The database is complex. Here is a discussion about double counting and partners that might help you for some issues.
Question :
At the moment, I am extracting sugar exports from France to Italy from the Toflit18. Data include exports to Venice, the kingdom of Naples and Italy. I used the classification system/source names to figure out what the term “Italy” means and which regions are included. I found
If so, then sugar exports from France to Italy also include exports to Venice and the kingdom of Naples. As exports to Venice and the kingdom of Naples are also listed separately, I am wondering whether double counting is occuring for several years. For example, we have exports to Italy and Venice for the years 1764-1779.
Or, have I thought incorrectly about that? I would be grateful for any advise.
Answer:
Depending on the year and the source, “Italy” does not mean the same thing. Basically, it means “everything else in Italie”. See (for the “National best guess partners x product” which I think interests you:
The French Revolution of 1789 was preceded by a crop failure crisis that played a decisive role in triggering the revolution (Labrousse, 1944). Although this crisis was one of the most severe of the French Old Regime, its demographic repercussions were very limited compared with the impacts of the 1693-94 and 1709-10 crises, which claimed approximately two million lives each, but did not have such radical political repercussions. Alexis de Tocqueville pointed in his day to a “paradox of growing dissatisfaction”, arguing that the political impacts of the difficulties of 1789 were greater because they were preceded by a prosperous period (Daudin, 2005).
However, this prosperity does not show up in the indices of real wages, which stagnated in the decades prior to the French Revolution (Ridolfi, 2019). Is this the right indicator? Real wages provide information on the situation of the urban working class, who spent most of their budget on food. They say nothing about the effect of industry- and foreign trade- related economic change on the urban middle classes or the peasant classes. This article contributes to research into other signs of economic change. It examines international trade price trends, since international trade was the most buoyant part of the French economy in the 18th century. It hence contributes to the literature on the economic causes of the French Revolution (Weir, 1991). Following a presentation of the database and the construction of our indices, we explore market integration trends, variations in terms of trade, and observed structural price differences.
Alexandre Aubourg, Loïc Charles and Guilllaume Daudin «La collecte du chiffre commercial au XVIIIe siècle» *Cahiers d’histoire des Douanes et des Droits Indirects*, n°59, 2nd semestre 2016
Coming back to TOFLIT18 TOFLIT18 aims at opening new alley of research on the French economy on the eve of the Industrial Revolution by studying data on more than 500,000 French external trade flows extracted from the archives of the Bureau de la Balance du Commerce (L. Charles et Daudin 2011). We have already discussed our objectives in Dialogue #43 in July 2016. Three years after, we want to provide here a brief summary of how it evolved and where it stands now. More details can be found at https://toflit18.hypotheses.org. I believe our experience can be useful to other people starting similar endeavours.
Analyse historique du commerce extérieur français – le projet TOFLIT 18. Auteurs : Corentin Ponton, Quentin Vidal et Guillaume Daudin
L’analyse des racines historiques du développement constitue un axe important du programme de travail de l’équipe DIAL. Les chercheur-se-s affrontent plutôt la poussière des archives que celle du terrain (notamment celui du Sahel) lors de la collecte de données d’enquête. Certain(e)s bravent même les deux poussières à la fois, mais ils (et elles) ne sont pas l’objet de ce focus. Le thème qui est ici présenté est celui du commerce international de la France au tournant de l’industrialisation de l’Europe.
Bienvenue à Paul Chagnaud, qui fait un stage encadré par Guillaume Daudin :
“Actuellement étudiant en classe préparatoire D2 au lycée Turgot ainsi qu’en L3 d’économie appliquée à l’Université Paris Dauphine-PSL, j’ai intégré le projet TOFLIT18 le 29 mai 2021 en tant qu’assistant de recherche dans le cadre d’un stage d’une durée d’un mois.
J’ai choisi l’option Histoire des faits économiques et sociaux en classes préparatoires au concours D2 de l’ENS Paris-Saclay. J’ai eu à cœur de découvrir cette discipline lors de mes trois premières années d’études supérieures. Son enseignement m’a permis de comprendre plus en profondeur les mécanismes économiques et de mieux analyser l’actualité.
Lors de mon stage, j’ai retranscrit des sources relatant le commerce de la France avec l’Outre-Mer autour de 1789. De plus, j’ai élaboré un processus automatisé permettant d’explorer les différences dans les chiffres sur le commerce de 1789 donnés par différentes sources (le « Résumé » se trouvant aux Archives Nationales à la cote F12 251 et des sources dispersée, souvent plus précises, aux Archives Nationales aux cotes F12 1666 et 1667 et aux Archives Nationales d’Outre-Mer à la cote Col F 2B 14. Le résultat de ce travail est ici. Ce travail m’a permis de prendre conscience de la nécessité de fiabiliser les outils et méthodes d’analyse des données.
