Human Capital and Industrialization:

Evidence from the Age of Enlightenment∗

Mara P. Squicciarini Nico VoigtländerKULeuven and FWO UCLA and NBER

First version: February 2014This version: May 2014

Abstract

While human capital is a strong predictor of economic development today, its importance for theIndustrial Revolution is typically assessed as minor. To resolve this puzzling contrast, we differ-entiate average human capital (worker skills) from upper tail knowledge both theoretically andempirically. We build a simple two-sector model, where worker skills raise the productivity inboth agriculture and manufacturing, and scientific knowledge affects the entrepreneurial ability tokeep up with a rapidly advancing technological frontier. The model predicts that the local presenceof knowledge elites is unimportant in the pre-industrial era, but drives growth thereafter; workerskills, in contrast, are not crucial for growth. To measure the historical presence of knowledgeelites, we use city-level subscriptions to the famous Encyclopédie in mid-18th century France. Weshow that subscriber density is a strong predictor of city growth after 1750, but not before the on-set of French industrialization. Alternative measures of development confirm this pattern: soldierheight and industrial activity are strongly associated with subscriber density after, but not before,1750. Literacy, on the other hand, does not predict growth. Finally, by joining data on Britishpatents with a large French firm survey from 1837, we provide evidence for the mechanism: uppertail knowledge raised the productivity in innovative industrial technology.

JEL: J24, N13, O14, O41

Keywords: Human Capital, Industrial Revolution, Regional Development

∗We would like to thank Ran Abramitzky, Quamrul Ashraf, Sascha Becker, Leonardo Bursztyn, Davide Cantoni,Nick Crafts, Christian Dippel, Paola Giuliano, Avner Greif, Oded Galor, Stelios Michalopoulos, Suresh Naidu, NathanNunn, Andrei Shleifer, Enrico Spolaore, Joachim Voth, Fabian Waldinger, Ludger Woessmann, and Noam Yuchtman,as well as seminar audiences at Berkeley, Brown, IPEG Barcelona, KU Leuven, Stanford, UCLA, and the Universityof Munich for helpful comments and suggestions. We are grateful to Petra Moser for sharing her data on exhibits atthe 1851 world fair in London, to Tomas E. Murphy for sharing digitized data of Annuaires Statistiques de la France,to Carles Boix for sharing data on proto-industrialization in France, to Jeremiah Dittmar for his data on ports andnavigable rivers, to David de la Croix and Omar Licandro for their data on ‘famous’ people, and to Daniel Hicks forgeo-coded data on soldier height.

The genre of modern industrial production requires extended knowledge of mechanics, no-tion of calculus, great dexterity at work, and enlightenment in the underlying principlesof the crafts. This combination of expertise . . . has only been achieved in this [18th cen-tury] period, where the study of science has spread widely, accompanied by an intimaterelationship between savants and artisans. (Chaptal, 1819, p.32)

1 Introduction

A rich literature documents an important role of human capital for economic development in themodern world. Schooling is a strong predictor of economic growth,1 and of per-capita income atthe national and regional level.2 Both theory and evidence explain these findings by worker skillsfacilitating technology adoption and innovation.3 In contrast, the role of human capital during theIndustrial Revolution is typically described as minor. In Britain – the cradle of industrialization –education was low and inessential for economic growth (Mitch, 1993).4 On the other hand, Scan-dinavia – which was fully literate as early as 1800 – fell behind and was one of the poorest regionsin Europe. Galor (2005, p.205) points out that "in the first phase of the Industrial Revolution . . .

[h]uman capital had a limited role in the production process, and education served religious, so-cial, and national goals." At a more systematic level, cross-country growth regressions applied tothe period of industrialization lead to the conclusion that "literacy was generally unimportant forgrowth" (Allen, 2003, p.433). The stark contrast to the findings in modern data is puzzling: didthe escape from millennia of stagnation really occur without a role for one of the most importantdeterminants of modern growth – human capital?

The previous non-results are based on education or literacy as skill measures of the average

worker. This may veil the role of scientifically savvy engineers and entrepreneurs at the top ofthe skill distribution. Mokyr (2005a) stresses the importance of this "density in the upper tail",and Mokyr and Voth (2009, p.35) conclude that "the Industrial Revolution was carried not by theskills of the average or modal worker, but by the ingenuity and technical ability of a minority."Recent research on contemporaneous economies chimes in, underlining the importance of math

1See Barro (1991) and Mankiw, Romer, and Weil (1992) for early empirical growth studies, and Krueger andLindahl (2001), Barro (2001), Cohen and Soto (2007), and Hanushek and Woessmann (2008) for more recent confir-mations based on richer data.

