French Colonial Trade Patterns and European Settlements

Abstract

We construct a new database relying on various primary historical sources containing information on the value of French sectoral trade between 1880 and 1913 in order to assess the contemporaneous effects of colonial European settlements on French trade patterns. Our empirical results show that French colonies with more European settlements traded more with France. The impact is stronger with respect to the imports of raw materials and exports of manufactured goods from France to their colonies, suggesting that those territories were a source of resources for France and a market for its products. European settlements in colonies other than the French ones did not impact the trade of those colonies with France. We also explore to what extent the impact of European settlements on trade was exerted through the channel of the institutions brought by the settlers. Separating the part of European settlements associated with institutions, we find that the settlements associated with stronger institutions in French colonies had a positive impact on trade with those colonies. No significant impact was found for non-French colonies.

Appendices

Appendix 1: European Settlements Data

See Table 11.

Table 11 The European settlement as constructed from Huillery

Appendix 2: Disaggregation of the Trade Data

In order to verify the extent of possible distortions produced by our disaggregation procedure, we compare our disaggregated data to the bilateral trade data from Mitchener and Weidenmier (2008), denoted M&W hereafter, for the set of countries available in both datasets. We compute France’s trade with each trade partner as a share of total French trade for the years 1988 and 1912. As can be seen in Table 12, the import and export shares using the data from M&W have the same order of magnitude of the ones computed using our dataset.

Although the differences in the trade shares are small between the two datasets, there are large discrepancies in percentage terms for some countries. These discrepancies may come from two sources: (1) divergences in the original data (M&W use mainly British sources, while we use the French official registry); or (2) errors we make in our procedure of assigning trade values to countries when they are part of a group of countries in our original data.

We can identify the first source of discrepancy, that is, data source, when the country is separately identified in our original data source so that we were able to retrieve their data directly from our original source, with no manipulation. We do observe large differences between our data and that of M&W for a number of such countries. In 1888, that is the case for Australia, China, Mexico and Senegal. The set of countries in that situation is even larger for the year 1912: Brazil, China, Guadeloupe, Guyana, Indonesia, Martinique, Reunion, Saint Pierre et Miquelon, Senegal and Thailand.

Since we compare trade values as a share of total trade, it is possible that we find different shares for the countries discussed in the previous paragraph because our denominator (i.e., total trade) is ‘wrong’ in our data, due to our manipulation in the data for other countries. If the numerator is the same in both data sets and the problem is in the denominator, then the percentage difference of the shares should be the same for all such countries. That is not the case though. For example, for the year 1888 the trade share for Australia is lower in M&W’s data compared to ours, while the trade share for China is more than 10 times larger in M&W compared to our data. Hence, the discrepancies in trade shares between M&W and our data cannot be attributed only to differences in total trade (the denominator). We can then conclude that, for countries with separate data in the French official registry, the trade data from M&W is different from the data in the French official registry.

For the countries which were part of a group in our original data source, the discrepancy may come either from our disaggregation procedure or from differences between our original source and that of M&W. Unfortunately, it is not possible to disentangle the two possible sources of divergence in that case. Given the large discrepancies in the original data sources for some countries as discussed in the previous paragraphs, we cannot rule out differences in the original data as the source of discrepancies here as well.

As robustness checks, we have used two alternative methods for the disaggregation, population and arable land area, and both yield the same results in our regressions. Furthermore, our results also remain unchanged when we estimate our main regression excluding the countries with very different data in our dataset compared to M&W, as presented in “Robustness Checks” section (Table 13).

Table 12 Data comparison

Table 13 Disaggregation of regions into countries