X-Axis show half plot

[Obaid Ur Rehman]

Hello Statalist, I use the below code which makes the below graph. I don't know why the year 2006 does not display the chart. any solution, please???

preserve
keep if inlist(year, 2002 , 2003, 2005, 2006)

gen event = year - 2004 + 100
tab event
xtreg st1 ib100.event i.year i.treat $cl, fe i(stkcd) cluster(stkcd)
eststo st1p
coefplot st1p, keep(*.event) vert rename(*.event="") xtitle("Years") ytitle("ST1") xlabel(1 "2002" 2 "2003" 3 "2005" 4 "2006") xscale(range(0.5 4.5)) name(st1_upper, replace)

Below is the data:

* Example generated by -dataex-. For more info, type help dataex
clear
input double(st1 event) float year double treat
                   .  98 2002 0
                   .  99 2003 0
                   . 101 2005 0
                   0 102 2006 0
  .09604616696929251  98 2002 0
  .28962622471456256  99 2003 0
  .06710824471689544 101 2005 0
  .08620013879570665 102 2006 0
   .6662144718507561  98 2002 0
   .6029833501337559  99 2003 0
   .4898758380528328 101 2005 0
   .2651543595305284 102 2006 0
   .5568158447725238  98 2002 0
   .5820257081493239  99 2003 0
  .39235363450130195 101 2005 0
    .441954794730697 102 2006 0
   .3098061216205315  98 2002 0
  .31966307412355016  99 2003 0
  .30206380882394623 101 2005 0
   .4081601935245437 102 2006 0
   .5270173664040717  98 2002 0
   .5838580271282523  99 2003 0
   .6602583189751736 101 2005 0
   .6090947192607512 102 2006 0
   .3613609940808387  98 2002 0
   .3345362221399254  99 2003 0
   .2502982744180982 101 2005 0
  .20914447364739341 102 2006 0
    .320553016823736  98 2002 0
   .3513748248164273  99 2003 0
   .4028876443683232 101 2005 0
  .04001263924960175 102 2006 0
  .34580344868702845  98 2002 1
  .36041677565924674  99 2003 1
   .4046321058365264 101 2005 1
   .3866869174226278 102 2006 1
   .5589146932798127  98 2002 0
   .5664086417948185  99 2003 0
  .30870132438471803 101 2005 0
   .2721147184531201 102 2006 0
   .3810061466231089  98 2002 1
  .27196101073408385  99 2003 1
  .21549063140258673  98 2002 0
  .31684958830123644  99 2003 0
  .15386760175179565 101 2005 0
  .29207055014036587 102 2006 0
  .03454964155295443  98 2002 0
.0037084165530211354  99 2003 0
   .0049528033575212 101 2005 0
 .002341256058221906 102 2006 0
  .18514068258937158  98 2002 1
  .20845330098283474  99 2003 1
   .2152675321755276 101 2005 1
     .20596810358499 102 2006 1
   .6330952325073145  98 2002 1
   .6975774214228182  99 2003 1
   .5103128519828921 101 2005 1
   .4280376449167185 102 2006 1
   .6098343936781896  98 2002 0
   .6359622517903386  99 2003 0
   .4213816921473605 101 2005 0
    .521120432391642 102 2006 0
   .6701561894010253  98 2002 1
   .6706641805010028  99 2003 1
   .6259713721975206 101 2005 1
  .44221327825139806 102 2006 1
 .000724298577907034  98 2002 0
.0010635444618339396  99 2003 0
                   0 101 2005 0
                   0 102 2006 0
   .5460841052795266  98 2002 0
   .3264975633382865  99 2003 0
   .4600696907500765 101 2005 0
   .6314494112229786 102 2006 0
   .5433364179462896  98 2002 0
   .6247981813914562  99 2003 0
    .325480416876663 101 2005 0
   .2268838619325507 102 2006 0
  .16258155907103944  98 2002 1
  .26201567935248044  99 2003 1
  .32809808873984503 101 2005 1
  .22732999369263898 102 2006 1
   .4898328134968033  98 2002 1
   .4207752846077022  99 2003 1
   .3839122176760559 101 2005 1
  .34959561429621605 102 2006 1
  .08065393240547168  98 2002 1
 .001995294934919864  99 2003 1
    .248751191068432 101 2005 1
   .6939867296277917 102 2006 1
  .06381824984791717  98 2002 1
  .03444830617789382  99 2003 1
  .04457600352451833 101 2005 1
  .10103184057715188 102 2006 1
    .239804659084622  98 2002 0
  .28604304438985106  99 2003 0
 .014733469046230997 101 2005 0
   .1346057898371778 102 2006 0
  .40398044237865693  98 2002 0
   .4143122286652363  99 2003 0
end
format %ty year

[Andrew Musau]

coefplot is from SSC, as you are asked to explain (FAQ Advice #12). Show us your xtreg results.

[Obaid Ur Rehman]

xtreg st1 ib100.event i.year i.treat $cl, fe i(stkcd) cluster(stkcd)
note: 100b.event identifies no observations in the sample.
note: 102.event omitted because of collinearity.
note: 2003.year omitted because of collinearity.
note: 2005.year omitted because of collinearity.
note: 2006.year omitted because of collinearity.
note: 1.treat omitted because of collinearity.

