The equally “bad” French and English farmers of Quebec: New TFP measures from the 1831 census

ABSTRACT

New TFP estimates drawn from the neglected census of 1831 for Lower Canada are used to test the controversial (but still dominant) traditional “poor French farmers” explanation for a prolonged economic crisis. The new evidence shows that French-speaking areas were equally as productive as English-speaking areas, something that upturns the established consensus and reinforces the minority viewpoint that culture had little to do with the crisis. Using a broad range of controls, the researchers find that this conclusion is robust and that other variables such as settlement recency, environment, and economic structure were much more significant determinants of TFP. These results warrant the abandonment of the cultural explanation and a shift toward other explanatory channels.

KEYWORDS:

• Agricultural productivity
• Canadian economic history
• total factor productivity

Acknowledgments

Vincent Geloso wishes to acknowledge the help of Germain Belzile, Stephen Broadberry, Alexandra Foucher, Chris Minns, and Gilles Paquet. Vadim Kufenko would like to thank Prof. Dr. Klaus Prettner and Prof. Dr. Harald Hagemann for their support during the work on the given paper. Michael Hinton acknowledges the help, comments, and guidance of Alan Green, Frank Lewis, and Marvin McInnis provided some 40 years ago at Queen's University at Kingston while he was writing an MA essay which he continued to benefit from in the writing of this article.

Notes

¹ For surveys, see Le Goff (1974), Paquet and Wallot (1975), McCallum (1980), McInnis (1982), Armstrong (1984), Little (1986), and Russell (2012).

² There has often been confusion regarding the points of the revisionist regarding efficiency. It is quite possible to say that two firms are equally efficient, but because one firm has fewer machines or fewer workers, have different levels of incomes. Thus, one can argue that farming efficiency was not relevant to the poverty of the French-Canadians without negating that they were poorer.

³ We put this in quotes because some, notably Gilles Paquet and Jean-Pierre Wallot (2007), would argue that there was no crisis.

⁴ We are perplexed that the scholars in the field who support this view have not attempted to link with the economics literature on loss-averseness. It might be possible that the French-Canadian peasantry was loss-averse and opted for strategies that limited potential losses, while also limiting potential for greater gains. This would be reminiscent of McCloskey's argument about scattered plot farming in pre-modern Europe (1976). Such an argument would probably cut across ethnic lines (loss-averseness is not inherently cultural), and has never been made. If it had been made, it would be easier to engage with, as McCloskey has laid out a framework to test for the presence of this hypothesis.

⁵ The areas considered were in the counties of Beauharnois (the townships of Hinchinbrooke and Hemmingford and the parishes of St-Timothée and St-Clément), the counties of Rouville and Mississquoi (the parishes of Foucault, St-Armand, DeLery, St-Mathias, St-Mary, and St-Athanase, and the townships of Stanbridge and Farnham) and the counties of Mégantic, Lotbinière, and Beauce (the parishes of St-Croix, Lotbinière, Ste-Marie, and St-Joseph, and the townships of Inverness, Ireland, Leeds, and Broughton) (Hinton 1977, 29). According to the census of 1831, these accounted for 7.6% of the population of Lower Canada when the urban agglomerations of Quebec and Montreal are included, and 8.9% when we exclude those urban areas. Note: The exclusion of field crops is probably the largest flaw in Michael Hinton's work since, as Little (1986) pointed out, the areas that were settled by English farmers tended to specialize in pastoral production. Indeed, as we will point out later in this article, English farmers tended to be located on lands unfit for crop growing but quite well-suited for pastoral production.

⁶ Since our goal is to determine whether cultural factors were at play, we decided (for the sake of this article) to eschew the role of seigneurial tenure. This is not because we believe that this is irrelevant, but we prefer to tackle the cultural issues first. See more in footnote 17.

⁷ There is a belief that the Appendix to the Journal excludes potatoes. This is not the case; the appendix does include potatoes. They are also included in the rolls themselves.

⁸ It would be improper to dismiss too promptly the census of 1831 for its drawbacks, as most censuses of the time in Canada had their own serious flaws (Fortier 1984; Thibeault 1989; Curtis 2001; Dillon and Joubert 2012, 2015; Kennedy and Inwood 2012).

⁹ To solve the problem, they grouped males over 14 years of age into five occupational categories (farmers, laborers, servants, persons of private means, and others). The proportion of laborers and servants attributed to agriculture in each district was taken to be the same as the ratio of farmers to the number of farmers plus others. In addition, they had missing information for 11 districts and must assume that each district was equal to the average of the county.

¹⁰ The census heading in French is “nombre de familles qui gagnent leur subsistence par les travaux de l'agriculture,” which translates into “number of families earning their subsistence by agricultural work.” The website of Library and Archives Canada states that the translation is “number of family members [our emphasis] earning their subsistence by agricultural work,” which is not a proper translation of the heading.

¹¹ The exact heading is the “nombre de serviteurs employés comme fermiers dans chaque famille” [number of males employed as farmers in each family].

¹² Dennis Cook's (1977) distance is one of the simplest methods for detecting outliers by using leverage and normalized squared residuals. The influential observations according to this are Brandon du Lac, Kilkenny, Mansfield, Litchfield, Clarendon, Bristol, Petite Nation, Deschambaults, Île Verte, St-Fabien, Ste-Flavie, Petite Rivière, Shefford, Milton, Barford for ln TFP1, column 2; and the same with an exception to Buckland and Standon and Lochaber for TFP2, column 6.

¹³ In addition to the previous outliers, in Tables 7 and 8 also detect the following outliers for both types of TFP: Ste-Anne, Longue Pointe and Frampton.

¹⁴ The MM-estimator is a two-step procedure involving the S-estimator as in Rousseeuw and Victor Yohai (1984) and the M-estimator as in Huber (1973).

¹⁵ The estimated scale parameter for the MM-regressions varies from 0.1873 to 0.232.

¹⁶ Even in their 1984 article, Frank Lewis and Marvin McInnis presented data suggesting that English counties were richer than French counties.

¹⁷ As we asserted above, we do not seek to question the issue of seigneurial tenure here. However, the data assembled here forms the basis for any paper that would aim to empirically assess the role of seigneurial tenure on depressing net incomes. Since poverty can even be found among those who are relatively efficient on the margin, if seigneurial tenure reduced the size of their surplus (net income), its effects would not be observed in the TFP measures. As such, the data developed here are the stepping stone for testing the arguments of Allan Greer, Fernand Ouellet, and Morris Altman.

¹⁸ Nafziger (2010) made a similar argument in the case of late-nineteenth-century Russian serfs by pointing out that there was substantial reallocation of labor inputs to non-farm activities in order to maximize income.

¹⁹ William Evans was the secretary of the Montreal Agricultural Society.