Pedro Miguel locks along the Panama Canal. Photo: Gonzalo Azumendi via Getty Images.
Annex 1: Methodologies
A1.1 Chatham House Maritime Analysis Tool (CH-MAT)
The Chatham House Maritime Analysis Tool (CH-MAT) is a first-of-its-kind (Excel-based) data-analysis tool for assessing the importance of maritime chokepoints to global food trade.330 The CH-MAT couples Chatham House’s Resource Trade Database (CHRTD, see Section A1.2 below) with a set of detailed assumptions on the most likely routes taken by food-carrying dry bulk vessels between any two regions. Here we outline the key steps in our approach to the CH-MAT, and discuss some of its limitations and ways in which these may be addressed.
Key steps
Underpinning the CH-MAT is a set of assumptions relating to the maritime routes via which food shipments are transported from one region to another. We applied these assumptions to bilateral trade data from the CHRTD in order to estimate both the weight and value of staple grains – wheat, maize, rice and soybean – and fertilizers passing through a given chokepoint each year.
Step 1: Maritime chokepoint identification
The first step in our approach was to identify those maritime chokepoints that are of strategic global importance. The basis of this initial assessment was existing analysis of international trade along sea lines of communication, both for general freight and for grain shipments specifically. An extensive literature review was undertaken, spanning transport network analysis and maritime logistics reports,331 Automatic Identification System (AIS)-driven mapping of maritime trade activity,332 and energy supply risk analysis.333
From this literature review, we identified eight global maritime chokepoints that lie along the major east–west trade channels: the Panama Canal, Dover Strait, Strait of Gibraltar, Turkish Straits, Suez Canal, Strait of Bab al-Mandab, Strait of Hormuz and Strait of Malacca. This grouping differs slightly from the list of global oil chokepoints drawn up by the US Energy Information Administration (EIA): we do not include the Danish Straits or Cape of Good Hope (while the EIA does not include the Dover Strait or Strait of Gibraltar). The reasons for this are as follows:
- The Danish Straits are an important channel for trade to and from the Baltic states, and to and from Russia via the Gulf of Finland. For grain and fertilizer, however, the majority of the Baltic states’ trade is with mainland Europe and is transported overland. While Russia’s export capacity on its Baltic coast is expanding, the vast majority of its grain and fertilizer shipments continue to be handled at the Black Sea ports.
- While the Cape of Good Hope is undoubtedly a recognized ‘chokepoint’ in naval settings and in common parlance, it is not considered a chokepoint in the context of our analysis. Unlike the eight maritime chokepoints identified above, the cape is neither a narrow strait nor a man-made canal; thus, while passage may be impeded by poor weather or other hazards, the nature of this impediment will differ fundamentally from the complete blockages possible through straits or canals.
The Danish Straits and Cape of Good Hope are nevertheless included in the overall CH-MAT database for added clarity in routing assignments. So too are Cape Horn, the Northern Sea Route, the South China Sea and the East China Sea (see Figure 26).
Figure 26: Chokepoints used in the elaboration of Chatham House Maritime Analysis Tool
—
Figure 26: Chokepoints used in the elaboration of Chatham House Maritime Analysis Tool
Step 2: Regional groupings
Next, each country was assigned to a regional grouping based on the most likely maritime route via which it would trade (see Figure 27). This assignation gave rise to a set of country groupings to which common trading routes may be applied; this step allowed for the 10,884 bilateral grain trade flows in the CHRTD to be converted into a more manageable list of 42 region-to-region flows.
This step implies no trade-off in terms of granularity; bilateral flows can still be interrogated and estimated at the country level. And it brings considerable gains in terms of developing plausible routes; standard regional breakdowns have little relevance to the consideration of convenient coastal access routes.
For the most part, the 42 regions identified comprise multiple – up to 26 – countries. However, the CH-MAT also has seven single-country regions, such as Canada, Russia and the US.334 Single-country regions are used for countries that have two or more major coastlines and enjoy access to multiple maritime routing options for imports and exports. Canada, for example, may export its agricultural goods from either its western or eastern coast, depending on whether the shipments in question are destined for Asian or European markets. Similarly, Russian fertilizers may be shipped from the country’s ports on the Baltic Sea, the Black Sea or the Sea of Japan; were Russia simply grouped with other exporting countries in Eastern Europe – Belarus or Ukraine, for example – important trading differences would be overlooked.