Ce stage constitue pour moi une première expérience dans le milieu de la recherche académique. J’y vois l’opportunité de découvrir le cadre d’un laboratoire de recherche, en l’occurrence le LEDa, de travailler en équipe sur plusieurs sujets et d’appréhender davantage le métier d’enseignant-chercheur. Les missions qui ont pu m’être confiées m’ont permis d’enrichir ma culture économique et d’apprendre un nouveau langage de traitement de données (STATA). Cette première expérience me conforte dans ma volonté d’approfondir mes connaissances en Histoire et de poursuivre mes études en suivant une formation axée sur la recherche en sciences économiques et sociales.”
The retranscribed variable is “quantity_unit” (1,564 different values).
Orthographic Normalization
The file “classification_quantityunit_orthographic.csv” establishes a correspondance between the variable “quantity_unit” and the variable “quantity_unit_orthographic” (701 values). 235,000 observations have “livres” in “quantity_unit_ortho” and 164,000 have “unité manquante” (out of 555,000).
Simplification
The file “classification_quantityunit_simplification.csv” establishes a correspondance between the variable “quantity_unit_orthographic” and “quantity_unit_simplification” & “conv_orthographic_to_simplification”. “quantity_unit_simplification” is the name of the simplified quantity unit. “conv_orthographic_to_simplification” is the conversion factor between quantity_unit_orthographic and quantity_unit_simplification. E.g. for “douzaine”, these are «nombre» and “12”
First pass for unambiguous “classification_quantityunit_metric1”
The file “classification_quantityunit_metric1” establishes a link between “quantity_unit_simplification” and “quantity_unit_metric” + “conv_simplification_to_metric” when possible directly. If not, we flag “needs_more_details” which indicate that we cannot know the metric equivalent without using more variables.
Simplification and Multi-key conversion
The file “classification_quantityunit_metric2” establishes a correspondance between : quantity_unit_simplification + product_simplification + product_revolutionempire + exportsimports + customs_region + partner_grouping with “quantity_unit_metric” + “conv_simplification_to_metric” In the end, 277,000 trade flows are expressed in kilograms, 164,000 have no quantity unit, 70,000 are expressed in numbers, 13,000 in liters, 10,000 in centimeters, 3000 in containers and 2,200 in sheets. 15,000 have not “quantity_metric” esquivalent.
Bienvenue à Lucas (le prénom a été changé), qui vient faire un Master que Guillaume Daudin va encadrer.
“Actuellement étudiant dans le Master d’Analyse Economique Quantitative, à l’Université Paris-Dauphine, au sein de la filière “Politique publique”, je prépare un mémoire sur l’existence d’un phénomène de maladie hollandaise dans l’économie française du XVIIIe siècle, du fait de l’importance du commerce avec les Antilles.
Passionné d’histoire économique, j’avais choisi une licence d’économie-histoire à l’université Paris-1, avant de basculer sur l’apprentissage de techniques quantitatives dans ce master. La base de données TOFLIT représente pour moi une excellente opportunité en terme de travail empirique quantitatif.
Par ce travail, j’espère mieux comprendre un des aspects de la trajectoire économique de la France moderne, ayant pu influer sur sa non-Révolution Industrielle.
Par ailleurs, je désire me former à l’analyse générale des phénomènes de maladie hollandaise.”
“Drap” and “Draperies” mainly refer to woolen cloths. However, in the Balance du Commerce source, one can find some occurence of “draps de lin” or “draps de soie”. As a consequence, up to now, we were agnostic as to the raw material of “draps” when there was no qualifier. In the SITC classification, we have put them in “Other threads and fabrics” (6i) instead of “Woolen threads and fabrics” (6e). In the “Revolution and Empire” classification, “Draps divers” were distinguished from “Étoffes de Laine”.
While listening to Jean-Marc Minovez’s seminar on londrins on December 9th, I realized that something was wrong with our woolen export data. Why were the national figures going to zero (1757, 1778, 1782-1813) on some years? Especially when nothing seemd wrong from Marseille’s point of view?
In red, the value of exports of woolens from Marseille to the Levant. In black, the value of exports from the whole of France to the Levant (purple are exports from Montpellier, BTW).
I then looked at the national level at the relationship between national exports of “Other threads and fabrics” (6i) and national exports of “Woolen threads and fabrics” (6e). Looking at 1756, 1782-1789 and 1811-1812 suggested something wrong was going on. Increases of “other threads and fabrics” are the symmetric of decreases of “Woolen threads and fabrics” and vis-versa.
in black, national exports of “Other threads and fabrics”, in red, national exports of “Woolen threads and fabrics”.
Looking at Savary and the 4th edition of the dictionnaire de l’Académie it is very clear that, when there is no qualifier, “drap” designates a woolen cloth. So I have decided to change the SITC and revolution and empire classifications to reflect that. The time series became much more convincing.
in red, national exports of “Other threads and fabrics”, in black, national exports of “Woolen threads and fabrics”.In purple, the value of exports of woolens from Marseille to the Levant. In red, the value of exports from the whole of France to the Levant.
