Documentation for “Trade, Institutions, and the Timing of GATT/WTO Accession in Post-Colonial States”
This file contains variable descriptions, a replication log to calculate statistical results presented in Table 2, and a list of Stata .do files to replicate the interaction charts (Figures 2-4) in the paper.
Replication dataset
The replication dataset is “RIO_Replication_Data.dta”. The file contains the following variables:
Variables
state = country name
entry = year/month of GATT/WTO accession
indep = year/month of independence
fromname = colonial ruler prior to independence
yearindep = years since independence
ukcol = British colony (dummy)
belgcol = Belgian colony
frcol = French colony
portcol = Portuguese colony
ccode1 = COW country code
kennedy = Kennedy round years (dummy)
tokyo = Tokyo round years (dummy)
uruguay = Uruguay round years (dummy)
wto = World Trade Organization years (dummy)
imf5 = IMF program in last five years (dummy)
gwcov5 = GATT/WTO trade coverage (5 year lagged moving average)
gwregcov5 = GATT/WTO regional trade coverage (5 year lagged MA)
pta5 = PTA trade coverage (5 year lagged MA)
open5 = trade/GDP (%, 5 year lagged MA)
ksgpolity5 = POLITY score (Gleditsch), 5 year lagged MA
ksg5xcov5 = ksgpolity5*gwcov5
rgdppc5l = real GDP per capita, log (5 year lagged MA)
rgdp5l = real GDP, log (5 year lagged MA)
cov5xpta5 = gwcov5*pta5
xz5 = gwcov5*pta5
xw5 = gwcov5*ksgpolity5
zw5 = kgspolity5*pta5
xzw5 = gwcov5*pta5*ksgpolity5
Replication log, Table 2
The following four Stata command lines will replicate the four models included in the empirical analysis. These correspond to the results in Models 1-4 of Table 2 in the paper.
Model 1
stcox gwcov5 pta5 ksgpolity5 gwregcov5 open5 imf5 rgdp5l rgdppc5l ukcol frcol portcol belgcol kennedy tokyo uruguay wto, efron cluster(ccode1) nohr
failure _d: member == 1
analysis time _t: yearindep
id: ccode1
Iteration 0: log pseudolikelihood = -162.69038
Iteration 1: log pseudolikelihood = -142.16116
Iteration 2: log pseudolikelihood = -140.29868
Iteration 3: log pseudolikelihood = -140.22679
Iteration 4: log pseudolikelihood = -140.2261
Iteration 5: log pseudolikelihood = -140.2261
Refining estimates:
Iteration 0: log pseudolikelihood = -140.2261
Cox regression -- Efron method for ties
No. of subjects = 61 Number of obs = 891
No. of failures = 50
Time at risk = 891
Wald chi2(16) = 63.22
Log pseudolikelihood = -140.2261 Prob > chi2 = 0.0000
(Std. Err. adjusted for 61 clusters in ccode1)
------
| Robust
_t | Coef. Std. Err. z P>|z| [95% Conf. Interval]
------+------
gwcov5 | .0288546 .0133843 2.16 0.031 .0026219 .0550873
pta5 | -.0176211 .0083674 -2.11 0.035 -.0340209 -.0012213
ksgpolity5 | .0804669 .0253798 3.17 0.002 .0307234 .1302104
gwregcov5 | .0459495 .0316723 1.45 0.147 -.0161269 .108026
open5 | .0092801 .0033474 2.77 0.006 .0027193 .0158409
imf5 | .5818388 .3380826 1.72 0.085 -.0807909 1.244469
rgdp5l | .1642131 .160312 1.02 0.306 -.1499926 .4784188
rgdppc5l | -.317387 .2369602 -1.34 0.180 -.7818205 .1470465
ukcol | .7585417 .4939484 1.54 0.125 -.2095793 1.726663
frcol | 1.178583 .6591783 1.79 0.074 -.1133827 2.470549
portcol | .5630939 .8075245 0.70 0.486 -1.019625 2.145813
