Fig 1.
How do we form the secondary networks?
We start with primary networks where nodes are countries and the links are total annual exports from one country to the other. We form 15 such networks for each year starting from 1995 until 2009. We calculate PageRanks and Strength of each country in these networks and form time series of PageRanks and Strengths for each country. Finally we calculate positive and negative correlations between these time series and form 4 different secondary networks, where nodes are countries and links are one of the following: positive correlations between PageRanks series, negative correlations between PageRanks series, positive correlations between Strengths series or negative correlations between Strengths series. We repeat the same procedure for the sectors.
Fig 2.
Evolution of the Country network over time.
Top 6 ranked countries by Pagerank (left) and by normalized strength (right).
Fig 3.
Evolution of sectors over time.
Top 6 ranked sectors by pagerank (left) and strength (right).
Fig 4.
1995 and 2009 Country-WION community structure.
In this map countries of the same community are of the same color, with the exception of white countries (non-coloured countries), which are those that are not in the World Input-Output Dataset and dashed countries which are on the dataset but make a community on their own. It can be observed that the communities in this network are mostly geographical and how the countries of the former Eastern Block changed their trade from Russia to Germany.
Fig 5.
Correlations of Countries’ PageRank.
Left: Correlated countries - Two countries are connected with a dark grey link if the Pearson coefficient between their PageRank time series is above if the coefficient is only above
the link is light grey. Right: Anticorrelated countries - Two countries are connected in dark grey if their PageRank time series have a Pearson coefficient below
if the coefficient is only below
the link is light grey. In both the size of of the nodes is proportional to the strength of the node (regarding light edges).
Fig 6.
Countries Normalised Strength correlation.
Left: Correlated countries - Two countries are connected with a dark grey link if the Pearson coefficient between their strengths time series is above if the coefficient is only above
the link is light grey. Right: Anticorrelated countries - Two countries are connected in dark grey if their strengths time series have a Pearson coefficient below
if the coefficient is only below
the link is light grey. In both the size of of the nodes is proportional to the strength of the node (regarding light edges).
Fig 7.
Correlation of Sectors’ Pagerank.
Left: Correlated sectors - Two sectors are connected with a dark grey link if the Pearson coefficient between their PageRank time series is above if the coefficient is only above
the link is light grey. Right: Anticorrelated countries - Two sectors are connected in dark grey if their PageRank time series have a Pearson coefficient below
if the coefficient is only below
the link is light grey. In both the size of of the nodes is proportional to the strength of the node (regarding light edges).
Fig 8.
Correlation of Sectors’ Normalised Strength.
Left: Correlated sectors - sectors are connected with a dark grey link if the Pearson coefficient between their strengths time series is above if the coefficient is only above
the link is light grey. Right: Anticorrelated sectors - Two sectors are connected in dark grey if their strengths time series have a Pearson coefficient below
if the coefficient is only below
the link is light grey. In both the size of of the nodes is proportional to the strength of the node (regarding light edges).