There were still two puzzles. What about 1778 ? How come French exports of woolens to the Levant are nil whereas Marseille’s are not? It turned out that the “Objet Général” mislabeled the destination of Marseille’s londrins to “Flandres” (which would have been a rather export destination). That was corrected in the Objet Général, using our new absurd and guess flagging feature. On the other hand, the Objet Général used a price (6 lt per pound) that is much more plausible than the Marseille price for this year (4 lt per pound) (You can look here at the prices of londrins in the database.) So the Marseille data were corrected as well.
The second puzzle is 1808 and the big increase of exports of “Other threads and fabrics”. It turns out that comes from an 11 million l.t. export of “Étoffes diverses” to Germany. So that seems to be legitimate. Of course, part of them could be “Étoffes de laines”, but that item is also explicitely present. So there does not seem to be anything we should do about it.
You can go and see the latest versions of the graphs here for total exports and here for exports to the Levant.
Thanks to further funding provided by DIAL we were able to hire again the great developers at OuestWare to update the datascape. We are very excited by the changes they made and we urge you to go and check them out !
The main change is the creation of a new visualization that allows to view individual trade flows. Less important changes are below the line.
More “best guess” fake sources. It is difficult to keep track of all the sources in the data and how they interact. So, instead of a source, you can select “best guesses” that automatically selects the source that for a year corresponds to what you want. There are currently five “best guesses”.
Best Guess national product x partner autoselects the best source about trade by product x partner for the whole of France in each year.
Best Guess national product x partner autoselects the best source about trade by product x partner for the whole of France in each year.
Best Guess national product autoselects the best source about trade by product for the whole of France in each year.
Best Guess national tax department autoselects the best source about trade by tax department for the whole of France.
Best Guess tax department product x partner autoselects the best source about trade by product x partner x tax department each year for individual tax departement (not for the whole of France).
Change in vocabulary. Some the of the vocabulary was confusing (like “Directions”, “Country”, “Type”…). This has been replaced by clearer terminology (“Tax department”, “Partner”, “Export or import”…)
Better treatment of flows with a single value for multiple quantities
Better treatment of values and quantities that are identified as coming from the sources but which are errors nonetheless. They are replaced in most viz by guesses by the team. You can find them in the trade flow viz (look for the variables “absurd_values” and “absurd_quantities”.)
Thanks to the funding provided by DIAL we were able to hire the great developers at OuestWare to update the datascape. We are very excited by the changes they made and we urge you to go and check them out !
The two main improvements are optimization and permalinks. And you can still use regular expressions. More details below the line.
Loïc Charles et Guillaume Daudin ont présenté un travail sur “Haut de gamme, excédent commercial et dynamique des exportationsdans la France pré-Industrielle” au colloque “Vers le haut de gamme Made in France” le 22 novembre 2019.
Qu’avons-nous fait ?
Une base riche… et problématique
Ce qu’on sait du rôle du luxe dans la politique économique
Comment isoler le luxe dans les flux commerciaux au XVIIIe siècle?
Le rôle du luxe dans les exportations françaises
Alors ?
Il est possible de tirer des choses des données de la Balance du Commerce
La recherche de la qualité était à la base de la politique économique française
Le commerce de soierie (et donc haut de gamme) jouait un rôle croissant dans l’excédent commercial français – en partie à cause de la perte des marchés d’Outre-Mers
Par contre, il n’est pas facile d’établir un rôle spécifique pour la qualité relative dans le travail d’un matériau donné, à part le déclin (mais à partir d’un niveau bas) dans les exportations nationales
Et il faudrait regarder d’autres produits : bonneterie, gants, chapeaux, vins…
We welcome Stephen Jackson in the team. He will stay for June and July 2019, and will work in the beginning on a new 200 category classification of merchandises based on the documents from the Revolution and the Empire. A short self-presentation follows.
After spending three years at the University of Warwick studying economics at undergraduate level, I decided to move back to France and follow up my bachelor’s degree with a master’s in development economics at Panthéon-Sorbonne University (Paris I).
The research-oriented nature of postgraduate study makes experiences in the field of academia all the more valuable. Joining the project as an intern is an opportunity to further explore an area of economics that has always interested me while experiencing first-hand what it takes to be engaged in research at the forefront of the discipline.
I’ll be joining the TOFLIT18 project for two months over the summer of 2019. My tasks as part of the team include elaborating a new classification of products based on earlier simplification work as well as harmonisation of data in the early years of Revolutionary France to address potential shifts in classification methodology and reporting.
Although economic history has always featured prominently in my choice of courses at university, this internship also represents a step into the unknown. Working with such a large collection of data and facing the challenges tied to historical records is something new and fascinating for me. In my short time on board, I hope I can contribute to the large body of work that has been carried out so far and pave the way for future use of the dataset.
Paul Girard presented the project on February 2nd at FOSDEM 19 in Brussels.
The presentation was a great success, and you can find the slides and a video recording here : https://fosdem.org/2019/schedule/event/graph_french_trade_study/
Transformations of the french economy through the lens of international trade, 1716-1821