2Gennaioli, La Porta, Lopez-de-Silanes, and Shleifer (2013). While the role of human capital as a fundamentaldeterminant of growth and development has been debated, its importance as a proximate cause, i.e., an essentialinput in the production function, is undisputed (Hall and Jones, 1999; Glaeser, Porta, de Silanes, and Shleifer, 2004;Acemoglu, Gallego, and Robinson, 2014).

3C.f. Nelson and Phelps (1966), Benhabib and Spiegel (1994), Caselli and Coleman (2006), and Ciccone andPapaioannou (2009).

4As late as 1855, at the end of the first Industrial Revolution, primary schooling enrollment in Britain was only11% (Flora, Kraus, and Pfenning, 1983). Also, modern technology typically replaced skilled workers, and the skillpremium remained unchanged until 1900 (Clark, 2005).

1

and science skills, or of entrepreneurial ability (Hanushek and Kimko, 2000; Bloom and Reenen,2007; Gennaioli et al., 2013).

In this paper, we ask whether distinguishing between upper-tail and average skills may rein-state the importance of human capital during the transition from stagnation to growth. Answeringthis question hinges on a historical proxy for the thickness of the upper tail, i.e., the presence ofknowledge elites.5 We use a novel measure from the eve of industrialization in mid-18th centuryFrance: subscriptions to the Encyclopédie – the cornerstone of Enlightenment, representing themost important collection of scientific and technological knowledge at the time. One of the pub-lishers kept a list of all (more than 8,000) subscriptions to the most prominent edition.6 Basedon this information, we calculate subscriber density for almost 200 French cities and use it as aproxy for the local concentration of knowledge elites. Figure 1 shows that regions with high andlow values are relatively evenly distributed and often immediately adjacent. In addition, subscriberdensity is uncorrelated with literacy rates in the same period (see Figures 1 and A.4), allowing usto differentiate between average and upper tail skills.

To guide our analysis, we build a simple two-sector model of industrialization. We assumethat upper tail knowledge enables entrepreneurs in manufacturing to keep up with advances at thetechnology frontier. Thus, in the spirit of Nelson and Phelps (1966), advanced skills are particularlyuseful when technological progress is rapid. The main prediction is that income and industrialemployment grow faster in regions with a thick knowledge elite. However, this relationship holdsonly after growth becomes rapid; in pre-industrial times, upper tail knowledge is inessential foreconomic growth. In contrast, average worker skills enter production in both sectors in the standardlabor-augmenting way. They thus affect income in the cross-section but not growth over time.In other words, while worker skills raise labor productivity for a given technology, upper tailentrepreneurial skills drive growth by changing technology.

We collect a host of outcome variables to test the model predictions. First, our main analysisuses city size in France between 1400 and 1850. We find that subscriber density was stronglyassociated with city growth after 1750 – when French industrial growth began – but not before thatdate. Figure 2 illustrates this finding, using a consistent set of cities that are observed over ourfull sample period. The coefficient on subscriber density is almost six times larger after 1750 – it

5Following Mokyr (2005a), we use a broad definition of "upper tail knowledge": it reflects an interest in scientificadvances, motivated by the Baconian notion that knowledge is at the heart of material progress. This concept comprisesinnovative and entrepreneurial capabilities in adopting and improving new technology, but also lower access costs tomodern techniques; it is thus compatible with Mokyr’s notion of (economically) ‘useful knowledge’. When referringto the local presence of people embodying such knowledge, we use the term "knowledge elite".

6The names of individual subscribers survived only in a few cases – where they did, a substantial share of sub-scribers were progressively minded and scientifically interested noblemen, administrative elites, and entrepreneurs.

2

rises from 0.037 to 0.218, and the difference is highly significant, with the two 95% confidenceintervals not overlapping. Second, we use soldier height as a proxy for income at the French de-partment level. We find that after 1820, soldier height is significantly associated with encyclopediasubscriptions, while this relationship disappears before 1750. Third, mid-19th century census datafor almost 90 French departments reveal that those with higher subscriber density had significantlyhigher disposable income, industrial wages, and industrial employment. In addition, the wage ef-fect is strongest within modern industries (such as textiles), and it is muted for sectors that saw lessrapid technological change before 1850, such as food processing and agriculture. For all outcomevariables, we also confirm the model prediction that literacy or schooling (proxying for averageworker skills) are positively associated with development in the cross-section, but do not explaingrowth.

When interpreting our results, we do not argue that the encyclopedia caused scientific knowl-edge at the local level. In fact, knowledge elites were present prior to 1750, and their spatialdistribution was stable over time. We show that early scientific societies are a strong predictor ofsubscriber density. The same is true for the share of Huguenots in 1670 – the suppressed Protes-tant minority that is typically associated with the French knowledge elite (Scoville, 1953; Hornung,2014). In addition, locations with higher subscriber density brought up a larger proportion of fa-mous people in scientific professions between the 11th and 19th centuries, and they also presentedmore innovations (per capita) at the 1851 London World Fair. In sum, there is compelling evidencethat subscriber density reflects a stable spatial distribution of knowledge elites.7 Our argument isthat these elites began to foster growth when knowledge became economically ‘useful’, and tech-nological progress became rapid. This follows Mokyr’s (2002; 2005b) seminal work on the risingimportance of upper tail knowledge during the period of Industrial Enlightenment.

