The research on which this article is based was supported by a postdoctoral fellowship from the Foreign Area Fellowship Program of the Social Science Research Council and American Council of Learned Societies. El Colegio de Mexico and the Banco de Mexico assisted me by making research and office facilities available; while in research and early drafts I benefited from discussion with, and suggestions from Gerardo Bueno, Leopoldo Soils, Clark W. Reynolds, and Clara J. de Bialostozky. A referee also contributed useful suggestions. Their help and support are gratefully acknowledged.

¹ See El Colegio de México, Seminario de Historia Moderna de México, Estadísticas económicas del Porfiriato: Fuerza de trabajo y actividad económica por sectores, Mexico, D.F., no date (c. 1964). Their labor force data are explained on pp. 1–4 and presented on pp. 38–60. As an example of a misleading datum in this section of the study, they report “professionals” in 1900 as numbering 137,245 (p. 57), whereas the original census found 62,207 people in the “free professions” by including not only everyone who would be considered professional by any stretch of modern definitions (including musicians, artists, actors, midwives, pharmacists, and draftsmen), but also mechanics, typists, stenographers, business agents, and construction foremen. To this dubious list the El Colegio study seems to have added launderers, cleaners, barbers, and prostitutes as well as bankers, acrobats, bullfighters, florists, and “administrative and clerical personnel of industrial establishments” (p. 4). As another example drawn from the same data on services, a category “other services” appears at the end (p. 59) without any explanation that this is a sum of the five categories previously given, including “professionals.”

² See for example Nacional Financiera, La economía mexicana en cifras, Mexico, D.F., 1966, Cuadro 4, giving semi-official figures; or Presidencia de la República y Nacional Financiera, Cincuenta años de Revolución Mexicana en cifras, Mexico, D.F., p. 29 (this is not to be confused with another volume with a similar title quoted in the next and subsequent footnotes).

³ For the initial effort to extend Mexican GDP series backward over time, see Enrique Pérez López, “El producto nacional” in México: 50 años de Revolución, I, Tomo, La economía, Mexico, D.F.: Fondo de Cultura Económica, 1960, pp. 571–92Google Scholar. A revised version has now been published by the same author as “The National Product of Mexico: 1895 to 1964,” in the collective volume Mexico's Recent Growth: The Mexican View, with essays by López, Pérez and five others, introduction by Tom E. Davis, and translation by Urquidi, Marjory (Austin: University of Texas Press, 1967), pp. 23–44Google Scholar. Partial revisions sticking to rather similar procedures have been made by the Banco de México (see Nacional Financiera, La economía, Cuadro 5); Clark W. Reynolds (his forthcoming book); and Leopoldo Solís (unpublished estimates used later in this study). All these studies to date have relied for the years 1895 to 1910 on the El Colegio volume's index of manufacturing output (Estadísticas, pp. 105–6 with explanation on pp. 5–9). Yet this index was based on a simple blowup of the output trends in the handful of sectors that were then relatively mechanized—textiles, tobacco, sugar, alcohol, and (after the first modern plant was built in 1902) iron and steel (ibid., p. 7). The ratio of output in these, relative to other sectors, was based on the relationships shown in the first industrial census in 1930 (ibid., p. 8), which incidentally seems to have relied in part on the 1930 population census data used here. This procedure ignored the fact that in the 1895 to 1910 period, most employment in manufacturing still consisted of handicraft artisans, whose numbers changed only slowly and started to diminish from 1900 to 1910 in the face of mechanized competition. The census labor force data for sectors such as commerce, construction, and services could also be used to improve and check the available output estimates. These data suggest, for example, that Pérez López' procedure of estimating 1895–1938 output outside agriculture, mining, manufacturing, and transportation by projecting backward the 1939 proportions yields a strong upward bias for the early years. Solís' unpublished estimates, which extend the Péiez López coverage into a few other sectors, stand the test of these comparisons more successfully, especially when the additional sectors are aggregated (cf. Table 2); his data imply a more rapid rate of growth from 1895 to 1930 than do earlier figures. But it may prove useful to supplement output data by assigning assumed productivity weights to particular categories of labor, where direct output data are scarce and unreliable.