Fixed-effects (within) regression               Number of obs     =      5,490
Group variable: stkcd                           Number of groups  =      1,528

R-squared:                                      Obs per group:
     Within  = 0.0961                                         min =          1
     Between = 0.0026                                         avg =        3.6
     Overall = 0.0202                                         max =          4

                                                F(12,1527)        =      22.88
corr(u_i, Xb) = -0.0860                         Prob > F          =     0.0000

                              (Std. err. adjusted for 1,528 clusters in stkcd)
------------------------------------------------------------------------------
             |               Robust
         st1 | Coefficient  std. err.      t    P>|t|     [95% conf. interval]
-------------+----------------------------------------------------------------
       event |
         98  |   .0822488   .0097897     8.40   0.000      .063046    .1014515
         99  |   .0845076   .0087797     9.63   0.000      .067286    .1017291
        100  |          0  (empty)
        101  |    .019103   .0038677     4.94   0.000     .0115163    .0266896
        102  |          0  (omitted)
             |
        year |
       2003  |          0  (omitted)
       2005  |          0  (omitted)
       2006  |          0  (omitted)
             |
     1.treat |          0  (omitted)
        size |   .0253561   .0104489     2.43   0.015     .0048603    .0458518
         lev |   .0255446   .0332853     0.77   0.443     -.039745    .0908343
        prof |  -.3001358    .041417    -7.25   0.000    -.3813761   -.2188955
           q |   .0087745   .0065781     1.33   0.182    -.0041286    .0216776
        tang |   .1099002   .0218948     5.02   0.000     .0669531    .1528474
        dlev |   .0134716     .02131     0.63   0.527    -.0283285    .0552716
      accint |  -.2210474   .0225716    -9.79   0.000     -.265322   -.1767728
        inef |   .0143573   .0609268     0.24   0.814    -.1051518    .1338664
        fage |    .000579   .0017897     0.32   0.746    -.0029316    .0040896
       _cons |   -.273388    .218039    -1.25   0.210    -.7010756    .1542995
-------------+----------------------------------------------------------------
     sigma_u |  .18979683
     sigma_e |   .1234366
         rho |  .70275531   (fraction of variance due to u_i)
------------------------------------------------------------------------------

. eststo st1p

[Andrew Musau]

There is nothing to show here as the coefficient is omitted.

------------------------------------------------------------------------------
| Robust
st1 | Coefficient std. err. t P>|t| [95% conf. interval]
-------------+----------------------------------------------------------------
event |
98 | .0822488 .0097897 8.40 0.000 .063046 .1014515
99 | .0845076 .0087797 9.63 0.000 .067286 .1017291
100 | 0 (empty)
101 | .019103 .0038677 4.94 0.000 .0115163 .0266896
102 | 0 (omitted)

If you want to indicate that the coefficient is omitted and show a marker at 0, use the option -omitted-.

coefplot st1p, keep(*.event) omitted vert rename(*.event="") xtitle("Years") ytitle("ST1") xlabel(1 "2002" 2 "2003" 3 "2005" 4 "2006") xscale(range(0.5 4.5)) name(st1_upper, replace)

This however may be confusing in the case that the reader interprets it as the base level.

[Obaid Ur Rehman]

i don't know why year 2006 observation dropped. below are my data characteristics:

tab treat

      treat |      Freq.     Percent        Cum.
------------+-----------------------------------
          0 |      3,007       54.50       54.50
          1 |      2,510       45.50      100.00
------------+-----------------------------------
      Total |      5,517      100.00

. 
. gen event = year - 2004 + 100

. tab event

      event |      Freq.     Percent        Cum.
------------+-----------------------------------
         98 |      1,275       23.11       23.11
         99 |      1,332       24.14       47.25
        101 |      1,416       25.67       72.92
        102 |      1,494       27.08      100.00
------------+-----------------------------------
      Total |      5,517      100.00

[Obaid Ur Rehman]

.

[Andrew Musau]

As you have 4 event years

. tab event

event | Freq. Percent Cum.
------------+-----------------------------------
98 | 1,275 23.11 23.11
99 | 1,332 24.14 47.25
101 | 1,416 25.67 72.92
102 | 1,494 27.08 100.00
------------+-----------------------------------
Total | 5,517 100.00

including

xtreg st1 ib100.event i.year

is equivalent to specifying a non-existent base. So Stata has to break the collinearity and takes 2006 (or event= 102) as the base. For an explanation, google the dummy variable trap.

[Obaid Ur Rehman]

Thanks, Andrew, i get the idea from your suggestion. below is the solution:

keep if inlist(year, 2002 , 2003, 2005, 2006, 2007)

gen event = year - 2004 + 100
tab event

xtreg st1 ib100.event i.year i.treat $cl, fe i(stkcd) cluster(stkcd)
eststo st1p
coefplot st1p, keep(*.event) vert rename(*.event="") xtitle("Years") ytitle("ST1") xlabel(1 "2002" 2 "2003" 3 "2005" 4 "2006") xscale(range(0.5 4.5)) name(st1_upper, replace)