To support our interpretation, we discuss detailed historical evidence that connects scientificknowledge to entrepreneurship and technological improvements, both via innovation and via theadoption of modern techniques. Modern technology was often imported from Britain, and aninterest in science helped entrepreneurs both to learn about these techniques in the first place, andto understand the underlying principles needed to implement and run them.8 We further support

7In this dimension, our empirical analysis is similar in spirit to Voigtländer and Voth (2012), who take the spatialpattern of anti-Semitism as given and use historical persecution of Jews to measure it. The spatial dispersion ofscientific activity in early modern Europe is well-documented (Livingstone, 2003). We take this pattern as given, andargue that subscriber density is a powerful proxy to capture it.

8French entrepreneurs often also sought advise from British technicians. In the early 19th century, 10,000-20,000English artisans – many of them mechanics – worked in France (Kelly, Mokyr, and Ó Grada, 2014). The same authorsalso point out that "the French learned quickly, and as soon as local workmen had acquired the basic skills, the seniorBritish operative became more of a rarity." However, it is important to note that France did not merely adopt Britishtechnology. Instead, it became an important source of innovations itself: "Technological progress became indigenous,

3

our argument by providing systematic evidence for the mechanism from a survey of more than14,000 French firms in 1837. Based on sector-specific British patent data, we split these firms into‘modern’ (innovative) and ‘old’. We show that firms in ‘modern’ (but not in ‘old’) sectors weremuch more productive in regions with higher subscriber density, even after controlling for sectorfixed effects. This supports the interpretation that upper tail knowledge favored the adoption andefficient operation of innovative industrial technology.

Importantly, we do not claim that upper tail knowledge was necessarily a fundamental driverof economic growth during industrialization.9 The spatial variation in scientific knowledge may bedue to deeper determinants such as culture, institutions, or geography that may also affect growthvia channels other than human capital. While we ultimately cannot exclude this possibility, weconstruct a host of control variables to account for it. In a similar vain, we control for numerouspotential confounding factors that may be associated with both encyclopedia subscriptions andgrowth. For example, we control for total book sales per capita at the city level in the 18th century.While these are strongly correlated with encyclopedia subscriber density, they do not affect growth.We also shed light on the potential role of institutions, using the fact that the French Revolutionoccurred approximately at the mid-point of our main period of analysis (1750-1850). We find thatsubscriber density was strongly associated with city growth under both regimes, despite their rad-ically different institutions.10 Our results are also robust to controlling for other factors, includinggeographic characteristics, proto-industrial activity, the density of noble families, possible effectsof the ‘Reign of Terror’, and early book printing. In sum, the historical evidence in combinationwith our empirical findings makes it hard to imagine that (upper tail) human capital did not play amajor role during industrialization – at the very least as a proximate driver.

Our paper is related to a large literature on the transition from stagnation to growth (for anoverview see Galor, 2011), and in particular to the role of human capital during industrializa-tion. For England – the technological leader – the consensus view is that formal education did notcontribute to economic growth (Mokyr, 1990; Mitch, 1993; Crafts, 1996; Clark, 2005). For thefollower countries, the evidence is mixed. O’Rourke and Williamson (1995) and Taylor (1999)conclude from country-level cross-sectional and panel analyses that human capital was not a cru-

built in to the economy, so that ... France became at mid-[19]century a centre of invention and diffusion for moderntechnologies" (Crouzet, 2003, p.234).

9The distinction of fundamental vs. proximate determinants of growth goes back to North and Thomas (1973). Asdiscussed above, evidence abounds that human capital is a proximate driver of modern development, but whether it isalso fundamental is debated (see the references cited in footnote 2).

10In addition, after the 17th century France was a centralized absolutist state, allowing for relatively little localvariation in institutions (Braudel, 1982; DeLong and Shleifer, 1993). Our analysis is thus less affected by the typicallimitations of cross-country studies. We also control for regions where the king exerted particularly strong control –the pays d’élection – and find that these differed neither in subscriber density nor growth.

