Discussing Chevalier’s Data on the Efficiency of Tariffs for American and French Canals in the 1830s

This article revisits Michel Chevalier’s work and discussions of tariffs. Chevalier shifted from Saint-Simonism to economic liberalism during his life in the 19th century. His influence was soon perceived in the political world and economic debates, mainly because of his discussion of tariffs as instruments of efficient transport policies. This work discusses Chevalier’s thoughts on tariffs by revisiting his masterpiece, Le Cours d’Économie Politique. Data Envelopment Analysis (DEA) was conducted to test Chevalier’s hypothesis on the inefficiency of French tariffs. This work showed that Chevalier’s claims on French tariffs are not validated by DEA.


Introduction
Michel Chevalier was one of the most emblematic Frenchmen of the 19th century. He was not only an economist who modified Saint-Simonian positions to liberal ones but also a special world traveler and an influential advisor to Napoleon III. This work will analyze Chevalier's arguments on the role of canal tariffs in economic growth. Although Chevalier also studied import duties, this article will focus on Chevalier's thoughts concerning transport tariffs. As we shall see, Chevalier was a common reference in the French debates on the definition of tariffs in the transportation sector.
For Chevalier, this topic became of the utmost importance after his journey to the United States in the 1830s. Henceforth, Chevalier's discussion of tariffs became one of the most relevant discussions of European canal tariffs produced by a 19th-century economist. The relevance of Chevalier's discussion of tariffs culminated in a free trade treaty between England and France, notably named the Cobden-Chevalier Treaty.
This work will examine Chevalier's discussion of canal tariffs. We will do so considering his life and his main economic work, Le Cours d'Économie Politique. There are several aspects that emphasize the importance of the early 19th-century discussions about tariffs or tolls. First, tolls were realized to be elements that negatively impacted economic growth. Second, they were key elements in the debate pitting individual freedom against public financing. Third, reinforced by Saint-Simon and his disciples (including Chevalier), tolls were perceived as decelerating the rhythm of industrialization. Finally, the French debates on canal tariffs produced the basis for the economic thought of the next generation of European thinkers, such as Léon Walras and the Lausanne School.
We tested a crucial element of Chevalier's discussion: the inefficiency of 1830s French tariffs. This testing was conducted using Chevalier's data and a modern instrument of efficiency evaluation: Data Envelopment Analysis (DEA). It is important to perform this exercise (revise Chevalier's claims) because using current statistical techniques, we can enlighten the field as to whether the analysis of Chevalier was robustly founded by a proper statistical discussion.
The remainder of the article is organized as follows. In section "A Note on Chevalier's Life and Bibliography," we review Chevalier's biography and published works. We also use references not only to historians but also to contemporaries to show Chevalier's impact on academics and politics. In section "Chevalier's Cours d'Économie Politique and His Discussions of Tariffs," we focus on his thoughts regarding tariffs, referring to his masterpiece Le Cours d'Économie Politique and to his letters. In section "Toward a Discussion on Chevalier's Claims Using DEA Analysis," we discuss Chevalier's claims about the inefficiency of French tariffs using DEA techniques. The final section concludes. Michel Chevalier (1806-1879 was a French economist living in a very interesting period of the 19th century. The France known by Chevalier was a country characterized by a proindustrialization movement.

A Note on Chevalier's Life and Bibliography
Michel Chevalier's social, economic, and political thoughts changed over the course of his life. Many authors consider the young Michel Chevalier to be a disciple of Saint-Simon, primarily concerned about improving the quality of life for his contemporary French industrial workers (Takashi, 1980). As he aged, however, Chevalier increased his distance from the orthodox Saint-Simonians, 1 although he retained the idea that social and economic development should be shared among all. As Drolet (2008) noted, "Chevalier moved from being one of the fiercest critics of France's political establishment to one of its most ardent defenders" (p. 1229). Campagnolo (2009) stated that during his life, Chevalier always believed that both productive entrepreneurship and state intervention were necessary for a competitive environment. Of course, this universality was not well received by the most orthodox thinkers, like Carl Grun (1845), who wrote the following about Michel Chevalier: M. Chevalier still refers with great sympathy to the industrialists . . . But to the disciple, the industrialists are no longer, as they were for his master, the proletarians; he includes capitalists, entrepreneurs and workers in one concept, that is to say, he includes the idlers in a category which should only embrace the poorest and most numerous class. (Grun,1845, p. 102) The reasons for this distance from the orthodoxy are both physical and ideological. In other words, this distancing from orthodoxy is due to his journey across the Atlantic Ocean, where he had the opportunity to experience different people, different economic organizations, and, especially, a new inspiration for his ideas to develop the French economy (Adamson, 2005).