Figure 27: Regional groupings used in Chatham House Maritime Analysis Tool
— Source: Authors’ own analysis.
Figure 27: Regional groupings used in Chatham House Maritime Analysis Tool
Step 3: Trade routes and chokepoint assumptions
The third step involves assigning each region-to-region flow to one or several assumed maritime trading routes. In light of the focus of our research, maritime chokepoints are used as a proxy for trading routes; queries can be run in the CH-MAT on the basis of trade transiting the Panama Canal specifically, for example, but not on the basis of trade travelling westbound or eastbound via unspecified routes between the Atlantic and Pacific oceans.
Our routing assumptions are based primarily on distance and shipping time. Where two or more routing options are available (for example, trade between Western Europe and East Africa may be shipped via the Suez Canal, or via the South Atlantic and under the Cape of Good Hope), a decision is made on whether to assign the totality of the region-to-region flow to one route or to split it between two. In those cases where the difference in shipping time between two routes is three days or greater, the shortest route is preferred; where two routing options are of comparable distance – with a difference in shipping time of less than three days – the region-to-region flow is split equally between both routes. The distance between region x and region y is calculated using the online searates.com tool.
Where possible, additional data and anecdotal evidence are used to sense-check and fine-tune routing assumptions. In those regions that span an extensive seaboard and include multiple countries – and those regions that include more than one seaboard – we use available port throughput data and existing analysis of principal grain trade hubs to pinpoint likely entry and exit points for the region as a whole. By way of example, ports at the southern end of our Western Africa region are the most important for trade in cereals within that region, and so we use these ports as the basis for all trade route assumptions to and from the Western Africa region. In the case of Mexico, national port throughput data indicate that both the east and west coasts have important grain-handling terminals, though the eastern terminals have roughly five times’ the capacity of those on the west coast. We therefore assume that 15 per cent of cereals imports enter the country from the Pacific and 85 per cent via the Gulf of Mexico.
For certain regions, more than one grain trade hub is identified. In these cases, we apply sub-regional routing divisions and apply these consistently, irrespective of the origin or destination of trade. This consistency maximizes the replicability of our approach and ensures that our assumptions are based as far as possible on the actual capacity of alternative port areas. This is the case for Russia, which, as noted above, has three possible export and import routes: via the Baltic Sea or Black Sea in the west of Russia, or via the Sea of Japan in the east. Relatively detailed port capacity data for Russia, made available by the US Department of Agriculture (USDA), provide a basis on which to apportion grain imports and exports between the three areas. These data, together with anecdotal evidence from other sources, indicate that the far eastern trade terminals – in and around Vladivostok – are expanding but not yet handling significant volumes of grain shipments (in part because the country’s wheat production is centred in Russia’s western and southern regions). The Baltic ports handle a reasonable share of grain shipments, but the Black Sea ports provide the primary access point for international markets. We therefore route 90 per cent of Russia’s agricultural exports and imports via the Black Sea, in line with port throughput and capacity data, and do so consistently for all region-to-region flows.
Common routing assumptions are applied to all four grains in all regions except the US. The US is unique both in the granularity of its port throughput data and in the marked differences between export patterns for wheat, maize and soybean. Grain export data are broken down both by grain and by port region, allowing us a degree of nuance not possible in other regions. On the basis of these data, we divide the US into three major sub-regions (the Gulf Coast, Pacific northwest coast and east coast), and apply different assumptions for each crop type (see Table 8).
Table 8: Share of US exports for staple crops shipped from major US port areas
|
Port area |
Soybean and rice |
Wheat |
Maize |
|---|---|---|---|
|
Pacific northwest coast |
25% |
47% |
17% |
|
Gulf Coast |
59% |
40% |
67% |
|
East coast |
7% |
7% |
2% |
|
Other |
9% |
6% |
13% |
Source: Authors’ analysis based on USDA (2015), Grain Transportation Report – February 12, https://www.ams.usda.gov/services/transportation-analysis/gtr/archive-2015 (accessed 25 Jan. 2017).
Each region-to-region routing assumption is stored within a two-dimensional array of 1,766 rows (representing every possible region-to-region flow in the database), and 14 fields (representing the eight maritime chokepoints deemed of global strategic importance, plus the Danish Straits, Cape of Good Hope, Cape Horn, Northern Sea Route, South China Sea and East China Sea). Trade flows attributed to ‘areas not elsewhere specified’ in the CHRTD are not assigned a routing assumption in our model.