belgcol | 2.43822 1.048511 2.33 0.020 .3831758 4.493264
kennedy | -.0302001 .8697317 -0.03 0.972 -1.734843 1.674443
tokyo | -1.752029 .8041979 -2.18 0.029 -3.328228 -.1758296
uruguay | 1.299486 .4224427 3.08 0.002 .4715138 2.127459
wto | -.1684491 .8466518 -0.20 0.842 -1.827856 1.490958
------
Model 2
stcox gwcov5 pta5 ksgpolity5 cov5xpta5 gwregcov5 open5 imf5 rgdp5l rgdppc5l ukcol frcol portcol belgcol kennedy tokyo uruguay wto, efron cluster(ccode1) nohr
failure _d: member == 1
analysis time _t: yearindep
id: ccode1
Iteration 0: log pseudolikelihood = -162.69038
Iteration 1: log pseudolikelihood = -140.76042
Iteration 2: log pseudolikelihood = -138.75043
Iteration 3: log pseudolikelihood = -138.64794
Iteration 4: log pseudolikelihood = -138.64695
Iteration 5: log pseudolikelihood = -138.64695
Refining estimates:
Iteration 0: log pseudolikelihood = -138.64695
Cox regression -- Efron method for ties
No. of subjects = 61 Number of obs = 891
No. of failures = 50
Time at risk = 891
Wald chi2(17) = 60.00
Log pseudolikelihood = -138.64695 Prob > chi2 = 0.0000
(Std. Err. adjusted for 61 clusters in ccode1)
------
| Robust
_t | Coef. Std. Err. z P>|z| [95% Conf. Interval]
------+------
gwcov5 | .0608419 .0217244 2.80 0.005 .0182628 .103421
pta5 | .076242 .0400621 1.90 0.057 -.0022782 .1547622
ksgpolity5 | .0874467 .0269937 3.24 0.001 .03454 .1403535
cov5xpta5 | -.0010603 .0004719 -2.25 0.025 -.0019851 -.0001355
gwregcov5 | .0457585 .0324434 1.41 0.158 -.0178294 .1093464
open5 | .0085843 .0031805 2.70 0.007 .0023507 .0148179
imf5 | .6096328 .3475024 1.75 0.079 -.0714595 1.290725
rgdp5l | .153749 .1597883 0.96 0.336 -.1594303 .4669283
rgdppc5l | -.2985881 .259526 -1.15 0.250 -.8072498 .2100735
ukcol | .5467845 .5149018 1.06 0.288 -.4624046 1.555974
frcol | 1.034103 .6177637 1.67 0.094 -.1766913 2.244898
portcol | .4904823 .8098424 0.61 0.545 -1.09678 2.077744
belgcol | 2.224141 1.060643 2.10 0.036 .1453188 4.302963
kennedy | -.0914022 .8986997 -0.10 0.919 -1.852821 1.670017
tokyo | -1.697216 .8098502 -2.10 0.036 -3.284494 -.1099392
uruguay | 1.467396 .435286 3.37 0.001 .6142515 2.320541
wto | .1211219 .8993775 0.13 0.893 -1.641626 1.883869
------
Model 3
stcox gwcov5 pta5 ksgpolity5 ksg5xcov5 gwregcov5 open5 imf5 rgdp5l rgdppc5l ukcol frcol portcol belgcol kennedy tokyo uruguay wto, efron cluster(ccode1) nohr
failure _d: member == 1
analysis time _t: yearindep
id: ccode1
Iteration 0: log pseudolikelihood = -162.69038
Iteration 1: log pseudolikelihood = -140.90804
Iteration 2: log pseudolikelihood = -139.09646
Iteration 3: log pseudolikelihood = -139.02984
Iteration 4: log pseudolikelihood = -139.0293
Iteration 5: log pseudolikelihood = -139.0293
Refining estimates:
Iteration 0: log pseudolikelihood = -139.0293
Cox regression -- Efron method for ties
No. of subjects = 61 Number of obs = 891
No. of failures = 50
Time at risk = 891
Wald chi2(17) = 66.66
Log pseudolikelihood = -139.0293 Prob > chi2 = 0.0000
(Std. Err. adjusted for 61 clusters in ccode1)
------
| Robust
_t | Coef. Std. Err. z P>|z| [95% Conf. Interval]
------+------
gwcov5 | .0395869 .0138292 2.86 0.004 .0124821 .0666917
