Research ArticleCOMPLEX SYSTEMS

Systemic trade risk of critical resources

See allHide authors and affiliations

Science Advances  13 Nov 2015:
Vol. 1, no. 10, e1500522
DOI: 10.1126/sciadv.1500522
  • Fig. 1 The worldwide trade risk network for nonfuel minerals.

    (A to C) The worldwide trade risk network for nonfuel minerals, represented as a multiplex trade network Vijr(t), where each layer corresponds to one mineral resource: (A) copper, (B) lithium, and (C) platinum group metals. We study the network topology of each of these layers and compute both regional (node-based, country-specific) and global (network-based) measures. We study the relationships between supply risk, price volatility, network centrality, and trade barriers for the United States and the EU (world regions highlighted in green on the world map).

  • Fig. 2 TradeRisk versus price volatility for the EU and the United States.

    Each point represents a mineral resource. (A and B) The country-specific TradeRisk indicator for (A) the EU and (B) the United States is significantly correlated with both the average yearly price volatility of the specific mineral and the composite supply risk, indicated by color. Resources with high Sr tend to be on the right-hand side. We also show the correlation coefficients ρvol and ρCSR of the price volatility with TradeRisk and composite supply risk, respectively, together with the P values to reject the null hypothesis that the true correlation coefficient is 0.

  • Fig. 3 Ranks of TradeRisk in the EU and the United States.

    Each point represents a single resource. Rank 1 is given to the resource with the highest TradeRisk in the given region, rank 2 is given for the second highest TradeRisk, and so on. Resources where information is only available for either the EU or the United States are shown outside the plot area. Major metals are shown as black boxes, minerals that are by-products are shown as gray circles, and other minerals are shown as light gray diamonds. It is clearly visible that minerals that have high TradeRisk values in both regions are mined as by-products, whereas the major metals exhibit intermediate TradeRisk values.

  • Table 1 Global properties of the trade networks for each resource r.

    The elements are Pearson correlation coefficients. The composite supply risk Sr is negatively correlated with the largest eigenvalue, λr, and the size of the SCC, Cr. Cr is positively correlated with both the total trading volume and the scarcity of the resource. The higher the scarcity of the mineral is, the lower is the resilience to shocks of the trade risk network. These correlations cannot be explained by a potentially confounding influence of the trade volume itself, as seen by the nonsignificant correlations of λr and Cr with vr.

    Correlation coefficientSrsrvr
    Largest eigenvalue, λr−0.32*0.47**0.21
    SCC size, Cr−0.41**0.45***0.05

    *Significant at P < 0.05.

    **Significant at P < 0.01.

    ***Significant at P < 0.001.

    • Table 2 Regional results for the correlations of TradeRisk indicators, price volatilities, and trade barriers.

      Price volatility of mineral resources is best explained using the TradeRisk indicator for both the EU and the United States. There are also significant correlations between price volatility and import reliance, PageRank, and In-Strength TradeRisk. The level of applied protection (trade barriers) bir is negatively correlated with TradeRisk in the United States but not in the EU.

      Correlation withCommentsσEUrσUSrbEUrbUSr
      TradeRisk TirFull network effects and import reliance0.71***0.58***−0.11−0.39**
      Import reliance IirNo use of trade networks0.48**0.51***−0.15−0.10
      PageRank PirFull network effects, no import reliance used0.56***0.45***−0.23−0.43***
      In-Strength TradeRisk Tistr,rNo network effects (only contributions from the nearest neighbors and import reliance)0.39*0.50***−0.12−0.11

      *Significant at P < 0.05.

      **Significant at P < 0.01.

      ***Significant at P < 0.001.

      Supplementary Materials

      • Supplementary material for this article is available at http://advances.sciencemag.org/cgi/content/full/1/10/e1500522/DC1

        Materials and Methods

        Fig. S1. Dependence of the correlations between price volatilities and TradeRisk on α.

        Table S1. Network-based properties, supply risk, and indicators obtained from trade data for 71 nonfuel mineral resources.

        Table S2. Global properties of the randomized trade networks Mrfix degree(t).

        Table S3. Pearson correlation coefficients between various price volatilities with TradeRisk, import reliance, PageRank, and in-strength for the EU and the United States and for several variants of the calculations.

      • Supplementary Materials

        This PDF file includes:

        • Materials and Methods
        • Fig. S1. Dependence of the correlations between price volatilities and TradeRisk on α.
        • Table S1. Network-based properties, supply risk, and indicators obtained from trade data for 71 nonfuel mineral resources.
        • Table S2. Global properties of the randomized trade networks Mrfix degree(t).
        • Table S3. Pearson correlation coefficients between various price volatilities with TradeRisk, import reliance, PageRank, and in-strength for the EU and the United States and for several variants of the calculations.

        Download PDF

        Files in this Data Supplement:

      Stay Connected to Science Advances

      Navigate This Article