⁴ The 1895 census reported occupational data for three categories of people, those “present” (presentes), “in transit” (de paso), and “absent” (ausentes). All calculations reported here are based on the sum of the first two, or what the census calls the population “of fact” (de hecho). Analogous data for 1900 were reported separately by state, but the national totals have been summed in unpublished research by Clara J. de Bialostozky of El Colegio de México. Naturally, the rough similarities between the findings of the first two censuses do not mean that any of the Mexican census data can be considered very reliable. There is plenty of room to doubt their accuracy; but we have practically no other sources of information on the composition of the labor force up to 1930, and in most cases we cannot be sure of the biases.

⁵ See the tables in the Resumen General of the 1910 census itself, El Colegio, Estadísticas, pp. 38–60, and Economía, Secretaría de, Dirección General de Estadística, Estadísticas sociales del Potfriato, 1877–1910, Mexico, D.F., 1956, pp. 18–19Google Scholar.

⁶ See for example the comments in the official foreword to Secretaría de la Economía Nacional, Censo de funcionarios y empleados públicos, 30 de Noviembre de 1930, Mexico, D.F., 1934, p. 7Google Scholar, which also indicated that the most common ambiguous designation was clerical employee (empleado), followed by production worker (obrero) and others.

⁷ The data were given, not in the Resumen General of the census as in the preceding censuses, but in a Parte Especial, published separately in 1954.

⁸ For example, fully 22.8 percent of the insufficiently specified workers of that year were clerical and related workers, a larger proportion than in any of the eight broad sectors for which occupational distributions were given. See the Resumen General of the 1950 census, Cuadro 6.

⁹ There are three basic sources for subsequent years. One is the unpublished sample from the 1960 population census made by the Center for Economic and Demographic Studies of El Colegio de Mexico, essentially a 1.23 percent sample that can be blown up to give a generally but not perfectly reasonable picture. Second, there have been industrial censuses, of which that of 1965 is easily the most reliable; but even that one underreports small-scale enterprises and is only dependable in the few manufacturing industries consisting mainly of large enterprises. A notion of the margin of error in the other industrial censuses can be gleaned by the fact that the 1960 one showed less than 830,000 manufacturing workers, compared to 1,556,315 in the population census of the same year; while the 1955 industrial census totals, in sharp contrast, were too high, with an employment distribution by industries that disagreed sharply with all available data from the population censuses, subsequent industrial censuses, and output trends. The third source is working estimates by official bodies. Several such estimates have been made, but only for very broad groups of industries; and they have been constructed very simply on the basis of assumed coefficients and output trends. Needless to say, these data are no better than the unproved assumptions and generally imperfect output statistics on which they are based.

¹⁰ This conventional division is emphasized, for example, by Cline, Howard F. in Mexico: Revolution to Evolution 1940–1960 (New York: Oxford University Press, 1963)Google Scholar, and by Clark W. Reynolds in his forthcoming book, The Mexican Economy: Twentieth Century Structure and Growth, to be published by Yale University Press for the Economic Growth Center.

¹¹ For a renowned interpretation of these relationsips see Vemon, Raymond, The Dilemma of Mexico's Development (Cambridge: Harvard University Press, 1963)Google Scholar; see also Vernon, Raymond, ed., Public Policy and Private Enterprise in Mexico (Cambridge: Harvard University Press, 1964)CrossRefGoogle Scholar.

¹² See for example United Nations, The Economic Development of Latin America in the Postwar Period, New York, 1964Google Scholar, ch. 1 and passim, and Bruton, Henry J., “Productivity Growth in Latin America,” American Economic Review, LVII (Dec. 1967), 1099–1116Google Scholar.

¹³ Indeed, in my view much of the success of Mexico's development policies since 1940 can be attributed to the pragmatism of the post-Revolutionary leaders in staying with successful older solutions to basic economic problems, and making modifications rather than radical reversals to correct unsatisfactory aspects of these past policies. In this light, the success of Mexican policies has been partly due to a carry-over and continuity from the liberal-progressive, yet laissez-faire and stability-minded, policies initiated in the later nineteenth century. Mexico, by a certain conservatism, has avoided many of the more recent fashions that have caused grievous distortions in development elsewhere in Latin America. Let me add that in my judgment rather similar continuities with late nineteenth-century progressive policies have also played favorable roles in the development of much of the Far East (including Japan) and southern Europe (including Italy).