pta5 | -.0194233 .0081988 -2.37 0.018 -.0354926 -.0033539
ksgpolity5 | -.2802562 .15398 -1.82 0.069 -.5820514 .021539
ksg5xcov5 | .0040421 .0017657 2.29 0.022 .0005814 .0075029
gwregcov5 | .0544092 .0333224 1.63 0.103 -.0109015 .11972
open5 | .0083725 .0034342 2.44 0.015 .0016416 .0151033
imf5 | .5557548 .3366187 1.65 0.099 -.1040057 1.215515
rgdp5l | .1871881 .1588481 1.18 0.239 -.1241484 .4985247
rgdppc5l | -.2959381 .251913 -1.17 0.240 -.7896785 .1978022
ukcol | .7088346 .4887037 1.45 0.147 -.2490071 1.666676
frcol | 1.118685 .6135677 1.82 0.068 -.0838854 2.321256
portcol | .5306706 .8215666 0.65 0.518 -1.07957 2.140911
belgcol | 2.32518 1.043537 2.23 0.026 .2798857 4.370474
kennedy | -.0714685 .9135847 -0.08 0.938 -1.862062 1.719125
tokyo | -1.740184 .8039042 -2.16 0.030 -3.315808 -.164561
uruguay | 1.336772 .432955 3.09 0.002 .4881959 2.185348
wto | -.2548981 .9514136 -0.27 0.789 -2.119635 1.609838
------
Model 4
stcox gwcov5 pta5 ksgpolity5 xz5 xw5 zw5 xzw5 gwregcov5 open5 imf5 rgdp5l rgdppc5l ukcol frcol portcol belgcol kennedy t
> okyo uruguay wto, efron cluster(ccode1) nohr
failure _d: member == 1
analysis time _t: yearindep
id: ccode1
Iteration 0: log pseudolikelihood = -162.69038
Iteration 1: log pseudolikelihood = -138.96895
Iteration 2: log pseudolikelihood = -136.40749
Iteration 3: log pseudolikelihood = -136.23481
Iteration 4: log pseudolikelihood = -136.23253
Iteration 5: log pseudolikelihood = -136.23253
Refining estimates:
Iteration 0: log pseudolikelihood = -136.23253
Cox regression -- Efron method for ties
No. of subjects = 61 Number of obs = 891
No. of failures = 50
Time at risk = 891
Wald chi2(20) = 64.90
Log pseudolikelihood = -136.23253 Prob > chi2 = 0.0000
(Std. Err. adjusted for 61 clusters in ccode1)
------
| Robust
_t | Coef. Std. Err. z P>|z| [95% Conf. Interval]
------+------
gwcov5 | .0778504 .0287117 2.71 0.007 .0215765 .1341242
pta5 | .0740456 .0940104 0.79 0.431 -.1102114 .2583025
ksgpolity5 | -.4419072 .2890963 -1.53 0.126 -1.008526 .1247113
xz5 | -.0011206 .0010239 -1.09 0.274 -.0031274 .0008862
xw5 | .0064317 .0033991 1.89 0.058 -.0002305 .0130939
zw5 | .0057688 .0121586 0.47 0.635 -.0180616 .0295993
xzw5 | -.0000848 .0001311 -0.65 0.518 -.0003418 .0001721
gwregcov5 | .0380264 .0316177 1.20 0.229 -.0239432 .0999959
open5 | .0072596 .0030908 2.35 0.019 .0012017 .0133175
imf5 | .6513271 .35665 1.83 0.068 -.0476942 1.350348
rgdp5l | .1819448 .1620265 1.12 0.261 -.1356213 .499511
rgdppc5l | -.2208115 .2860196 -0.77 0.440 -.7813996 .3397766
ukcol | .4164957 .559272 0.74 0.456 -.6796572 1.512649
frcol | 1.096902 .6184539 1.77 0.076 -.1152453 2.309049
portcol | .5666998 .7792969 0.73 0.467 -.960694 2.094094
belgcol | 2.258241 1.13122 2.00 0.046 .0410902 4.475391
kennedy | -.068955 .9791696 -0.07 0.944 -1.988092 1.850182
tokyo | -1.752097 .813389 -2.15 0.031 -3.34631 -.1578842
uruguay | 1.607806 .416246 3.86 0.000 .7919793 2.423634
wto | .6717705 .802743 0.84 0.403 -.9015768 2.245118
------
Interaction charts, Figures 2-4
The included Stata .do files (“GATTWTO_coverage*PTA.do”, “GATTWTO_coverage*POLITY.do”, and “3way_interaction_Model 4_Table_2.do”) will reproduce Figures 2-4 in the paper.