June 5th meeting
Attending & program:
Alain Becchia, Loïc Charles, Federico D’Onofrio, Guillaume Daudin, Béatrice Dedinger, Pierre Gervais, Paul Girard, Jérôme Héricourt, Reka Juhasz, Emmanuelle Lavallée, Ivan Ledezma, Silvia Marzagalli, Cyril Mouton, Guillaume Plique, Pierrick Pourchasse, Pierre Hollegien.
After having presented the various participants, the meeting took place in accordance with the following agenda:
9h30-10h30: Alain Becchia, “Imports and exports of woolens in the balance of trade registers”
10h30-11h30: Pierre Hollegien, “Democratization of medicinal drugs in France (1700-1850)”
11h30-12h30: Béatrice Dedinger & Paul Girard, “Ricardo: presentation of a preliminary version of the database”
14h-15h: Guillaume Daudin & Loïc Charles: “Coding standards and merchandises, units and measures classifications in the TOFLIT 18 database, and other associated problems”
15h-16h: Paul Girard & Guillaume Plique: “Thoughts on the data template and associated tools”
16h15-17h15: Open discussion on next steps.
On the woolen trade (A. Becchia)
The work is specific to cloth trade from Elbeuf because of its importance. The data Alain Becchia has been working on are from the trade balance, localized in Rouen or elsewhere.
First Alain Becchia retraced the construction of his database. He highlighted a few important points that can alter the quality of data:
- Employees from the “Direction” determine the designation of commodities. Depending on the knowledge of each employee designation of a same commodity might vary across time and space. To correct those mistakes aggregation is sometime necessary.
- Prices and values can be easily retrieved. On the other hand, quantities are sometimes in different units or can be estimated based on value. It is interesting to evaluate whether prices fluctuate over time or if employees are referring to a standard price over the years.
On the importance of textile trade
50% of woolen exports are from Rouen in the 18th century. Textile is a driving force of the regional economy as exportations in value accounted for about 800 worker’s annual wages.
It is specifically interesting when studying the effect of exogenous shocks on trade: from diseases decimating Spanish herds and increasing wool prices, to trade treaties with England. Especially the 1786 treaty with England had important repercussions on French manufacturers. Although drapers thought they could resist the competition with English manufacturers, merchandises sent to England remained unsold thus resulting in huge losses for the French.
Exports to colonies through other countries such as Spain can also complicate the visualization of trade.
Finally an important characteristic of this trade is its role in providing species. It is in the interest of local producers to equilibrate financial flows: France exports woolens to Italy partly to pay for raw silk imports.. Trade works thanks to a compensation system in order to ensure sufficient liquidity for merchants.
Comparing with other sources
The balance of trade data can be compared to the enquiry in 1765-1766 by John Holker.
On the democratization of medicinal products (P. Hollegien)
Three main themes were discussed: methodological issues, descriptive analysis and some proofs of the democratization of the usage of medicinal products.
How to classify products?
Considering the wide diversity of remedies, identifying medicinal products is a challenge. Some are specifically dedicated to medicine (mostly products from purgative medicine) while others can be used for other purposes and can hardly be representative of the evolution of the medical habits. To tackle this problem, the classification is based on the following principle: products are classified as medicinal products if they have as primary usage or are exclusively used for medicine purposes.
Another problem to be dealt with is the variation of names and measures across time and sources as the TOFLIT project gathers multiple sources. However after a comparison between sources (1788), the retained classification seems to be proper enough to compare different sources.
Although there was a mistake with the specification of the weight in the regression, the descriptive analysis enables us to conclude that the value of medicinal products traded over the period 1718-1839 has increased more than the population growth, especially for the period after 1771 where imports per capita increased at an annual rate of 1.5%. It also highlighted that the main products imported are those from America and the Levant.
Finally it raised the problem of import flows that are re-exported. This is an important limit in the current presentation.
