   .188287
34 1972  71.3 9.601122 . 37.2 .20204525
13 1972 179.4 9.547482 . 30.6 .18328904
31 1972   103 9.630849 . 39.1  .1823062
15 1972 136.5  9.59714 . 41.5  .1716557
26 1972 124.4 9.779579 . 38.4 .17586027
32 1972 111.5 9.569716 . 41.6  .1795947
20 1972 104.9 9.746475 . 41.1  .1766969
29 1972 134.4 9.770211 . 41.6  .1845411
33 1972 108.6 9.675209 . 46.9 .18547577
22 1972 296.2 9.736376 . 36.1 .17393607
27 1972 121.7 9.625027 . 39.8 .17759047
28 1972   109 9.784274 . 44.7 .16679317
 8 1972 126.1 9.651839 . 33.7  .1740643
11 1972 109.4 9.730265 . 41.9 .17282845
19 1972 122.2 9.694641 . 40.1  .1784993
 4 1972 134.3 9.805548 . 31.2 .19234496
35 1972   138 9.673069 .   40 .18295917
 3 1972 123.5 9.930814 . 39.9  .1804344
30 1972 118.7 9.509309 . 34.1  .2006698
21 1972 198.6 9.944233 . 40.6 .16599335
12 1972   111 9.768118 . 38.8  .1825754
end
label values state state
label def state 1 "Alabama", modify
label def state 2 "Arkansas", modify
label def state 3 "California", modify
label def state 4 "Colorado", modify
label def state 5 "Connecticut", modify
label def state 6 "Delaware", modify
label def state 7 "Georgia", modify
label def state 8 "Idaho", modify
label def state 9 "Illinois", modify
label def state 10 "Indiana", modify
label def state 11 "Iowa", modify
label def state 12 "Kansas", modify
label def state 13 "Kentucky", modify
label def state 14 "Louisiana", modify
label def state 15 "Maine", modify
label def state 16 "Minnesota", modify
label def state 17 "Mississippi", modify
label def state 18 "Missouri", modify
label def state 19 "Montana", modify
label def state 20 "Nebraska", modify
label def state 21 "Nevada", modify
label def state 22 "New Hampshire", modify
label def state 23 "New Mexico", modify
label def state 24 "North Carolina", modify
label def state 25 "North Dakota", modify
label def state 26 "Ohio", modify
label def state 27 "Oklahoma", modify
label def state 28 "Pennsylvania", modify
label def state 29 "Rhode Island", modify
label def state 30 "South Carolina", modify
label def state 31 "South Dakota", modify
label def state 32 "Tennessee", modify
label def state 33 "Texas", modify
label def state 34 "Utah", modify
label def state 35 "Vermont", modify
label def state 36 "Virginia", modify
label def state 37 "West Virginia", modify
label def state 38 "Wisconsin", modify
label def state 39 "Wyoming", modify

[Jared Greathouse]

You don't say how you intend to use levelsof, but no loops needed. Note I use allsynth from ssc.

u "http://fmwww.bc.edu/repec/bocode/s/synth_smoking.dta", replace

qui xtset state year, y
cls

loc int_time = 1989

tempfile cali

loc others (1980(1)1988)

loc beer (1984(1)1988)
cls
allsynth cigsale beer`beer' ///
    lnincome`others' retprice`others' ///
    age15to24`others' ///
    cigsale(1988) ///
    cigsale(1980) ///
    cigsale(1975), ///
    fig trunit(4) ////
    trperiod(`int_time') ///
    gapfig(classic placebos lineback) pval keep(smokingresults) rep
    
    
u smokingresults, clear

keep if inlist(state,4,5,21,34)

This estimates all your palcebos as you want, and then retains the states of interest. I recognize these as the donors for California. Why do you want to only estimate an SC for these?

[Luciana Pando]

Hi, thanks for your answer. I want to run this command using levelsof but it does not work. It says "4,state invalid name r(198)". Do you have any idea to solve the problem?

Many thanks.

levelsof state if (state==4,state==5, state==21,state==34),local(levels)

foreach i of local levels {
 qui synth cigsale retprice cigsale(1988) cigsale(1980) cigsale(1975), ///
xperiod(1980(1)1988) trunit(`i') trperiod(1989) keep(ps`i', replace)
}

[Jared Greathouse]

I just said you don't need levelsof. allsynth takes care of this, no loops or levelsof needed. You could loop over a numlist for this issue, but as I said, this is wholly unnecessary and can be accomplished without any looping at all. You also didn't answer my second question- why exactly do you want to do this? These are the selected donors for California from the Abadie 2010 paper. Why do you want these placebos, specifically? I'm not even saying it's wrong, I'm just curious on why you want these specific placebo estimates.

EDIT:
If you really want to use a loop, which again, I will say is not needed, you can just do

foreach i of num 4 5 21 34 {

qui synth cigsale retprice cigsale(1988) cigsale(1980) cigsale(1975), ///
xperiod(1980(1)1988) trunit(`i') trperiod(1989) keep(ps`i', replace)
}

This does as you wish, but again, as someone who writes SCM methods in Stata, I'm telling you there's a better way to accomplish what you desire.