The Impact of a ‘Colonizing River’: Colonial Railways and the Indigenous Population in French Algeria at the turn of the Century

ABSTRACT

Colonial railways eased settlement and altered the economic activity of the surrounding areas. Thus, they provide a good testing ground for the impact of settlement expansion. By taking advantage of unique territorial population data and digitized historical colonization maps in the Constantine region, this paper assesses the effect of railways on the indigenous population in Algeria during the colonial years. The indigenous population growth and density are first analysed in a cross-section multivariate regression framework that permits controlling for various forms of settlement. As a robustness check to the results, the paper implements differences-in-differences combined with a propensity score matching methodology that allow analysing the impact in relatively isolated areas where the infrastructure arrived later. The main conclusion of the paper is that, if settlement did have a positive effect on the indigenous population growth – as many historians tend to argue – it was channelled through railways only after 1900, when cereal cultivation improved, and the trade policy changed. The lack of significance before the 1900s is most likely explained by geographic-specific factors that limited the potential effects of railways.

Keywords:

• colonial railways
• French Algeria
• transportation
• population density

Acknowledgements

The author thanks Markus Lampe, Joan Rosés, James Simpson, Alfonso Herranz-Loncán, Giovanni Federico, Ewout Frankema, Joerg Baten, Alexandra López-Cermeño and the participants at the Economic History Seminar in Vienna University of Economics and Business (May 2016), the Economic History Workshop at the Carlos III University (January 2017), the seminar at the University of Tübingen (January 2018), the Economic History Society Annual Conference (May 2018) and the World Economic History Congress (August 2018).

Disclosure statement

No potential conflict of interest was reported by the author.

Notes

2 As Good (1961) says, after the 1870s the indigenous population experienced a continuous growth explained by its natural increase. Still, there is no consensus on the real census figures for the period before 1900 and the estimated annual growth rates range between 3 and 1%. For more detail on population trends see e.g. Kateb 2001; Good 1961; Biraben 1969.

3 This section is based largely on historical material provided by Bernard Venis and Francis Rambert on the site “Alger Roi”, http://www.alger-roi.fr/, accessed 15 January 2019.

4 According to Belkacemi (1984), the total Algerian colonial investment in the infrastructure amounted to 633 million francs, without including the 367 million interest paid to shareholders. It has been shown that railways represented the largest expenditure within colonial budgets in Africa and had long-lasting effects even after independence (Jedwab & Moradi, 2016).

5 This quote has been extracted from Belkacemi (1984: 322–323). The original source is Burdeau 1892.

6 Belkacemi (1984: 343) provides evidence of cases where irrigation was used as an instrument to improve fertility and ensure the success of the railway infrastructure.

7 The value for South Africa was obtained from Herranz-Loncán & Fourie (2018: 11), changed into ton-km, and divided by the population value from Mitchell (1994). For Algeria, the freight data was obtained from Mitchell (1994) and the population data was collected from various Annuaires Statistiques de la France and the 1948 Algerian census. The missing observations were adjusted based on vital rates. For Uruguay and Argentina, the values were obtained from Herranz-Loncán 2011: 25.

8 Retrieved from the World Bank Development Indicators, https://data.worldbank.org/indicator/, accessed 15 January 2019.

9  For example, the Statistique Générale de l’Algérie (1911) shows that viticulture occupied approximately 14% of the settler-cultivated area.

10 The line concession and financing system were also blamed for the high tariffs since they did not provide the companies with the right incentives to decrease rates. For instance, the high rates were partly explained by the lack of route competition between the railway companies (Belkacemi 1984: 285).

11 Excluding the Territoires du Sud; see Statistique Général de l'Algérie (1911).

12 The 1884 volume reflects the population situation on the 30th of September; the 1892 and 1897 volumes refer to the 1st of January and the 1902 reports the population; and the 1902 volume refers to the 1st of November.

13 The applied census technique changed according to the surveyed population category. The populations in the civil territory and those located in the settlement centres in the military territory completed a family questionnaire, while the populations in the military territory were inferred by counting the number of tents (assuming each tent hosted five to seven people).

14 The douars were municipally self-governing territorial units to which the sénatus-consulte had been applied. The sénatus-consulte was a law passed in 1863 that aimed to disintegrate tribal areas and divide them into territorial units known as douars. It delimited and registered indigenous properties and provided legal land titles in accordance to French law (Bellahsene 2006: 169).

15 The prolongation after the 1900s to Khenchela is not included in this study.

16 Within the line Bone–Guelma, the track Bône–Duvivier opened to transit in 1876 while Duvivier–Guelma opened in 1877.

17 The information on the presence of settler infrastructure was obtained from the Dictionnaire des Communes, Villes & Villages de l’Algérie published by Fillias (1878).

18 The author quotes Auguste Burdeau in the Chambre d’Agriculture in 1891.

19 There was also a strong military component in the construction of railways. For instance, Belkacemi (1984: 321) notes the following statement from the Courrier d’Oran, 19 October 1881: ‘[France] should construct railways in the territories of hostile (or suspect) tribes and make the stations military strong point and fortified blockhouses from which to carry out surveillance and control of the surrounding areas.’

20 The number of regions (n=170) with less than 30 settlers in 1884 had on average a density of 0.0008 settlers per hectare (min=0 and max=0.056).

21 Moreover, given that these regions do not appear to be colonization centres in a 1902 official colonization map, it is reasonable to assume that they were not projected to be future settlements in 1884.

22 This methodology requires two assumptions regarding latitude: (1) conditional independence (i.e. once you control for latitude the potential density of population is independent of gaining railway access) and (2) the common support (for each value of latitude the probability of being treated (or not) is positive).

23 Banerjee et al. (2012: 17) use this method to predict the construction of the railway: ‘If there are two cities (or ports) where the difference in distances are less than 100km, we draw a line to both. The line is continued past the city until it hits a natural barrier (e.g. Tibetan Plateau, coast line), or a border to another country.’

24 Occasionally, the information was complemented with additional material. Such as the posterior 1903 Dictionnaire des Communes, Villes & Villages de l’Algérie.

25 The main port in Constantine was Bône, followed by Philippeville and Bougie. These were built during French occupation for military purposes and after the 1850s trade began to expand, particularly because of tariff changes. I do not include the tertiary ports Djidjelli, Collo, Herbillon and La Calle.