Methodological limitations
As the first tool of its kind, the CH-MAT includes a number of assumptions and limitations which warrant further probing if the tool’s accuracy and robustness are to be strengthened. Below, we briefly outline these and identify priorities for future research.
Assumptions
On the basis of an extensive literature review and a series of conversations with maritime industry experts, we make the assumption that all grain trade is transported by dry bulk vessels. This is because, while the use of containers in the transport of grain is increasing, available evidence suggests that containerized shipments as a share of total shipments remain low.335 Our routing assumptions therefore imply single point-to-point voyages. If liner routes were also considered, our approach to region-to-region routing assumptions (based on the distance and shipping time between major trade hubs in region x and region y) might no longer be valid; liner vessels would likely call at other ports en route – picking up or dropping off grain shipments from other trading partners – rather than steaming directly from region x to region y.
Furthermore, the CH-MAT does not account for the movement of vessels empty of cargo. This is because the underlying data for the CH-MAT consist of bilateral trade data rather than port throughput data. Taking only cargo-carrying vessels into account makes sense when assessing the strategic importance of certain routes to food security; in the case of a major blockage to the Panama Canal, for instance, vessels carrying grain shipments, not those that are empty, will be of concern to the food security community. But as empty vessels can also be used in identifying potential points of congestion, their omission from our analysis may lead to an underestimate of usage levels at certain chokepoints or along certain trade routes.
We apply to fertilizer flows the same set of routing assumptions as those developed for grain flows. Having consulted available data on fertilizer shipments to and from major consumers and producers, we are confident that this approach is valid. However, further analysis of the specific patterns of fertilizer trade and of specialized fertilizer-handling terminals would be needed to be sure that no important differences between fertilizer and grain trading patterns have been missed.
Where both maritime and overland trade routes exist between region x and region y and are of comparable distance, we assume that the maritime route will be taken. The exceptions are where seaborne shipments would imply a significant detour: for example, in the case of trade between landlocked countries in Central Asia.
Limitations
In developing our approach to the CH-MAT, we faced a number of challenges arising from poor data availability. Very little analysis has been undertaken of the seaborne movement of grain and/or fertilizer; and while there exists a relatively rich body of literature on global transport networks and patterns of international maritime trade, the idiosyncrasies of shipping data are such that commodity-specific analysis is extremely difficult.
AIS data distinguish between vessel types, for example, but do not detail the cargo carried by any given vessel. Apps such as MarineTraffic allow users to follow the movement of a particular dry bulk carrier but do not indicate the nature of the dry bulk cargo on board; as such, it is not possible to distinguish between coal shipments and wheat shipments, for example. And while rudimentary data – the vessel type and forecast route, for example – are open-access, further details such as the previous and next ports of call are behind a pay wall and therefore likely to be unavailable to many civil society and public actors.
The quality and availability of national data on port capacity, export flows and import flows are extremely variable. Some countries (such as the US and Saudi Arabia) collect and make public fine-grained data on port capacity and throughput, sometimes broken down by commodity type. Others differentiate only between container shipments and dry bulk shipments. Turkey, for example, reports on dry bulk throughput at its ports, but does not distinguish between grain and other dry bulk such as coal and sand. Other countries (such as the Pacific Island states) have little or no open-access data on import and export shipments, let alone statistics for individual port areas.
Consistency in port throughput data across countries is further limited by discrepancies in reporting. While certain countries report in weight-in-tonnage, others report in value-in-currency; and, while certain countries are consistent in their annual reporting, others provide data on an ad hoc basis. We apply consistent rules when selecting data for use in the CH-MAT – data in weight are preferred over data in value wherever possible; the most recent data available in the period 2010–16 are always used – but discrepancies certainly remain.
Finally, the use of bilateral trade data itself implies certain limitations in analysing the patterns of trade in grain. Trade data are reported only annually, and as such the CH-MAT estimates aggregate chokepoint throughput volumes on a yearly basis. This masks important seasonal fluctuations in the export and import of grain; peak flows from South America during the soybean harvest season are balanced out by yearly lows earlier in the cropping cycle, for instance. While a valuable indication of port areas of systemic interest, the CH-MAT therefore offers little basis for capturing periods of relatively high risk when a major disruption to Brazil’s southern ports could, for example, have a disproportionately large impact on global soybean supply.