¹⁴ For relevant data see especially México: 50 años de Revolución, Tomos, I and II, Mexico, D.F.: Fondo de Cultura Economica, 1960Google Scholar; Wilkie, James W., The Mexican Revolution: Federal Expenditures and Social Change Since 1910 (Berkeley: University of California Press, 1967), ch. 9Google Scholar; National Financiera, La economía; and the forthcoming book by Clark W. Reynolds.

¹⁵ Cf. Reynolds, ch. 1, Table 1:2.

¹⁶ The GDP data for 1895 and 1930 are from Leopoldo Solís' unpublished estimates, which are subject to further revision. Other estimates are slightly higher than his 1930 figure (i.e., around 15 billion pesos) and are considerably higher (closer to 9 billion pesos) for 1895.

¹⁷ Solís' estimates. Here again, the 1895 figure in particular is speculative and subject to dispute. For references see fn. 3.

¹⁸ When women corn-grinders (molenderas) are included as nonagricultural labor, as they are in summary data of the 1900 and 1910 censuses themselves, agriculture and related pursuits (forestry, hunting, and fishing) employed just 63.8 percent of the labor force in 1895 but 66.5 percent in 1910. With corn-grinders excluded, according to my calculations, the figures become 66.5 percent in 1895, 65.9 percent in 1900, and 68.1 percent in 1910. Immediately after the Revolution in 1921, with these women no longer counted, the equivalent figure fell to 66.8 percent, though this proportion may have been biased downward by underreporting of the rural population in the 1921 census. In 1930 the proportion had risen to 68.6 percent. By 1940, this ratio had begun to fall, and stood at 65.4 percent. It then declined sharply to 58.3 percent in 1950 and 54.2 percent in 1960.

A small absolute fall in manufacturing employment from 1900 to 1910 is clear (cf. El Colegio, Estadísticas, p. 48). I have posited a similar mild decline from 1921 to 1930 based on my own interpretation of the 1921 census. From 1930 to 1940, in turn, though the numbers rose, the proportion of the labor force in manufacturing seems to have remained roughly constant, between 9.9 and 10 percent. My agricultural figures differ from those of the El Colegio study (which would imply percentages of 62.5 in 1895 and 67.1 in 1910; cf. pp. 39 and 45), in that I have excluded entirely from my computations what El Cofegio interpreted to be insufficiently specified workers, who constituted 6.7 percent of their ‘labor force” in 1895 but only 1.2 percent in 1910. As their own table (p. 60) shows, almost three-quarters of the people in this category were women. Well over half were found in the then largely rural states of San Luis Potosi and Quer´taro, and the rest largely in Nuevo León and Oaxaca. The 81,339 men included just 2,166 in the capital district. I have entirely omitted all these people with insufficiently specified “occupation” from my own calculations for 1895. Even under my procedure, the participation rate in 1895 was higher than in later years (35.2 percent compared to slightly over 32 percent in 1930 and 1950), which makes it almost certain that very few of these people could have been economically active by modern definitions.

¹⁹ Other special assumptions employed in my computations will be spelled out later.

²⁰ In one respect his computations are not independent; he has added a deflating factor to his early commerce series, to take into account the spread of a commercial economy and to bring the productivity implications more into line with'those for other sectors.

²¹ By my computations, from 1950 to 1960 output per worker grew 46.2 percent in utilities, 26.8 percent in manufacturing, and from 5.2 percent (mining) to 19.1 percent (agriculture) in the other six broad sectors. These figures, especially the one for manufacturing, have a slight upward bias since insufficiently specified workers were far more numerous in 1950 than in 1960; the computations were based on the censuses and standard GDP data from Nacional Financiera, La economía, Cuadro 8. By comparison, to judge by Gonzalo Robles' index of industrial output, industrial productivity grew more rapidly, at rates of around 50 percent per decade, from 1900 to 1910, from recovery circa 1922 to 1930, and again from 1930 to 1940 (México: 50 años de Revolución, Tomo I, p. 197).

²² In 1895 they included 3,240 “administrative and office personnel of industrial establishments,” 45,435 “production workers of industrial establishments,” 5,686 “mechanics,” and 24,431 “day laborers” outside agriculture.

