Hi everyone,

I am evaluating the effect of Grab's entry in Vietnam on firm performance indicators such as gross profit using a difference-in-difference design. To visualize the parallel trends assumption, I need to make a graph showing how the values of gross profit change for province per year.


I ran the following command:

Code:
xtreg grossprofit2 placetimedummy i.year##i.provinc
Code:
xtreg grossprofit2 placetimedummy i.year##i.provinc
note: 2015.year#80.provinc omitted because of collinearity

Random-effects GLS regression                   Number of obs     =      2,554
Group variable: firmid                          Number of groups  =        867

R-sq:                                           Obs per group:
     within  = 0.0065                                         min =          1
     between = 0.0328                                         avg =        2.9
     overall = 0.0256                                         max =          3

                                                Wald chi2(14)     =      40.55
corr(u_i, X)   = 0 (assumed)                    Prob > chi2       =     0.0002

---------------------------------------------------------------------------------
   grossprofit2 |      Coef.   Std. Err.      z    P>|z|     [95% Conf. Interval]
----------------+----------------------------------------------------------------
 placetimedummy |   1408.406   196247.1     0.01   0.994    -383228.7    386045.6
                |
           year |
          2013  |   115433.1   117971.8     0.98   0.328    -115787.4    346653.5
          2015  |     116295   155326.6     0.75   0.454    -188139.5    420729.4
                |
        provinc |
     Hai Phong  |   55391.53   205587.9     0.27   0.788    -347553.4    458336.5
     Khanh Hoa  |  -256938.5   268226.2    -0.96   0.338    -782652.1    268775.1
          HCMC  |   450535.3   165177.6     2.73   0.006     126793.1    774277.4
       Long An  |  -346668.4   227219.2    -1.53   0.127    -792009.9    98673.09
                |
   year#provinc |
2013#Hai Phong  |  -334925.7   177399.3    -1.89   0.059    -682621.9    12770.51
2013#Khanh Hoa  |     175364   230810.5     0.76   0.447    -277016.3    627744.3
     2013#HCMC  |  -149433.5   142382.3    -1.05   0.294    -428497.6    129630.6
  2013#Long An  |  -118933.7   197166.2    -0.60   0.546    -505372.3    267504.9
2015#Hai Phong  |  -345398.7   204305.2    -1.69   0.091    -745829.6    55032.08
2015#Khanh Hoa  |   11192.34   251957.3     0.04   0.965    -482634.9    505019.6
     2015#HCMC  |  -26003.57   143888.4    -0.18   0.857    -308019.7    256012.5
  2015#Long An  |          0  (omitted)
                |
          _cons |   709729.5   137203.6     5.17   0.000     440815.4    978643.5
----------------+----------------------------------------------------------------
        sigma_u |  1435055.7
        sigma_e |  1090994.5
            rho |  .63372441   (fraction of variance due to u_i)
---------------------------------------------------------------------------------

After that, I ran the margins command:

Code:
margins year#provinc
Code:
. margins year#provinc

Predictive margins                              Number of obs     =      2,554
Model VCE    : Conventional

Expression   : Linear prediction, predict()

---------------------------------------------------------------------------------
                |            Delta-method
                |     Margin   Std. Err.      z    P>|z|     [95% Conf. Interval]
----------------+----------------------------------------------------------------
   year#provinc |
   2011#Ha Noi  |          .  (not estimable)
2011#Hai Phong  |          .  (not estimable)
2011#Khanh Hoa  |          .  (not estimable)
     2011#HCMC  |          .  (not estimable)
  2011#Long An  |          .  (not estimable)
   2013#Ha Noi  |          .  (not estimable)
2013#Hai Phong  |          .  (not estimable)
2013#Khanh Hoa  |          .  (not estimable)
     2013#HCMC  |          .  (not estimable)
  2013#Long An  |          .  (not estimable)
   2015#Ha Noi  |          .  (not estimable)
2015#Hai Phong  |          .  (not estimable)
2015#Khanh Hoa  |          .  (not estimable)
     2015#HCMC  |          .  (not estimable)
  2015#Long An  |          .  (not estimable)
---------------------------------------------------------------------------------
I looked around on other threads already, but I haven't been able to find a solution yet. If I'm understanding correctly, the issue is caused by multicollinearity? I'm then surprised that the correlations between the variables are not all that high when checking for this.

Is there any way for me to fix this? Alternatively, is there another command I can use that can also help me visualize parallel trends?

Thank you in advance!

Note: This is my first time posting to the forum. If I missed anything that should be included, please let me know and I will upload any information needed as soon as possible.