Review
A number of maritime industry experts and food security analysts provided valuable feedback on our approach to the project since its inception in early 2015. In the summer of 2016, we undertook a more formal process of peer review through which we presented the key steps, assumptions and limitations of the CH-MAT. The reviewers were Dr Tristan Smith (University College London Energy Institute), Dr Michael Traut (University of Manchester Tyndall Centre) and Dr Conor Walsh (University of Manchester Tyndall Centre). We are extremely grateful to these reviewers for their feedback and guidance. However, we remain solely responsible for the development of the CH-MAT and for any errors it may contain. We welcome feedback on how this important tool may be improved and expanded in the future.
A1.2 Chatham House Resource Trade Database (CHRTD)
Richard King, Felix Preston and Siân Bradley
The Chatham House Resource Trade Database (CHRTD) is a repository of statistics on bilateral trade in natural resources between more than 200 countries and territories.336 The database includes the monetary values and masses of trade in over 1,350 different types of natural resources and resource products, including agricultural, fishery and forestry products, fossil fuels, metals and other minerals, and pearls and gemstones. It contains raw materials, intermediate products and by-products.
Dealing with complexity
Bilateral statistics are critical to understanding global resource trade, but existing data are often difficult to access and use. The original data source for the CHRTD is the International Merchandise Trade Statistics (IMTS). IMTS data are collected by national customs authorities and compiled into the United Nations Commodity Trade Statistics Database (UN Comtrade)337 by the United Nations Statistics Division. UN Comtrade utilizes three distinct trade classification systems: the Harmonized Commodity Description and Coding System (HS), the Standard International Trade Classification (SITC), and Broad Economic Categories (BEC). Of these, the CHRTD employs the HS taxonomy (1996 revision), which assigns HS codes to all forms of traded goods in a hierarchical structure (two-, four- and six-digit codes respectively represent commodity chapters, headings and subheadings).
Across the resource landscape alternative repositories of trade data are available, but these do not offer the breadth and depth of analysis that UN Comtrade permits. For example:
- Data from the United Nations Conference on Trade and Development (UNCTAD)338 have the greatest temporal availability, with some aggregate categories dating back to 1948, but even recent series lack commodity-level (six-digit HS code) detail.
- The FAO339 provides comprehensive bilateral trade data for agriculture, forestry, fisheries and aquaculture, but not for other resource domains. Much of the available data share the same origins as UN Comtrade data and are not necessarily more accurate.
- The USDA Global Agricultural Trade System (GATS)340 is similarly domain-constrained and focuses on US trading partners. Similarly, the European Commission’s EUROIND database341 covers only trade with and between EU countries.
- The IEA342 provides comprehensive energy balance and energy flow statistics, but trade statistics are limited to gas flows within and to Europe.
- US EIA data343 include national energy consumption, production, and import and export statistics, but lack bilateral detail.
- The world oil and gas databases developed by the Joint Organisations Data Initiative (JODI-Oil344 and JODI-Gas345) were developed as transparency tools rather than trade databases. Unlike other databases, JODI does not adjust reported figures or substitute missing figures, so coverage is incomplete.
- Commercial sources such as the BP Statistical Review of World Energy346 provide comprehensive energy trade data, but no bilateral dimension.
UN Comtrade is therefore arguably the most comprehensive source of merchandise trade statistics available; volumetric and monetary value data are catalogued under more than 5,000 HS codes, and the monetary values of trades are available as far back as 1962. However, it does present several challenges for users focusing on resource trade, which the CHRTD and Chatham House’s resourcetrade.earth site address:
- The HS system is not easy to use. Its nomenclature has evolved historically as a pragmatic and comprehensive industrial taxonomy for the broad range of internationally traded goods.
- The scale hinders simple queries: with over 3 billion trade records since 1962,347 finding the right data is not always easy and the size of data queries can be difficult to manage.
The IMTS data are of variable quality: missing data, trade mispricing, unreported and illegal trade, and general mistakes and inconsistencies all cast doubt over the reliability of certain reported trade flows.
The presence of between one and four data points for every trade flow complicates use: both exporters and importers are expected to report trade values and trade masses. If these records are incomplete or do not correspond with one another, there may be uncertainties about the reliability of data and/or about which reporter is the most authoritative.
