The dataset provided by Gleditsch (2002) includes 196 countries for which there are data on trade flows from 1948 to 2000. However, trade data contain many missing (or badly reported) values before 1970. In addition, there are some countries with zero total exports in some years.
Notice also that our analysis requires to match trade data with real GDP (both in levels and per capita).
This is because: (i) weights are defined as exports divided by GDP; (ii) one wants to cross-sectionally correlate network measures with country-specific variables like per-capita GDP.
We have therefore selected countries in such a way to have: (i) a time horizon and a country sample size as long as possible; (ii) no missing values in trade data and GDP (both in levels and per capita); (iii) non-zero total exports.
By applying conditions (i) and (ii) we get only 83 countries from 1960-2000. This number becomes 138 for the period 1970-2000; 152 for the period 1970-2000; 163 for the period 1981-2000; and 168 for the period 1990-2000. We thus decided to select the time interval 1981-2000 using 163 countries. However, 4 of them (San Marino, Andorra, Liechtenstein, Monaco) have total exports equal to zero in some years. This leaves us with N=159 countries, whose list is in Table 1.
Table1:ListofCountriesintheBalancedPanel. IdAcroNameIdAcroNameIdAcroNameIdAcroNameIdAcroNameIdAcroName 2USAUnitedStates150PARParaguay395ICEIceland520SOMSomalia678YEMYemen910PNGPapua 20CANCanada155CHLChile402CAPCapeVerde522DJIDjibouti690KUWKuwait920NEWNewZealand 31BHMBahamas160ARGArgentina403STPSaoTome530ETHEthiopia692BAHBahrain935VANVanuatu 40CUBCuba165URUUruguay404GNBGuinea-Bissau540ANGAngola694QATQatar940SOLSolomon’s 41HAIHaiti200UKGUnitedKingdom411EQGEq.Guinea541MZMMozambique696UAEArabEmirates950FJIFiji 42DOMDominicanRep.205IREIreland420GAMGambia551ZAMZambia698OMAOman970KBIKiribati 51JAMJamaica210NTHNetherlands432MLIMali552ZIMZimbabwe700AFGAfghanistan971NAUNauru 52TRITrinidad/Tobago211BELBelgium433SENSenegal553MAWMalawi710CHNChina972TONTonga 53BARBarbados212LUXLuxembourg434BENBenin560SAFSouthAfrica712MONMongolia973TUVTuvalu 54DMADominica220FRNFrance435MAAMauritania570LESLesotho713TAWTaiwan 55GRNGrenada225SWZSwitzerland436NIRNiger571BOTBotswana731PRKNorthKorea 56SLUSaintLucia230SPNSpain437CDICoteDIvoire572SWASwaziland732ROKSouthKorea 57SVGSt.Vincent235PORPortugal438GUIGuinea580MAGMadagascar740JPNJapan 58AABAntigua260GFRGermany439BFOBurkinaFaso581COMComoros750INDIndia 70MEXMexico290POLPoland450LBRLiberia590MASMauritius760BHUBhutan 80BLZBelize305AUSAustria451SIESierraLeone591SEYSeychelles770PAKPakistan 90GUAGuatemala310HUNHungary452GHAGhana600MORMorocco771BNGBangladesh 91HONHonduras325ITAItaly461TOGTogo615ALGAlgeria775MYAMyanmar 92SALElSalvador338MLTMalta471CAOCameroon616TUNTunisia780SRISriLanka 93NICNicaragua339ALBAlbania475NIGNigeria620LIBLibya781MADMaldives 94COSCostaRica345YUGYugoslavia481GABGabon625SUDSudan790NEPNepal 95PANPanama350GRCGreece482CENCentrAfricanRep.630IRNIran800THIThailand 100COLColombia352CYPCyprus483CHAChad640TURTurkey811CAMCambodia 101VENVenezuela355BULBulgaria484CONCongo645IRQIraq812LAOLaos 110GUYGuyana360RUMRumania490DRCCongo(Zaire)651EGYEgypt816DRVVietnam 115SURSurinam365RUSRussia500UGAUganda652SYRSyria820MALMalaysia 130ECUEcuador375FINFinland501KENKenya660LEBLebanon830SINSingapore 135PERPeru380SWDSweden510TAZTanzania663JORJordan840PHIPhilippines 140BRABrazil385NORNorway516BUIBurundi666ISRIsrael850INSIndonesia 145BOLBolivia390DENDenmark517RWARwanda670SAUSaudiArabia900AULAustralia
Table 2: Countries in the core
1981 1985 1990 1995 2000
USA USA USA USA USA
Japan Japan Germany Germany Germany
Germany† Germany† Japan Japan Japan
UK UK France France France
France France UK UK UK
USSR USSR Italy Italy China
Italy Italy USSR Belgium Italy
Saudi Arabia Netherlands Netherlands Netherlands Korea
† Up to 1989 data refers to West Germany only.
