I hope you are doing well, I am using a gravity model in the context of bilateral remittances which is basically oneway flows in contrast to trade, export and import.
home is recipient country and the host is the remittances sending country.
however, when I use the second specifications with pairid, the results significantly changes.
ppmlhdfe remit lgdpp lgdpp_hos lcost2 lmig_st , absorb(home host year) vce(robust) ------1
ppmlhdfe remit lgdpc lgdpc_hos mig_st lcost2 , absorb(pairid year) vce(robust)-----------2
Also, the z-value of migrant_stock variable is very high in specification 1. why with paired id the results significantly changes.
Do I stick to specification 1, is there any other way to deals with this specific problem.
Thanks for your comments on it.
PHP Code:
ppmlhdfe remit lgdpp lgdpp_hos lcost2 lmig_st , absorb(home host year) vce(robust)
(dropped 14 observations that are either singletons or separated by a fixed effect)
Iteration 1: deviance = 1.790e+05 itol = 1.0e-04 subiters = 8 min(eta) = -3.80 [p ]
Iteration 2: deviance = 6.237e+04 eps = 1.87e+00 itol = 1.0e-04 subiters = 6 min(eta) = -5.67 [ ]
Iteration 3: deviance = 4.722e+04 eps = 3.21e-01 itol = 1.0e-04 subiters = 6 min(eta) = -6.73 [ ]
Iteration 4: deviance = 4.518e+04 eps = 4.52e-02 itol = 1.0e-04 subiters = 6 min(eta) = -6.97 [ ]
Iteration 5: deviance = 4.489e+04 eps = 6.39e-03 itol = 1.0e-04 subiters = 4 min(eta) = -6.98 [ ]
Iteration 6: deviance = 4.486e+04 eps = 7.14e-04 itol = 1.0e-04 subiters = 8 min(eta) = -7.44 [p ]
Iteration 7: deviance = 4.486e+04 eps = 6.40e-05 itol = 1.0e-04 subiters = 3 min(eta) = -8.13 [ ]
Iteration 8: deviance = 4.486e+04 eps = 6.91e-06 itol = 1.0e-06 subiters = 3 min(eta) = -8.53 [ ]
Iteration 9: deviance = 4.486e+04 eps = 2.79e-07 itol = 1.0e-06 subiters = 2 min(eta) = -8.63 [ ]
Iteration 10: deviance = 4.486e+04 eps = 7.04e-10 itol = 1.0e-08 subiters = 2 min(eta) = -8.63 [ s ]
Iteration 11: deviance = 4.486e+04 eps = 2.44e-14 itol = 1.0e-10 subiters = 13 min(eta) = -8.63 [pso]
Iteration 12: deviance = 4.486e+04 eps = 3.80e-14 itol = 1.0e-10 subiters = 13 min(eta) = -8.63 [pso]
------------------------------------------------------------------------------------------------------------
(legend: p: exact partial-out s: exact solver o: epsilon below tolerance)
Converged in 12 iterations and 74 HDFE sub-iterations (tol = 1.0e-08)
HDFE PPML regression No. of obs = 1,100
Absorbing 3 HDFE groups Residual df = 985
Wald chi2(4) = 1902.16
Deviance = 44859.05854 Prob > chi2 = 0.0000
Log pseudolikelihood = -26589.13444 Pseudo R2 = 0.9825
------------------------------------------------------------------------------
| Robust
remit | Coef. Std. Err. z P>|z| [95% Conf. Interval]
-------------+----------------------------------------------------------------
lgdpp | -.8177773 .1726019 -4.74 0.000 -1.156071 -.4794838
lgdpp_hos | 1.174428 .5027084 2.34 0.019 .189138 2.159719
lcost2 | -.0138842 .0428241 -0.32 0.746 -.0978179 .0700495
lmig_st | .9379309 .02305 40.69 0.000 .8927538 .983108
_cons | -11.40849 8.36286 -1.36 0.173 -27.79939 4.982417
------------------------------------------------------------------------------
Absorbed degrees of freedom:
-----------------------------------------------------+
Absorbed FE | Categories - Redundant = Num. Coefs |
-------------+---------------------------------------|
home | 76 0 76 |
host | 30 1 29 |
year | 7 1 6 ?|
-----------------------------------------------------+
? = number of redundant parameters may be higher
PHP Code:
. ppmlhdfe remit lgdpc lgdpc_hos lcost2 lmig_st , absorb(pairid year ) vce(robust)
(dropped 54 observations that are either singletons or separated by a fixed effect)
Iteration 1: deviance = 1.394e+05 itol = 1.0e-04 subiters = 5 min(eta) = -2.61 [p ]
Iteration 2: deviance = 3.346e+04 eps = 3.17e+00 itol = 1.0e-04 subiters = 4 min(eta) = -3.60 [ ]
Iteration 3: deviance = 1.893e+04 eps = 7.67e-01 itol = 1.0e-04 subiters = 3 min(eta) = -4.59 [ ]
Iteration 4: deviance = 1.644e+04 eps = 1.51e-01 itol = 1.0e-04 subiters = 3 min(eta) = -5.58 [ ]
Iteration 5: deviance = 1.609e+04 eps = 2.21e-02 itol = 1.0e-04 subiters = 3 min(eta) = -6.53 [ ]
Iteration 6: deviance = 1.605e+04 eps = 2.35e-03 itol = 1.0e-04 subiters = 4 min(eta) = -7.40 [p ]
Iteration 7: deviance = 1.605e+04 eps = 2.08e-04 itol = 1.0e-04 subiters = 2 min(eta) = -8.10 [ ]
Iteration 8: deviance = 1.605e+04 eps = 2.07e-05 itol = 1.0e-04 subiters = 2 min(eta) = -8.50 [ ]
Iteration 9: deviance = 1.605e+04 eps = 8.18e-07 itol = 1.0e-06 subiters = 2 min(eta) = -8.60 [ ]
Iteration 10: deviance = 1.605e+04 eps = 2.16e-09 itol = 1.0e-08 subiters = 2 min(eta) = -8.61 [ s ]
Iteration 11: deviance = 1.605e+04 eps = 1.44e-14 itol = 1.0e-10 subiters = 6 min(eta) = -8.61 [pso]
Iteration 12: deviance = 1.605e+04 eps = 4.75e-16 itol = 1.0e-10 subiters = 6 min(eta) = -8.61 [pso]
------------------------------------------------------------------------------------------------------------
(legend: p: exact partial-out s: exact solver o: epsilon below tolerance)
Converged in 12 iterations and 42 HDFE sub-iterations (tol = 1.0e-08)
HDFE PPML regression No. of obs = 1,077
Absorbing 2 HDFE groups Residual df = 821
Wald chi2(4) = 22.02
Deviance = 16047.05859 Prob > chi2 = 0.0002
Log pseudolikelihood = -12105.96398 Pseudo R2 = 0.9919
------------------------------------------------------------------------------
| Robust
remit | Coef. Std. Err. z P>|z| [95% Conf. Interval]
-------------+----------------------------------------------------------------
lgdpc | .0409308 .0669247 0.61 0.541 -.0902393 .1721008
lgdpc_hos | .3785502 .0865878 4.37 0.000 .2088411 .5482592
lcost2 | .0181827 .0349164 0.52 0.603 -.0502522 .0866176
lmig_st | -.0126245 .0722956 -0.17 0.861 -.1543212 .1290722
_cons | 1.990043 1.690595 1.18 0.239 -1.323462 5.303547
------------------------------------------------------------------------------
0 Response to PPMLHDFE Gravity model for remittances
Post a Comment