Introducing clarity
The CHRTD reorganizes data around natural resources. As the IMTS and HS systems contain all types of traded goods – including manufactured goods – analysing natural resource trade flows in UN Comtrade typically requires amalgamating a variety of HS codes. The difficulty of this varies: products that have a long history of being traded extensively are captured in greater detail than those that are traded less frequently. For example, there is a single HS code associated with rare-earth elements, but several hundred codes assigned to steel and steel products. The CHRTD overcomes this problem by selecting over 1,350 HS codes that are identifiable as raw materials or relatively undifferentiated intermediate products, and by grouping them by resource type. For example, copper ores and concentrates, and intermediate copper products such as mattes, bars, wires and scrap are all classified into a single ‘copper’ category, enabling global copper trade to be tracked at different stages of the value chain. The CHTRD employs a five-tier resource taxonomy permitting queries to be as atomized or aggregated as required.
The CHRTD employs a systematic approach to identify and manage data gaps and errors. The CHRTD is subject to the same data gaps and weaknesses as are apparent in other sources of international merchandise trade data.348 However, it exploits the maximum information available within UN Comtrade to assess the reliability of individual trade records, and to present as complete and reliable a picture as possible. The approach taken relies on two assumptions. First, for each trade flow the values (in US$) and masses (in kg) reported by the exporter and importer should approximate to one another. The reported monetary values are unlikely to be exactly the same, since exports are typically reported on a ‘free on board’ (fob) basis, whereas imports are typically reported on a ‘cost, insurance and freight’ (cif) basis. Second, we expect the reported prices per tonne to relate to world market prices. Unlike in some alternative approaches to reconciling importer and exporter reports, we make no assumptions about the general reliability of country reporting across multiple commodities or years; each individual report is assessed on its own merit.
Logical operations are used to produce a transparent decision on the relative reliability of each data point and to reconcile the importer and exporter reports into a single record. Each record incorporates the value and mass of trade in the given commodity between the two countries in the given year. In each case we consider the degree of similarity between the importer and exporter reports. In cases where either trade partner reports the monetary value and the mass of the trade (some reports contain only the value), the reported price per tonne is assessed relative to the global distribution of unit prices for the same commodity in the same year. If both partners report a non-outlying unit price, then a weighted average of the two reports is recorded; the weighting factor is calculated according to the relative divergence of each unit price from world average market prices. Data points that are deemed unreliable and irreconcilable are labelled as such and quarantined. A manual review of some of the larger flows that have been excluded from the database by this process allows us to reintroduce important flows at the global level, using external sources where necessary.
A full specification of this methodology will be made available at a later date.
A1.3 Chatham House Food Security Dashboard (CH-FSD)
The Chatham House Food Security Dashboard (CH-FSD) provides a framework with which to combine existing measures of food insecurity with a new understanding of chokepoint risk. It assesses the chokepoint risk of 205 countries in terms of their exposure (at national level) and vulnerability (at national level and at household level) to chokepoint disruption. The 18 indicators are shown in Table 9.
Table 9: Indicators and sources used in Chatham House Food Security Dashboard
|
Domain |
Indicator |
Data source |
|
Trade dependence |
Cereal import dependency ratio (%) |
FAO |
|
Chokepoint reliance |
Aggregate maize, wheat, rice and soybean imports passing through at least one maritime chokepoint (% by weight) |
CH-MAT |
|
Aggregate maize, wheat, rice and soybean imports passing through a critical maritime chokepoint (% by weight) |
CH-MAT |
|
|
Food insecurity |
Household income spent on in-home food consumption (%) |
World Bank |
|
Prevalence of undernourishment (%) |
FAO |
|
|
Food availability |
Aggregate stock-to-use ratio for maize, wheat, rice and soybean (%) |
USDA |
|
Average dietary energy supply adequacy (%) |
FAO |
|
|
State fragility |
Fragile state? (Y/N) |
OECD |
|
Political stability and absence of violence/terrorism. Composite indicator (approx. -2.5 to 2.5, mean=0, higher values correspond to better governance) |
World Bank |
|
|
Infrastructure quality |
Existence of adequate crop storage facilities? (Y/N) |
EIU |
|
Road infrastructure quality (1–7, 1=worst, 7=best) |
World Economic Forum |
|
|
Rail infrastructure quality (1–7, 1=worst, 7=best) |
World Economic Forum |
|