19800 1985 1990 1995 2000
19800 1985 1990 1995 2000
0.1
19800 1985 1990 1995 2000
0.1
19800 1985 1990 1995 2000
0.1
Figure 1: Global network indicators vs. years. Top-left: Network density. Top-right: Percentage of bilateral links. Bottom-left:
S index (not standardized) for BUNs. Bottom-right: S index (not standardized) for WUNs.
Figure2:TheWTWasaBUN.Degreedistributionsin1981,1990,2000.Toppanels:Size-rankplots(dashedline:lognormalfit).Bottompanels:Kerneldensityestimates. 2.533.544.555.5
−5
−4
−3
−2
−10 Log(Degree)
Log(Rank)
1981 Degree LogNormal Fit 2.533.544.555.5
−5
−4
−3
−2
−10 Log(Degree)
Log(Rank)
1990 Degree LogNormal Fit 2.533.544.555.5
−5
−4
−3
−2
−10 Log(Degree)
Log(Rank)
2000 Degree LogNormal Fit 0501001500
0.002
0.004
0.006
0.008
0.01 Degree
Kernel Density
1981 0501001500
0.002
0.004
0.006
0.008
0.01 Degree
Kernel Density
1990 0501001500
0.002
0.004
0.006
0.008
0.010.012 Degree
Kernel Density
2000 Note.Size-rankplots.X-axis:logofdegree.Y-axis:Logofrankofx-axisobservation.
1985 1990 1995 2000 0
20 40 60 80 100 120 140 160
Year
Node−Degree Statistics
Average StdDev
1980 1985 1990 1995 2000
1.95 2 2.05 2.1 2.15 2.2
Kurtosis
Year
−0.2
−0.1 0 0.1 0.2 0.3
Skewness
Figure 3: The WTW as a BUN. Left panel: Average and standard deviation of degree distributions. Right panel: kurtosis (circles) and skewness (asterisks) of degree distributions.
Figure4:TheWTWasaWUN.Strengthdistributionsin1981,1990,2000.Toppanels:Size-rankplots(dashedline:lognormalfit).Bottompanels:Kerneldensityestimates. −8−6−4−202−7−6−5−4−3−2−10 Log(Strength) Log(Rank)
1981 Strength LogNormal Fit −8−6−4−202−6−5−4−3−2−1
0 Log(Strength)
Log(Rank)
1990 Strength LogNormal Fit −5−4−3−2−101−7−6−5−4−3−2−10 Log(Strength) Log(Rank)
2000 Strength LogNormal Fit 0.511.522.50123456 Strength
Kernel Density
1981 0.511.522.501234567 Strength
Kernel Density
1990 0.511.522.50123456 Strength
Kernel Density
2000 Note.Size-rankplots.X-axis:logofstrength.Y-axis:Logofrankofx-axisobservation.
1985 1990 1995 2000
Figure 5: BUN vs. WUN. Left panel: Degree-strength correlation vs. years (dashed lines: 5% and 95% confidence intervals).