Before this journey to America (1833-1834), Chevalier was focused on the European reality and wrote in favor of pan-Mediterranean free trade, which he argued was also important for peace between the East and West: The Mediterranean was an arena, a closed field, where for thirty centuries, the East and the West fought battles. Henceforth the Mediterranean must be like a vast forum where all the previously divided nations will unite together. The Mediterranean will become the wedding-bed of the East and the West. (Chevalier, 1832 2 ) His journey to the United States from 1833 to 1834 (essential to understanding his future critiques of French tolls) was described in his book Society, Manners, and Politics in the United States. 3 This book is a composition concerning the American economy (or society) observed as "a model of industry." According to Chevalier, it should be highlighted how poor Irish men thrived after so little time in America. The reasons for success point not just to some particularly American institutions (such as a business culture and proemployment education) but mainly to efficient American means of transport and internal tariffs; namely, it was the tariffs on the Erie Canal that exerted a special effect on Chevalier's observations, as is documented in detail in the following section. The transport and tariffs that Chevalier observed were the result of intense federal and state debates that had begun in the 1820s (initially with the strong opposition to protectionist tariffs from Southern congressmen; DiLorenzo, 2004;Temin, 1999).
After visiting the United States, a trip that is well documented by Jennings (2006), Chevalier also traveled to Mexico and Cuba (1835), where he strengthened the idea of a Panama Canal linking the Atlantic to the Pacific and developed his principles of a Pan-American alliance of countries. According to Monteiro (2008), after this tour, Chevalier became one of the first authors to explore the concept of a "Latin American country." This was mainly due to his support of Latin American countries being freed from the administrations of Portugal and Spain.
These ideas were influential in South American countries, developing, for instance, the basis of Poucel's thoughts regarding the need for reinforcing French support to the young South American countries that were reaching independence. 4 After returning to Europe, Chevalier visited Great Britain in 1851, recognizing that England's prosperity was due to industrialization. This industrialization was symbolized by the Crystal Palace in Sydenham, which was built for the London Great Exhibition of 1851. The construction of this palace impressed him deeply, especially its rapidity and the building's design and transparency to light. However, he was mainly impressed by British tariffs, which could be noted as the first signal of his favor of the future Cobden-Chevalier Treaty 5 (although Nye, 1991, confirmed that British tariffs were consistently lower than those of France).
As Maddison (1995)  Although recognized mostly for his later political influence in the Second Empire and as one of Napoleon III's closest advisors, 6 Chevalier's academic role as a professor at Collège de France was not limited to his influential books. 7 For instance, Chevalier, along with Villermé and Hyppolite Passy, was also one of the founders of the Societé Statistique de Paris (Statistical Society of Paris;Martin, 2001).
By this time, his influence was already being recognized not only by economists but also in the political world. For instance, Chevalier's bibliography led to fevered debates and significant replicas, such as Gottfried Duden's critique of Chevalier's (1837) Letters on North America, as noted by Rowan (2009), or Carl Grun's critique of Chevalier's Saint-Simonism, which Marx and Engels (1845) severely reproved. Other remarkable influences were assumed by Carl Menger. According to Campagnolo (2009), Menger cited Chevalier 56 times in his leading Grundsätze.

Chevalier's Cours d'Économie Politique and His Discussions of Tariffs
Returning to the original edition (published by Capelle, Libraire-Editeur, in Paris 8 ), we confirm that the book was written for both the academic year of 1841 to 1842 and the first year of graduation. As usual, the lectures had been collected by an assistant to Chevalier (M. A. Broët) and authorized by the professor. The first two sections of this manual are composed from two opening discourses by  All of these lectures were written according to a historicist perspective typical among Saint-Simonian disciples (see Khalil, 1995;Wolfson, 1994). The author begins by pointing out the conclusions of his arguments. He then exposes these arguments while referencing a few statistical data points and many historical cases. Finally, he synthesizes the observations, highlighting the previously introduced conclusion.
