Hi everyone,

I am currently trying to run a fixed effects model on Stata using the ppml command, but the regression only shows the intercept for some reason. Can you help me figure out what's wrong in my code? In our model, we want to determine the effect of the Belt and Road Initiative on bilateral exports. We also introduce interaction terms between (1) the dummy variable for BRI and the continuous variable 'infrastructure level' and (2) the dummy variable for BRI and the continuous variable for 'outward Chinese FDI' (OFDI). We do these to determine how BRI influences the effect of infrastructure level and OFDI on exports. Here is the regression result:

note: starting ppml estimation
note: tradeflow_baci has noninteger values

Iteration 1: deviance = 1.12e+09
Iteration 2: deviance = 1.08e+09
Iteration 3: deviance = 1.08e+09
Iteration 4: deviance = 1.08e+09
Iteration 5: deviance = 1.08e+09
Iteration 6: deviance = 1.08e+09

Number of parameters: 1
Number of observations: 93
Pseudo log-likelihood: -5.388e+08
R-squared: .
Option strict is: off
(Std. err. adjusted for 23 clusters in pair_id)
------------------------------------------------------------------------------
| Robust
tradeflow_~i | Coefficient std. err. z P>|z| [95% conf. interval]
-------------+----------------------------------------------------------------
_cons | 15.62473 .3956078 39.50 0.000 14.84936 16.40011
------------------------------------------------------------------------------

This is the code I used:
ppml tradeflow_baci PAIR_FE* EXPORTER_TIME_FE* IMPORTER_TIME_FE* ln_world_nomGDP ln_importer_nomGDP ln_exporter_nomGDP ln_importer_infra ln_exporter_infra ln_ofdi ln_distcap rta bri comlang_ethno contig col_dep_ever bri_ln_exporter_infra bri_ln_importer_infra bri_ln_ofdi ln_ofdi_ln_exporter_infra ln_ofdi_ln_importer_infra if exporter != importer, cluster(pair_id)

To get the marginal effect of BRI on OFDI and infrastructure level, I was hoping to get the fixed effects estimates for paired fixed effects, importer fixed effects and exporter fixed effects, then regress each of those on the corresponding variables they absorbed, effectively obtaining within estimates from the two-step regression following the method by Head & Mayer (2014).

Thank you so much in advance for your help!