Right panel: Degree-strength scatterplot in 2000.
1985 1990 1995 2000
Figure 6: The WTW as a WUN. Left panel: Average and standard deviation of strength distributions. Right panel: kurtosis (circles) and skewness (asterisks) of strength distributions.
1985 1990 1995 2000
Figure 7: Average Nearest-neighbor degree (ANND). Left: Population average vs. years. Right: Correlation between (ANND) and degree vs. years. Dashed lines: 5% and 95% confidence intervals.
1985 1990 1995 2000
Figure 8: Left: Population-average of weighted average nearest-neighbor degree (WANND) vs. years. Right: Population-average of average nearest-neighbor strength (ANNS) vs. years. Dashed lines: 5% and 95% confidence intervals.
1985 1990 1995 2000
Figure 9: Left: WANND-degree correlation vs. years. Right: ANNS-strength correlation vs. years. Dashed lines: confidence intervals.
Figure 10: Left: ANND-degree scatter plot. Mid: WANND-degree scatter plot. Right: ANNS-strength scatter plot. Year:
2000.
1982 1984 1986 1988 1990 1992 1994 1996 1998 2000
19800 1985 1990 1995 2000
0.05
% of Countries with Excess Clustering
Figure 11: Top-Left: Average of BUN (binary) clustering coefficient vs. years. Dashed lines: 5% and 95% confidence intervals.
Top-Right: Average and expected value of WUN (weighted) clustering coefficient vs. years. Bottom-Left: Kernel density of the WUN (weighted) clustering coefficient in 2000 vs. average and expected values. Bottom-right: Percentage of countries characterized by node clustering above its expected value. Note: Expected values computed by randomly reshuffling for 10000 times (in each year) the observed weights across the existing links.
1982 1984 1986 1988 1990 1992 1994 1996 1998 2000
1982 1984 1986 1988 1990 1992 1994 1996 1998 2000
−1
Figure 12: Top-left: Correlation between (binary) clustering coefficient and degree vs. years. Top-right: Correlation between (weighted) clustering coefficient and strength vs. years. Dashed lines: 5% and 95% confidence intervals. Bottom-left: Scatter plot of (binary) clustering coefficient and degree in year 2000. Bottom-right: Scatter plot of (weighted) clustering coefficient and strength in year 2000.
1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 0
Figure 13: Correlation between degree-pcGDP and strength-pcGDP vs. years. Dashed lines: 5% and 95% confidence intervals.
1985 1990 1995 2000
−1
−0.5 0 0.5 1
Year
Clustering−pcGDP Correlation
1982 1984 1986 1988 1990 1992 1994 1996 1998 2000
−1
−0.8
−0.6
−0.4
−0.2 0 0.2 0.4 0.6 0.8 1
Year
Weighted Clustering−pcGDP Correlation
Figure 14: Left: Correlation between clustering and pcGDP in a BUN vs. years. Right: Correlation between clustering and pcGDP in a WUN vs. years. Dashed lines: 5% and 95% confidence intervals.
Figure15:RandomWalkBetweennessCentralitydistributionsin1981,1990,2000:Kerneldensityestimates. 00.10.20.30.40.50.60.70
10
20
30
40
50
60
70 Random−Walk Betweenness Centrality
Kernel Density
1981 00.10.20.30.40.50.60.70102030405060 Random−Walk Betweenness Centrality Kernel Density
1990 00.10.20.30.40.50.60.70
10
20
30
40
50
60 Random−Walk Betweenness Centrality
Kernel Density
2000
1985 1990 1995 2000
Figure 16: Alternative weighting schemes. Left panels: Exports divided by importer country GDP. Right panels: Exports not scaled by any country size measure. Top: Correlation between strength and ANNS vs. years. Mid: Correlation between clustering and strength vs. years. Bottom: Correlation between strength and pcGDP. Dashed lines: 5% and 95% confidence intervals.