2017-2022, Towards a new version of RICardo

The first version of the RICardo database and RICardo website was officially released in December 2017. Over the last four years, there has been sustained activity around the project, especially from March 2019 when a new grant was awarded by the Fondation Del Ducca covering the period 2019-2022. It provided the financial support needed to allow for the achievement of three objectives discussed by the RICardo committee in May 2017:1

  • update the database with new primary sources;
  • add a political dimension to trade data;
  • improve the website with new exploratory analysis features.

 

An updated RICardo database: from the collection to the edition of new data

Four years ago, one of our main objectives was to complement the RICardo database by collecting new primary sources that proved to be more comprehensive than the Annales du commerce extérieur, the British statistical abstracts or the Memoranda of the League of Nations in that they provide more detailed tables of a reporting’s trade partners (Dedinger & Girard 2017).

Great support came in 2018 when Raymond Hicks offered to include in RICardo the database he had built with Joanne Gowa. Gowa and Hicks database (Gowa & Hicks 2013; Gowa & Hicks 2015; Gowa & Hicks 2018) covers the years 1900 to 1949 and gathers bilateral trade statistics for 110 reporting countries. Hicks benefited from access to the Princeton library which possesses a vast collection of historical national trade publications (primary sources). Furthermore, we consulted all the national trade archives available at the Bibliothèque nationale de France. The search for and digitization of these sources occupied most of the second semester of 2019. A third channel to find new primary sources was the downloading of digitized archives increasingly available on the Internet.

December 2019 to June 2020 was devoted to the transcription of data in Excel/CSV files with the precious help of Ariane Treguer, student in the Master of History at Sciences Po. To this day, more than 200,000 bilateral trade data were transcribed and will be added to the RICardo database (the RICardo.2017 version includes about 300,000 data).2

The next step was to conceive an efficient way of editing the new data directly in the structured database linked to the versioning system. Mengying Du, research engineer at Sciences Po médialab, created an edition tool that dramatically facilitates the integration of new data files in the data repository. The prototype she elaborated is a JavaScript React application hosted on Github. As illustrated by the figures below, this edition tool makes it very easily to add new files of trade data to the repository, making sure that the format specified by the datapackage is fully respected. Once logged in with a Github account, the user downloads a new data file. A user interface then reports errors found in the file. There are two types of errors: format errors (date format or enums) and foreign key errors (a linked value is missing). Each error has to be fixed by completing, through a dedicated form, the ‘sources’, ‘entity_names’ and ‘RICentities’ tables, and/or the ‘currencies’ and ‘exchange_ rates’ tables.3 The validated file is thereafter added to the data repository (through a git commit and push) that is automatically updated with the new rows created in the ‘sources’, entity_names, RICentities, ‘currencies’, or ‘exchange_rates’ tables.

Example: editing of the file ‘Greece bilateral trade data, 1890-1928’

  1. Review of errors in the file

Errors summary

  1. Source not found in the ‘sources’ table; creation of a new row

  1. Original name not found in the ‘enty_names’ table; creation of a new row

 

  1. Exchange rate not found in the ‘exchange_rates’ table; creation of a new row

We are currently working on the edition of the 200,000 new trade flows. The publication of the updated version of the RICardo database is scheduled for 2022.

 

 How RICardo led to GeoPolHist, a new dataset of the geopolitical entities of the world

One of the main challenges raised by the RICardo database is the complexity of its network of trading entities. As outlined at the 2017 RICardo workshop, we thought that this problem of heterogeneity needed to be clarified before implementing new ideas such as a political codification for the entities.4 However, at this time, our major concern was to exploit the database in order to provide an overview of the geographical distribution of world trade over the 19th-mid20th centuries, to know who trades with whom. We considered different options to allow for the reconstruction of a matrix of world trade flows as complete as possible. A peculiarity of the RICardo database is that the reporting and partner entities are not simply ‘countries’, they can be a ‘city/part of’, a ‘colonial area’, a ‘geographical area’, or a ‘group’. Our first idea was to disaggregate the trade flows of the heterogeneous entities by converting entities of the type ‘city/part of’, ‘colonial area’ and ‘group’ into the ‘country’ type. At the same time, we were working on the geo visualization of trade flows with such applications as Gephi or Arabesque. Advances in our research were presented at different seminars.5

Working on the reconstruction of a complete world trade matrix raised the issue of the completeness of the RICardo dataset. The list of RICentities of the type ‘country’ was elaborated by using two COW lists of geopolitical units.6 New RICnames were created for the ‘city/part of’, ‘colonial area’ and ‘geographical area. In the 2017 version of the ‘RICentities’ table, there are 366 ‘country’ entities. What part of the totality of world countries do these 366 entities represent? How many countries are there in the world since the beginning of the 19th century? We had these questions in mind when the Covid crisis began. One of the sided-effects of this crisis was to provide us with a lot of time to dig deeper into these issues. We gradually focused on the second one, ‘how many countries in the world.’ As developed in Dedinger & Girard 2021, a close examination of the COW lists led us to find some flaws: the same unit can have different COW names, the same COW name can have different COW codes, or different COW names can have the same COW code. Furthermore, a graphical representation of the COW lists of entities over time showed that they were more comprehensive from the 1880s as can be seen in the growth of the black curve (all of the geopolitical units identified by COW over 1816-1993/2016).