As previously noted, there are a significant number of authors who have discussed Chevalier's Saint-Simonism and bibliography as a whole. However, Chevalier's Cours d'Économie Politique is substantially important for understanding the evolution of many of Saint-Simonians to libertarian positions and for its reflection of the general European conviction that the best way for countries to grow economically was a rapid and substantial investment in transportation through tariff reductions.
The question of tariff reduction was one of the most important questions debated in Europe and America in the first half of the 19th century. At the time, there were intense debates in France among transportation specialists. 9 The foci of these debates were not only inland waterways (including canals) but also the emerging railroads. Authors like Chevalier himself or Jules Dupuit (Groenewegen, 1983;Klein, 1985;Simonin & Vatin, 2002) not only considered the importance of tariffs for economic growth but also were concerned with their redistributive function via the pricing system. It is also important to mention the debates among the economist-engineers from the Ponts et Chaussées, which bolstered contemporary economic thinking regarding pricing in transportation.
When we read Chevalier's Cours carefully, it is easy to see that it is not a general economics book in the way we conceptualize it today, as a general introductory book for a first-year undergraduate economics course. Chevalier's Cours is concerned primarily with free trade and arguing against tariffs. 10 The concluding claim is clear: France imposed excessive tariffs on the majority of goods and services, which diminished its rhythm of growth and, obviously, diminished the success of its distribution efforts.
As noted by Takashi (1980), Chevalier had tried to put these ideas into practice some years before. Seeking to reduce tariffs, 11 he proposed an enormous development plan for French ways of communication in 1838 to improve the linkage among usable rivers and railroads. This plan would have cost 1.184 million francs over 12 years. The plan that was actually approved by the ministry had a final cost of 2.8 million francs.
Chevalier observed that although France and England had the same area and almost the same population, they had very different systems and efficiencies regarding water canals. As a result, Chevalier stated, "les canaux français ne rendent pas la dixième partie des services qu'on est en droit d'en rendre" 12 (Chevalier,p. 303). Chevalier knew that tolls covered infrastructure costs. 13 This point is noteworthy, as the 19th-century transportation specialists argued (particularly Jules Dupuit, according to Simonin & Vatin, 2002). Chevalier wanted to discuss the fact that in some instances, toll levels were exceeding infrastructures costs, which consequently limited the extension of the market and eventually economic growth.
Chevalier used data on American canals from 1835 to 1839 and on French canals from 1821 to 1836. These data were collected directly from official sources. After 1842, Chevalier considered a more rigorous examination of these data because in the second edition of Chevalier (1842), that is, in 1855, he introduced notes about data beyond 1841, reprinting the original table and adding a note on the stability of canal tariff during the studied period.
When comparing French with American canals, Chevalier is even more emphatic: Les canaux américains sont ceux qui ont donné les résultats les plux extraordinaires. 14 (p. 303) Chevalier did not particularly focus on English, Belgian, and Dutch tariffs (which he considered low), but he did applaud American tariffs: Using a t test on the equality of the means between French and American tariffs collected by Chevalier at four French canals and at four American canals (Figure 1), we get the following probability that the null hypothesis is observed (French tariffs are the same than American ones): manure, 30.2%; coal, 36%; flour, 57.8%; wheat, 68%; wine/salt pork, 85.3%; iron, 73.3%; and textiles, 60.7%. Therefore, we can conclude now that French tariffs could not be so significantly higher than American ones as Chevalier was claiming.
It thus becomes pertinent to ask the following: Was Chevalier exaggerating due to his fascination with the British and American systems? Were his politics dominating his technical and academic abilities? Was he forgetting to consider the different economic and social conditions of France and America? Or, was he interested in increasing the rhythm of change by denouncing a larger gap than was actually present between French and British or American tariffs?

Toward a Discussion on Chevalier's Claims Using DEA Analysis
This section will use DEA to test Chevalier's argument that French tariffs were exorbitantly high for his contemporaries and fellow citizens. DEA is a nonparametric programming approach to technical frontier estimation (Coelli, 1996). It allows us to compute efficiency measures considering given outputs and a set of inputs. Although the number of academic works focusing on historical data using DEA is currently not as large as the number of works that use DEA to analyze more recent data, this approach has received increased attention even from those who are interested in researching data from the past (Enflo & Baten, 2007;Mourao, 2009;Nenovsky, Ivanov, & Mihaylova, 2008;Tankersley & Cuzan, 2004;Wheelock & Wilson, 1995).
