Research ArticleECONOMICS

Interpreting economic complexity

See allHide authors and affiliations

Science Advances  09 Jan 2019:
Vol. 5, no. 1, eaau1705
DOI: 10.1126/sciadv.aau1705
  • Fig. 1 Interpreting the ECI as a spectral clustering method.

    Each panel shows the ECI vector (in ascending order) (left) and the associated similarity matrix S (right), where rows and columns have been ordered by the ECI and colored by the Sij values. Panels correspond to similarity networks based on (A) randomly generated data with two clear components, (B) HS6 COMTRADE data for 2013, (C) data on employment concentrations in different industries in U.K. local authorities (LAs), and (D) data on employment concentrations in different occupations in U.S. states.

  • Fig. 2 ECI versus income per capita.

    (A) Relationship between the ECI and log GDP per capita for data on countries and exports. (B) Relationship between the ECI and log per-capita earnings for data on industrial employment concentrations in U.K. local authorities. As the scatterplot is too tightly clustered to show legible local authority labels, we provide the top and bottom 10 local authorities ranked by their ECI in the Supplementary Materials. (C) Relationship between the ECI and log GDP per capita for data on occupational employment concentrations in U.S. states.

  • Fig. 3 Ordering rows and columns of M by the ECI and PCI.

    In each matrix, rows are sorted by the ECI and columns are sorted by the PCI. (A) Country-product M matrix; (B) U.K. region-industry M matrix; (C) U.S. state-occupation M matrix.

  • Fig. 4 ECI versus diversity.

    Relationship between diversity and the ECI for data on (A) countries and exports, (B) U.K. regions and industries, and (C) U.S. states and occupations.

  • Fig. 5 Ordering rows and columns of M by diversity and ubiquity.

    In each matrix, rows are sorted by diversity and columns are sorted by ubiquity. (A) Country-product M matrix; (B) U.K. region-industry M matrix; (C) U.S. state-occupation M matrix.

  • Fig. 6 Diversity versus income per capita.

    (A) Relationship between diversity and log GDP per capita for data on countries and exports. (B) Relationship between diversity and log per-capita earnings for data on industrial employment concentrations in U.K. local authorities. (C) Relationship between diversity and log GDP per capita for data on occupational employment concentrations in U.S. states.

Supplementary Materials

  • Supplementary material for this article is available at http://advances.sciencemag.org/cgi/content/full/5/1/eaau1705/DC1

    Section S1. Diversity and degree equivalence

    Section S2. Relationship between the ECI and PCI

    Section S3. Interpretation of ECI as a diffusion map and relationships to correspondence analysis and kernel principal component analysis

    Section S4. ECI and PCI rankings for regional data

    Section S5. Eigengap heuristic analysis

    Section S6. Robustness of empirical results to alternative RCA thresholds

    Fig. S1. Application of diffusion map interpretation to country export data.

    Fig. S2. Top largest eigenvalues of the Formula matrix for data on exports, U.K. regional industrial concentrations, and U.S. state occupational concentrations.

    Fig. S3. Robustness of ECI versus GDP/cap relationship to varying the RCA export threshold.

    Fig. S4. Country-product M matrix with rows sorted by the ECI and columns sorted by the PCI constructed using different RCA thresholds.

    Fig. S5. Robustness of ECI versus GDP/cap relationship to varying the RCA per-capita threshold.

    Table S1. Top and bottom 10 U.K. local authorities ranked by ECI.

    Table S2. Top and bottom 10 industries ranked by PCI.

    Table S3. Top and bottom 10 U.S. states ranked by ECI.

    Table S4. Top and bottom 10 occupations ranked by PCI.

    References (3335)

  • Supplementary Materials

    This PDF file includes:

    • Section S1. Diversity and degree equivalence
    • Section S2. Relationship between the ECI and PCI
    • Section S3. Interpretation of ECI as a diffusion map and relationships to correspondence analysis and kernel principal component analysis
    • Section S4. ECI and PCI rankings for regional data
    • Section S5. Eigengap heuristic analysis
    • Section S6. Robustness of empirical results to alternative RCA thresholds
    • Fig. S1. Application of diffusion map interpretation to country export data.
    • Fig. S2. Top largest eigenvalues of the M~ matrix for data on exports, U.K. regional industrial concentrations, and U.S. state occupational concentrations.
    • Fig. S3. Robustness of ECI versus GDP/cap relationship to varying the RCA export threshold.
    • Fig. S4. Country-product M matrix with rows sorted by the ECI and columns sorted by the PCI constructed using different RCA thresholds.
    • Fig. S5. Robustness of ECI versus GDP/cap relationship to varying the RCA per-capita threshold.
    • Table S1. Top and bottom 10 U.K. local authorities ranked by ECI.
    • Table S2. Top and bottom 10 industries ranked by PCI.
    • Table S3. Top and bottom 10 U.S. states ranked by ECI.
    • Table S4. Top and bottom 10 occupations ranked by PCI.
    • References (3335)

    Download PDF

    Files in this Data Supplement:

Navigate This Article