²³ I have assigned 10,000 of the others to railroad transportation and two-thirds of the day laborers (16,288) to construction, leaving the remaining 13,108 unspecified. These, plus 2,758 fabricantes de carbón (makers of coal or charcoal), whom I hesitatea to classify, comprise the insufficiently specified workers in my Table 1.

²⁴ The 1930 census reported 56,742 workers in “sugar, alcohol, and partly refined sugar (piloncillo o panela).” This number is far out of line with all relevant data on these industries before or since; and I have corrected it by revising downward the number of workers in this portion of the food and beverage industry to just 15,000, assigning the other 41,742 workers in this category to agriculture, on the theory that this is where they would have been ordinarily classified. By comparison, almost no workers in these industries were recorded as such through 1921; while the 1935 and 1940 industrial censuses, covering the larger establishments which predominate in sugar, found only 13,791 and 11,837 workers in the sugar industry, respectively. Meanwhile, sugar cane production followed a pattern of steady growth (El Colegio, Estadísticas, p. 72, and México: 50 años de Revolución, p. 148). To clinch the point, there cannot have been many more than 15,000 workers in these industries in 1940, since there were just 93,968 food and beverage workers in all, compared to my estimates (including ihis correction) of 111,670 in 1930.

²⁵ Based on the approximate proportions in 1921, I have assigned 5,000 of the 18,866 hatmakers of 1895 to soft-fiber hats, and thus to the clothing industry, and the remaining 13,886 to straw hats, counted with textiles following modern classification procedures. Of the 11,636 workers in “manufactured metal products” in 1930, I have assigned three-fourths to fabricated metal products and one-fourth to machinery and electrical equipment. An obvious step affecting a large number of workers was to exclude from manufacturing (and from the labor force) molenderas, women who ground corn by hand, who numbered 211,504 in 1895. The 1900 and 1910 censuses counted these people in the food industry under manufacturing.

²⁶ México: 50 años, p. 197; Nacional Financiera, La economía, Cuadro 32.

²⁷ By 1921, there were 6,123 “administrative and clerical personnel of unspecified industries,” 108,596 “production workers of unspecified industries,” 26,875 “mechanics,” and 65,368 laborers, not in agriculture or construction,” or a total of 186,962 in categories roughly equivalent to those that accounted for 78,792 workers in 1895.

²⁸ El Colegio, Estadísticas, pp. 107, 108.

²⁰ See, for example, the tables in Horowitz, Morris A., Zymelman, Manuel, AND Herrnstadt, Irwin L., Manpower Requirements for Planning, Vol. II (Boston: Northeastern University, 1966)Google Scholar.

³⁰ For 1895 I have included two-thirds of the 6,796 stone-cutters (canteros), assigning the rest to construction based on subsequent proportions. The 1921 data are undoubtedly an underestimate, since the census recorded only 71 petroleros, despite the record extraction of petroleum in 1921.

³¹ For output data on mineral and petroleum production, see especially Pérez López' national product series, “El producto nacional”; Nacional Financiera, La economía, Cuadros 22 and 23; and José Campillo Sainz, “Los recursos naturales no renovables” in México: 50 años de Revolucion, Tomo I.

³² See Villegas, Daniel Cosío, ed., Historia modema de México, El Porfiriato: La vida ecónomica, Tomo I, Mexico, D.F. and Buenos Aires: Editorial Hermes, 1965, p. 568Google Scholar; Sainz, México: 50 años, p. 463 ff.; and National Financiera, La economía, Cuadro 76.

³³ The 1950 total included 3,275 in air transportation, 1,535 in lone-distance ocean transportation, 5,560 in other river and maritime transportation including port services, 61,702 in rail transportation, 35,628 in buses and streetcars, 20,243 in other land transport of passengers (including taxis), 48,121 in highway cargo transport (i.e., trucking), and 1,837 in warehousing.

³⁴ A sizeable proportion of these highways were not paved; see National Financiera, La economía, Cuadro 79. According to this table, in “1925–28” there were just 695 kilometers of highways, of which only 241 were paved.

³⁵ National Financiera, La economía, Cuadro 26.