During 2020-21, we have been working on the COW lists to create a dataset and visual documentation of the political status of the entities of the world from 1816 to the present. This led to the GeoPolHist database that shows remarkable linearity in the total number of the world’s entities during the last two centuries, the figure oscillating around an annual average of 1000 GPH entities.

Each entity is codified over the entire period by using ten different political statuses. The complete list of geopolitical entities and the definition of the political statuses are available on the GeoPolHist website that facilitates the visual exploration of the political history of the world’s entities.

Each geopolitical entity has a dedicated web page showing two timelines. The first timeline depicts the political status of the selected geopolitical entity over the whole period and the second timeline identifies the geopolitical entities that were under the control of the selected entity.

The elaboration of GeoPolHist, therefore, helped us to better define what is a ‘country’ and what are the geopolitical entities of the world over time. A first consequence for RICardo is that the term ‘GPH_entity’ is now used preferably to the term ‘country.’ Second, GeoPolHist solves the issue of the political (and also geographical) codification of the RICentities of the ‘GPH_entity’ type. It must be noted that the updating of the RICardo database can also help to update the list of the geopolitical entities. For example, Thursday Island, an island part of the Torres Straits islands, was added as a trade partner into the list of RICentities. This led to the creation of the new GPH entity ‘Torres Straits Is.’, part of Queensland since 1872.

 

New visualizations on the RICardo website

 In 2017, another objective was to improve the RICardo website. We aimed to create new features to allow for the visualization of aggregated entities, for the selection of partner entities (not just reporting), for the exploitation of the exchange rate database, and lastly for the inclusion of annotations giving information on structural breaks in the series due to historical events. The mission was entrusted to the OuestWare enterprise that Paul joined in 2020. Time and cost constraints eventually decided us to focus on three new features: a visualization of the exchange rates database; the inclusion of a political timeline that indicates, for a trading entity, the number of its political dependencies over time; the creation of a ‘Partner’ page that allows for the visualization of the trade of a selected partner by using bilateral trade data of its reporting entities. Details about the work done on the website during the summer 2020 can be found on OuestWare’s website. Here are examples of the new visualizations available on the RICardo website:

The new visualization of the exchange rate database

 Trade of Partner ‘Germany (Zollverein)’ before 1871 can be visualized from

bilateral trade statistics of its ‘reportings’

 

We continue our long journey through historical trade data. Visiting archives, collecting data, transcribing into a database, visualizing and discussing results are each a step that contributes to a better understanding of 19th century international trade. We hope that our work will help scholars and teachers to discuss both historical economy and long run quantitative analysis methodological issues.

 

References

Dedinger, B., & P. Girard (2021). How many countries in the world? The geopolitical entities of the world and their political status from 1816 to the present. Historical Methods: A Journal of Quantitative and Interdisciplinary History. doi:10.1080/01615440.2021.1939826

Dedinger, B., & P. Girard (2017). Exploring trade globalization in the long run: The RICardo project. Historical Methods: A Journal of Quantitative and Interdisciplinary History, 50:1, 30-48. doi: 10.1080/01615440.2016.1220269

Gowa, J., & R. Hicks (2013). Politics, Institutions, and Trade: Lessons of the Interwar Era. International Organization, 67, pp 439-467. doi:10.1017/S0020818313000118

Gowa, J., & R. Hicks (2015), Commerce and Conflict: New Data about the Great War. British Journal of Political Science, 47, 653–674. doi:10.1017/S0007123415000289

Gowa, J., & Hicks, R. (2018). “Big” Treaties, Small Effects: The RTAA Agreements. World Politics, 70(2), 165-193. doi:10.1017/S0043887117000302

  1. See ‘Future prospects’ in RIC_Workshop_2017_report []
  2. The complete RICardo dataset gathers all collected data since the beginning of the project and is freely available on a github data repository. The file provided on the RICardo website is an extraction of the complete data set. It eliminates duplicates and favors primary sources. The current available version is RICardo.2017. The release of the new version is scheduled for 2022 []
  3. (Dedinger & Girard 2017) describes the RICardo database relational schema []
  4. RIC_Workshop_2017_report, p. 10: “Entities are the basic information to analyze and to compare the distribution of trade. There are ‘reporting’ and ‘partner’ entities, the list of partner entities being much more longer and heterogeneous. A great part of the project has been devoted to the construction of a RICardo entity table in order to standardize the names of the entities and to establish a system of codification of these entities. Much more could be carried out, such as the creation of aggregated entities or the introduction of a political codification for the entities. But this raises problems due to the non-homogeneity of entities that need to be clarified” []
  5. Lund Seminar of the Department of Economic History, Lund University, 10 October 2018 RIC_WTW_Lund_2018; Workshop of the Center for History and Economics – Paris (CHE-P), Sciences Po Paris, 5 June 2019; International Conference on Spatial Analysis and Geomatics, Clermont Ferrand, 13-15 November 2019 ; TransNum Seminar, médialab SciencesPo, 25 September 2020 []
  6. The two files available on the COW website are states2016.csv , which identifies the states of the world since 1816  and Entities.pdf which gathers all colonies and dependencies in the world from 1816 to 1993 []

Laisser un commentaire

Votre adresse e-mail ne sera pas publiée. Les champs obligatoires sont indiqués avec *

Ce site utilise Akismet pour réduire les indésirables. En savoir plus sur comment les données de vos commentaires sont utilisées.