|
Port infrastructure quality (1–7, 1=worst, 7=best) |
World Economic Forum |
|
|
Climate vulnerability |
People killed/affected by floods, storms, droughts (per 100,000 people – worst year and annual average, 2005–14) |
EM-DAT |
|
Total economic damage caused by floods, storms, droughts (% of GDP – worst year and annual average, 2005–14) |
EM-DAT |
|
|
Social protection |
Social protection coverage for poorest quintile (%) |
World Bank |
|
Macroeconomic vulnerability |
Cash surplus/deficit (% of GDP) |
World Bank |
|
Value of food imports in total merchandise exports (%) |
FAO |
Data sources as follows: Centre for Research on the Epidemiology of Disasters (CRED) (undated), ‘EM-DAT: The International Disaster Database’, http://emdat.be (accessed 15 Jun. 2017); Chatham House Maritime Analysis Tool; Chatham House (2017), ‘resourcetrade.earth’ (2015 data); Economist Intelligence Unit (2016), ‘Global Food Security Index’, http://foodsecurityindex.eiu.com (accessed 3 Jun. 2017); FAO (2016), ‘Food security indicators’, http://www.fao.org/economic/ess/ess-fs/ess-fadata/en/#.WTRQthiZOHp (accessed 22 Mar. 2017); OECD (2015), States of Fragility 2015: Meeting Post-2015 Ambitions, Paris: OECD Publishing, doi: 10.1787/9789264227699-en (accessed 5 Jun. 2017); Schwab, K. (ed.) (2015), The Global Competitiveness Report 2015–2016, Geneva: World Economic Forum, http://www3.weforum.org/docs/gcr/2015-2016/Global_Competitiveness_Report_2015-2016.pdf (accessed 9 May 2017); USDA Foreign Agricultural Service (2017), ‘Production, Supply and Distribution’, https://apps.fas.usda.gov/psdonline/app/index.html#/app/advQuery (accessed 3 Jun. 2017); World Bank (undated), ‘ASPIRE Indicators at a glance’, http://datatopics.worldbank.org/aspire/indicator_glance (accessed 3 Jun. 2017); World Bank (2011), ‘International Comparison Program’, http://siteresources.worldbank.org/ICPEXT/Resources/ICP_2011.html (accessed 3 Jun. 2017); World Bank (2015), ‘Worldwide Governance Indicators’, http://info.worldbank.org/governance/wgi/index.aspx#home (accessed 3 Jun. 2017).
Intended purpose and limitations
The CH-FSD is intended to permit ready analysis and comparison of values within and across indicators, domains and countries. It permits interrogation of the relative and absolute levels of risk across traditional and non-traditional measures of food insecurity, and affords a greater appreciation of the varied interactions of different sources of exposure and vulnerability. We deliberately do not create one composite chokepoint risk indicator from the constituent parts because to do so would be reductionist – stripping out the complexities – and arbitrary, in terms of any decisions about how to weight components differentially. Although we believe this approach is justified, this does mean that the CH-FSD requires careful interpretation and an appreciation of which of the three data point values (see below) is most appropriate for any given interrogation. Additionally, the CH-FSD is constrained by data availability, and by data quality in the underlying indicators relied upon.
Approach
The indicators employed are selected to provide a broad measure of countries’ exposure and vulnerabilities within the domains specified. Among the criteria for inclusion were ready comprehension, lack of overlap with other indicators (composite indices measuring vulnerability across several domains were deliberately excluded), and broad and contemporary country coverage. Nonetheless, indicators vary in terms of country coverage, scale and latest year of data availability. In all instances the most recently available data (at the time of writing) from secondary sources were used. However, the data publication years range from 2008 to 2013; as such, and without annual data for each indicator, the CH-FSD reflects a snapshot based on the best available data, rather than an annual measure that can be compared year on year.
As the various source data series measure disparate phenomena, and are constructed on different scales, a process is required to standardize them for comparison. First, a 90 per cent winsorization is performed to reduce the influence of outliers on the data series. A 90 per cent winsorization sets the bottom 5 per cent of values to the value of the fifth percentile, and sets the top 5 per cent of values to the value of the 95th percentile. Thus countries that are outliers in a data series are not discarded from the dataset (unlike with trimming), but they are limited in their extremity.
Second, a standard minimum-maximum normalization is applied to the winsorized data. This rescales values from their original range and units to a comparable 0–100 scale. The minimum value present in the data is assigned a value of zero, the maximum value is set to 100, and all values in between are proportionally revalued. Min-max normalization captures the best and worst values present in the data, not the theoretical minimum and maximum values attainable. Where necessary, min-max normalized values are inverted so that no matter what the underlying data series, the maximum value (100) always represents the best score present in the data series and the minimum value (0) represents the worst score.