³⁶ In 1910 a census category that consisted mainly of construction workers reached 143,435. See also El Colegio, Estadísticas, p. 49, which shows an increase of 25,109, or over 50 percent from 1895 to 1910, in a restricted group of construction-related occupations that exclude carpenters and day laborers. The 1921 census, in contrast, showed only some 83,975 construction workers that year. For 1940 I get an estimate of 95,168 construction workers, dividing workers in ambiguous categories according to proportions midway between those of 1930 and 1950. The occupational composition is not very revealing except for one sign of the times—the number of plumbers grew from 488 in 1895 to 2,051 in 1921.

³⁷ The “proprietors” numbered 38,560; and a considerable proportion were probably in hotels, restaurants, and similar establishments that would more usually be classified under services. Using a scheme only marginally different from mine, El Colegio, Estadísticas, p. 52, shows a steady growth in commerce from 249,605 in 1895 to 293,753 in 1910. A rather narrower scheme of classification in 1921 showed some 268,490 people, about equal to the 1930 figures (267,556), before the latter was corrected by the addition of one-fifth of the insufficiently specified workers of that year.

³⁸ The census figures for employment in commerce, prior to any allowance for insufficiently specified personnel, were 509,836 in 1940 and 684,092 in 1950. Adding one-fifth of all insufficiently specified personnel, the 1940 figure would become 542,568, which is curiously high compared to 1930, in relation to other changes over the same decade.

³⁹ Like the other fastest-growing sectors—including such diverse activities as electric power, air transport, communications, and various services—finance is, of course, an especially skill-intensive sector requiring much technical know-how. For asset and other data on the rapid growth and diversification of the financial sector, see Brothers, Dwight S. and Leopoldo, Solís M., Mexican Financial Development (Austin: University of Texas Press, 1966)Google Scholar.

⁴⁰ Launderers, who were among the many humble workers displaced by mechanization, declined from 49,905 in 1895 to 32,838 in 1921 and then fell sharply to 12,711 persons—including employees of dry-cleaning, pressing, and ironing establishments as well as laundries—in 1930.

⁴¹ The Mexican figure is based on the Resumen General for that year, Cuadro 6. In 1960 the proportion was slightly smaller, 18.4 percent, perhaps because fewer service workers remained unclassified. For data on other countries see Horowitz, Zymelman, and Herrnstadt, Manpower Requirements.

⁴² Even as late as 1924, in a single year only 28 engineering and 16 architecture students were graduated; see the tables in Secretaría de Educación Pública, La educación publica a través de los mensajes presidenciales, Mexico, D. F., 1926, pp. 552–55Google Scholar. The anuario estadístico de los Estados Unidos Mexicanos, 1941, Cuadro 163, gives the total number of engineers and architects given professional titles from 1901 to 1927 as 1,323.

⁴³ There must have been from 30,000 to 35,000 teachers in 1930, but they cannot be distinguished from other public employees in the census of that year.

⁴⁴ Secretaría de la Economia Nacional, Censo de funcionarios y empleados públicos, 30 de Noviembre de 1930, Mexico, D.F., 1934Google Scholar.

⁴⁵ The figures are from El Colegio, Estadísticas, p. 55.

⁴⁶ To add support to this view, between 1900 and 1910 the reported number of male domestic servants fell from 93,947 to 59,351, while male empleados particulares rose from 31,945 to 75,054. A decline of 6,133 in female servants was offset by a growth of 6,393 in the number of women in this same category. (Ibid., pp. 58 and 55). Of course, even if this makes it fairly obvious that this tide became a euphemism by 1910, I will grant that the status of the title in 1895 is far from clear. It probably referred mainly to especially high ranking service personnel within a landed proprietor's household.

⁴⁷ It may also be worth noting that the 1921 census found 111,835 empleados “not included in other groups,” of whom 93,017 were males; many of these may have been empleados particulares rather than clerical personnel.

⁴⁸ Obviously the line between domestic servants and family members is a very ambiguous one in any case, under Mexican conditions.

⁴⁹ Computed from the U.S. Census of Population, 1960, Occupation by Industry All the other leading industrial countries appear quite close to the U.S. pattern in this regard; see the tables in Horowitz, Zymelman, and Hermstadt, Manpower Requirements.