With Chevalier's data and with his claims on inputs, we use an input-oriented model and a cost-minimization model characterized by variable returns to scale. We follow McGillivray, McLean, Pahre, and Schonhardt-Bailey (2001), and we also consider each toll as the result of the inputs of the transportation process.
Our model (a cost-minimization model) is described by the following minimization problem (Coelli, 1996): In our case (please refer the raw data in the appendix), w i is a 6 × 1 matrix of tolls 19 for the ith canal and x i * (which is calculated by the linear programming of DEA) is the costminimization vector of economic and geographical characteristics for the same ith canal. y i , x i * , or x i are also 6 × 1 matrixes. N1 is an 8 × 1 vector of ones (convexity constraint) as in Coelli (1996), with N being the number of firms (or, in our case, the number 8 of canals). λ is a 8 × 1 vector of constants. X is the 6 × 8 inputs matrix. Y is a 6 × 8 matrix of normalized outputs. Therefore, our minimization problem can be described as a problem in which we want to achieve the maximum output with the minimum value of inputs' expenses or inputs' costs (Coelli, 1996). For instance, for the same output, a canal with a toll of five monetary units paid for 5 miles exhibits a more efficient ratio (in the perspective of the toll payer) than another canal with a toll of five monetary units only paid for only 2 miles, keeping the other inputs (geographical and economic characteristics) constant.
The total cost efficiency (CE) or economic efficiency of the ith canal would be calculated as CE = ′ ′ wi xi wi xi * , meaning that CE is the ratio of minimum cost to observed cost.
Chevalier recognized that the tolls and the costs of each canal changed both because of the nature of each transported product and six other geographical and economic dimensions (our inputs): the distance of each canal; the minimum, maximum and differential of altitude; the market area; and the population density. Therefore, we have to identify these factors as inputs of our model. 20 Our first source for these variables is Chevalier's table that is reproduced in Figure 1. We must remember, however, that Chevalier's data exhibit some weaknesses: Besides being the product of individual collection from official reports, these data have not been conveniently discussed in terms of efficiency. Thus, DEA is an appropriate method for analyzing Chevalier's claims on tariffs for various reasons. First, it has been proven to be an appropriate method for analyzing efficiency related to decision-making units. Second, Chevalier's complaints concerned the modest efficiency of French tariffs. Third, they allow the assumption of variable (as opposed to constant) returns to scale to characterize the output.
For the topological characteristics, we also returned to Pierre (1997) for French canals and Shank (1986) and Bernstein (2005) for American canals. For the population dimensions of the regions around the canals, we used 19thcentury data from the National Historical Geographic Information System (NHGIS; for North American areas) and from the Institut National de la Statistique et des Etudes Economiques (for French areas).
The descriptive statistics are at Table 1. Our raw data are shown in the appendix. It is clear that our model should assume variable returns to scale because constant returns to scale is adequate only when we are observing decision-making units operating at an optimal scale (Banker, Charnes, & Cooper, 1984). When decision-making units face imperfect competition, constraints on finance, or different geographical areas, then it is argued that a variable return to scale model is the most appropriate. Our model is an input-oriented model because this orientation is associated with the cost-minimization model and because, as Coelli and Perelman (1996) assumed, this orientation is preferred when input dimensions are the primary decision variables.
We will pass over formal explanations of DEA and efficiency concepts. For those interested in these themes, we suggest Coelli (1996). To give an interpretation of the total CE values that we are going to report, we state that our efficiency measure identifies the ratio of minimum cost to the observed cost for each transported product:

CE of each canal
Minimum cost observed for all canals Cost observed f = o or each canal .
In our model, less efficient decision-making units (canals) face costs that are too high considering the set of inputs; the most efficient decision-making units require lower costs to produce the same unit of output produced by the less efficient units. We ran our estimations using WIN4DEAP v.1.1.2.
Our results on the efficiency of each canal detailed by Chevalier are reported in Table 2 and graphically represented in Figure 2.