Third, the quartile of each value in the data series is recorded. This provides a rapid intuitive understanding of how any one data point compares with any other, and of how skewness is affecting the data. For example, if there is positive skew (i.e. the mean value is greater than the median), min-max normalized values may be relatively low but still appear in the top quartile. This suggests that although their performance is noticeably inferior to the top performers in the dataset, they still compare relatively well with the majority of the other countries. A high min-max value but a low quartile value suggests the presence of a large number of countries with high min-max values. Thus, the quartile values are most useful for rapidly and crudely considering how a score compares with all other scores in the series, whereas the min-max value gives a better sense of the score relative to the best and worse scores (once extreme outliers have been accounted for). In the full version of the CH-FSD, for each data point we record three values, with the utility of each value depending on the current level and purpose of analysis. These values are:
- the raw value in the native units for the series;
- the winsorized max-min values, displayed on a four-colour gradient, with each colour representing a quarter of the range 0–100; and
- the quartile of each normalized value, displayed on a four-colour gradient, with each colour representing a quartile.
330 The CH-MAT was developed by Laura Wellesley and Felix Preston.
331 For example, Rodrigue, Comtois and Slack, B. (2017), The Geography of Transport Systems; Kaluza, P., Kölzsch, A., Gastner, M. T. and Blasius, B. (2010), ‘The complex network of global cargo ship movements’, Journal of the Royal Society Interface, doi: 10.1098/rsif.2009.0495 (accessed 8 Jun. 2017); UNCTAD (2016), Review of Maritime Transport 2016, New York and Geneva: UNCTAD, http://unctad.org/en/PublicationsLibrary/rmt2016_en.pdf (accessed 8 Jun. 2017).
332 For example, the UCL Energy Institute’s www.shipmap.org.
333 For example, Emmerson and Stevens (2012), Maritime Choke Points and the Global Energy System; Mitchell (2014), Asia’s Oil Supply: Risks and Pragmatic Remedies; Komiss, W. and Huntzinger, L. (2011), The Economic Implications of Disruptions to Maritime Oil Chokepoints, CNA, https://www.cna.org/cna_files/pdf/D0024669.A1.pdf (accessed 8 Jun. 2017); Nincic (2002), ‘Sea Lane Security and U.S. Maritime Trade: Chokepoints as Scarce Resources’.
334 The full list of single-country regions is as follows: Brazil, Canada, Mexico, Russia, Saudi Arabia, Turkey and the US.
335 Munro (2011), ‘Containers Move High-Value Exports – When do Containers Work Best For the Grain Buyer and the Grower?’; Lirn and Wang (2016), ‘The Determinants of Containerised Grain Shipping’ (accessed 8 Jun. 2017).
336 The CHRTD was developed by Jaakko Kooroshy. The development of ‘resourcetrade.earth’ was led by Felix Preston, Richard King and Siân Bradley.
337 UN Comtrade, http://comtrade.un.org/.
338 UNCTAD Stat, http://unctadstat.unctad.org/EN/.
339 UN FAOSTAT, http://www.fao.org/faostat/en/#data/TM.
340 USDA Foreign Agricultural Service, Global Agricultural Trade System (GATS), https://apps.fas.usda.gov/gats/.
342 IEA Statistics, https://www.iea.org/statistics/.
343 EIA International, https://www.eia.gov/beta/international/.
344 JODI Oil World Database, http://www.jodidata.org/oil/.
345 JODI Gas World Database, http://www.jodidata.org/gas/.
346 BP Statistical Review of World Energy, http://www.bp.com/en/global/corporate/energy-economics/statistical-review-of-world-energy.html.
347 UN Trade Statistics (undated), ‘What is UN Comtrade?’, https://unstats.un.org/unsd/tradekb/Knowledgebase/50075/What-is-UN-Comtrade.
348 For a full discussion of reporting asymmetries, see Markhonko, V. (2014), ‘Asymmetries in official international trade statistics and analysis of globalization’, discussion paper, International Conference on the Measurement of International Trade and Economic Globalization, Aguascalientes, Mexico, 29 September–1 October 2014, https://unstats.un.org/unsd/trade/events/2014/mexico/Asymmetries%20in%20official%20ITS%20and%20analysis%20of%20globalization%20-%20V%20Markhonko%20-%2018%20Sep%202014.pdf (accessed 8 Jun. 2017).