In this part regressions on prices and quantities supported the initial hypothesis. Indeed while prices are going down (which is even more obvious if we consider relative prices), imported and exported quantities are going up. In terms of nominal values it seems to indicate that there are enough quantities for the entire population to benefit from these products.
However this analysis is balanced by the classification weaknesses. Are we witnessing the rise of medicines usage or is it due to a transfer from polypharmacy to purgative medicine (which is better captured in our data)? Other concerns were raised on the Napoléon era as duty borders evolved and duties changed.
On the project Ricardo (P. Girard & B. Dedinger)
The presentation started with a summary of the objectives of the Ricardo projects and the website visualization. Today the project has reached 267,000 inputs on 4,500 locations and 1,459 aggregated entities. From that point a discussion started on what kind of aggregated data should be available via the website.
The key concern was how to display data with a friendly visualization when countries, denominations and norms evolved across time. What can be gathered and what cannot. A solution is to differentiate two categories: rough data and refined data.
The main conclusions were to either open the website and wait for feedbacks, or define a specific target publics for the website (e.g. school teachers or academics) and modify data according to what these publics are looking for.
On coding standards and classifications (G. Daudin & L. Charles)
As noted in previous presentation, data from old registers and manuscripts are subject to several difficulties:
- Different sources not standardized (different name for a same commodity, etc…)
- Different norms: norms are established by the administration but practices among clerks vary depending on the aggregation level (Local or national) and their personal involvement.
- Transcription errors
Thus how should we gather merchandises, geographic areas and unit measures?
As for Ricardo, the debate was articulated around the level of correction / simplification / aggregation we should perform on the data base.
The idea of any kind of classifications (SITC for example) has been widely criticized and the rule we decided to apply is the following: we should keep one column where we would perform orthographic corrections only (Semicolons will be kept and we will use the singular for all merchandises). Classification aspects will be dealt with in the next section.
The initial presentation is available here :
On the data template (P. Girard & G. Plique)
Paul and Guillaume proposed a system based on 6 facets:
- Measurement units
The idea is to be able to explore every flow through these 6 facets (flows from specific countries, merchandises, sources… through the spectrum of the other facets). Users could choose what they want to visualize and then download the data.
On a different tab, the user could select its own classification. To deal with the problem of aggregated data, each user could create a classification according to what he needs (or use classifications from other users). From the first classification (the orthographic one), a user can add new layers of classification and aggregate more data. If one user believes one layer is wrong, he can simply create a new one based on a less aggregated layer.
This is illustrated in this (blurred) picture
The idea is thus to move from free tags to classified tags. The user could use a tool working like Open Refine so he can navigate through classification easily and find phonetic or morphologic resemblances.
- How can we treat metrics as they depend on multiple facets ? Paul wants to convert the database into the metric system before so the user has not to choose.
- A problem remains for observation (mostly countries) that can be normalized as A for a specific point in time and B for another (for example annexed regions).
- We must think on how to add a user friendly .pdf file to link products to their definitions.
- Finally it must be possible in the first tab to compare two different ways of aggregating data.
Economists need to make a list of information / tools they want.
- Guillaume Plique starts working on the datascape in July
- All the available data should be in the box at the end of August.
- Pierre Gervais is in charge of the orthographic corrections from merchandises and thus creating the first layer of aggregations (only orthographic corrections).
- 01/2016: Guillaume Plique will stop working with us by the end of January, thus the datascape must be ready.
- 01/2016: Big internal meeting to evaluate the datascape.
- 03-06/2016: Workshop in Chicago to showcase the work realized (it would be great if some economists had papers to present)
- End 2016: Open meeting on databases (with O’Rourke, David Jacks, Nikolaus Wolf, Giovanni Federico, etc…)
- ??? : Compare data available on the Navigocorpus project
- Set an economist meeting to discuss on tools needed.
Soon, Guillaume D. will upload new database on the Live Drive so economists can have a look at the latest version available. The credentials to connect to the LiveDrive are available from Guillaume
Finally we need to upload an Excel file with data available according to their sources.
Next meeting is July 2nd at 10am at the Medialab.