To help the reader, we offer an interpretation of the values in Table 2 and Figure 2. Focusing on a given productmanure-we observe that the canals that exhibited the minimum cost (Efficiency estimate = 1) were Midi, Briare, and Schuylkill. In contrast, the Pennsylvania Canal can be confirmed as the canal characterized by the maximum cost (Efficiency estimate = 0). The Centre Canal can be highlighted as a canal showing median costs (Efficiency estimate = 0.4) for manure and considering the set of inputs.
Our estimates indicate that Chevalier (1842) was not far off. In fact, our estimates provide a mixed lecture on the efficiency of tariffs. Two American canals, Schuylkill and Delaware, were identified as having the most efficient tariffs. The two other American canals analyzed by Chevalier (1842), however, exhibit less efficient estimates-namely, his highly favored Erie Canal and the Pennsylvania Canal. However, the Briare Canal also shows highly significant efficient tariffs, whereas Saint-Quentin Canal exhibits slightly lower values. The Centre Canal and Midi Canal are identified as the French canals with the least efficient tariffs.
Our values show that, considering the evolving reality, only the Centre and Midi canals were as costly as Chevalier argued. However, examining within the American reality and considering the set of inputs analyzed in our model, the Erie and Pennsylvania tariffs were too high for the users, even if their values were lower than French values.
The advantages of DEA method are known: We do not need to know a priori the form of production frontier. Nonparametric methods allow the DEA test to ignore the specification of a particular functional form for the technology. Moreover, the use of small samples is possible (in our article, eight canals considered as eight Decision Making Units [DMU]): From this perspective, the methodology required for using Coelli's (1996) model is correct. We also know, however, that this model is very sensitive to misspecified data and errors due to the deterministic feature of the method: Each discrepancy from the frontier is immediately interpreted as inefficiency and statistical noise and inefficiency are not distinguished. To account for these limitations, we also tested different compositions of the set of inputs using the 63 different combinations of the six inputs, 21 and our results retained the previously signaled pattern: Schuylkill and Delaware were the most efficient American canals, and Saint-Quentin was the most efficient French canal. After running the 63 combinations and using SimLab (data sensitivity software), we obtained a central value for each estimate of efficiency and a corresponding standard deviation. Full details are available under request. Table 2 also expresses these values.   Source. For canals' tariffs, see Chevalier (1842). For distance, minimum, maximum, and differential of altitude, see Pierre (1997), Shank (1986), and Bernstein (2005). For area of market and for population density, see National Historical Geographic Information System (2009)  Although we experienced data scarcity, we also tried to include a proxy for the price of each product using the mean value of the inverse of quantities produced in France and the United States between 1850 and 1860 and those suggested by Chevalier. Even with this innovation, our DEA results did not change.
Our results show that even when using modern techniques like DEA, we are unable to confirm the efficiency of all American canals' tariffs over French canals' tariffs. In fact, French canals exhibit efficiency values that are central values when compared with the extreme values assumed for American canals.
This points to a political motivation behind Chevalier's emphasis on American canals. His argument was based only on estimates for gross tariffs, and his claims were mainly aimed at accelerating the improvement of the French system's methods of transportation. Highlighting the differences in tariffs, Chevalier reinforced the need for France to deepen the process of industrialization through improved canals, railroads, and land ways.

Conclusion
This work revisited the main aspect of the liberal Saint-Simonian Michel Chevalier's economic thought, namely, his discussion of the role of tariffs in economic growth. Chevalier was a Frenchman involved not only in the controversies of Saint-Simonism but also in a global perspective of economic growth and redistribution processes. His journeys around the world, especially to the Americas, clarified his thoughts and oriented them toward proindustrialization preferences using convenient policies of transportation. His reflections led him to discuss tariffs as a main theme in his bibliography.
His book Cours d'Économie Politique collected his main economic thought as distilled by the classes he taught at the Collège de France in the 1840s. In this work, Chevalier claimed that France was losing its rhythm of economic growth because French canal tariffs were too high compared with British and American tariffs.
We proved in this article that Chevalier's claim was essentially political. Using DEA techniques, we calculated estimates of French tariffs that were not as inefficient as Chevalier claimed. In fact, in our sample, some French tariffs, such as the Briare Canal, exhibited the most efficient results even when considering American tariffs. In this finding, we can agree with authors like Nye (1991) and argue that Chevalier was trying to denounce a larger gap than actually existed to accelerate the changes he wished to see in French transportation policy.