4

cial driver of catch-up in the 19th century. In contrast, Becker, Hornung, and Woessmann (2011)document that elementary education predicts employment in metals and other industries, but notin textiles in 19C Prussia. This is in line with O’Rourke, Rahman, and Taylor (2008), who empha-size that industrial innovation in sectors such as textiles was initially unskill-biased, reducing thedemand for skilled workers.11 We shed new light on this debate by distinguishing between averageand upper-tail skills, following Mokyr (2005b) argument that the expansion and accessibility of‘useful knowledge’ during the period of Enlightenment was a cornerstone of industrial develop-ment.12 Our paper also relates to a literature showing that book production had a positive impact onpre-industrial economic development (Baten and van Zanden, 2008; Dittmar, 2011).13 Our mainexplanatory variable, encyclopedia subscriber density, offers two advantages over printing loca-tions or local book production. First, it is a more precise measure, identifying readers within thenarrow category of scientific publications.14 Second, subscriptions measure the local demand forknowledge, rather than the supply of books from printing locations.

Relative to this literature we make several contributions. To the best of our knowledge thispaper is the first to empirically differentiate between average worker skills (literacy/schooling) andupper tail knowledge during the early days of industrialization. Along this dimension, our studyis the first to provide systematic evidence for Mokyr’s (2005b) hypothesis about the importanceof ‘useful knowledge’ for industrialization. We also shed light on the mechanism, showing thatupper tail knowledge probably fostered growth by raising firm productivity in modern, innovativeindustries. Finally, we show that basic education was related to economic development in thecross-section, but – in contrast to upper tail knowledge – not to growth.15

The paper is organized as follows: Section 2 discusses the historical background of industri-alization in France with a particular emphasis on the intersection of science and entrepreneurship.

11This led to the famous incident of the Luddites – skilled textile workers – wrecking modern machinery that wasoperated by unskilled labor.

12Kelly et al. (2014) also emphasize the importance of highly skilled, technically capable individuals. In the con-temporaneous context, Hanushek and Woessman (2012) show that the share of cognitively high-performing studentsis strongly associated with growth, independent of basic literacy.

13Baten and van Zanden show for a panel of 8 European countries over the period 1450-1850 that wage growth wasmore rapid where book printing was more pronounced. Remarkably, however, after 1750 countries with particularlyhigh book printing, such as Sweden or the Netherlands, saw a decline in real wages. This suggests that the observedpattern is driven by variation before the onset of industrialization. Cantoni and Yuchtman (2014) also document earlyeffects, showing that universities fostered commercial activity in medieval times.

14Total book production, in contrast, contains books from cooking manuals to religious works. For example, booksabout natural science, math, and engineering account for less than 5% of overall book sales by the large publishinghouse STN in the late 18th century, while 70% of all sales occurred in Belles lettres (e.g., romans and poetry), history,and religion.

15The cross-sectional result is in line with Hornung (2014), who shows that the migration of skilled Huguenots toPrussia had strong positive effects on the productivity of textile firms around 1800, i.e., before Prussia industrialized.

5

We also discuss encyclopedia subscriptions as a proxy for the presence of knowledge elites. Sec-tion 3 presents a model of industrialization that illustrates our main argument. In Section 4 wedescribe the data, and Section 5 presents our empirical results. Section 6 concludes by discussingthe implications of our findings for the literature on human capital and development.

2 Historical Background: Industrialization and Upper Tail Knowledge

In this section, we discuss the role of upper tail knowledge during the French industrialization. Wealso provide background on the Encyclopédie, and discuss why its subscriptions are a good proxyfor the local presence of knowledge elites.

2.1 Industrialization in France

Starting in the mid-18th century, France experienced significant economic growth; its industrialoutput more than doubled during this century (Rostow, 1975), and machines were introduced inthe main industrial sectors – textiles and metallurgy (Ballot, 1923). By the mid-19th century,France had become "a centre of invention and diffusion of modern technology" (Crouzet, 2003,p.234).16

At the micro level, French firms were smaller than their English counterparts: most of themwere family owned; in 1865 they had 9.5 workers on average (Verley, 1985), and in 1901, 71percent had no hired employees (Nye, 1987). However, the small size of French firms is not a signalof inefficiency or lack of entrepreneurial skills, since they suited the economic and technologicalconditions of the time and "would not have stood to gain much in efficiency by being larger" (Nye,1987, p. 668). For example, firms would often not expand because the political environment madeinvestments risky. Overall, France had its own path to industrialization, which was different, butnot inferior to the British one (Crouzet, 2003).

British know-how reached France via several channels: First, scientific reports published bylearned societies, an intense correspondence among "industrially minded" people in the two coun-tries, as well as industrial spies sending regular reports on English technology drove the rapidtransfer of industrial knowledge (Mokyr, 2005b; Horn, 2006). Learned circles played a centralrole spreading this knowledge. Second, there were many initiatives to import English machines,and third, thousands of British workers were hired by French producers with the specific aim to

16Figure A.1 in the appendix shows that GDP per capita was relatively stable until approximately 1750, and thenstarted to grow steadily. Timing the French industrial takeoff is difficult, as it lacks a clear structural break (Roehl,1976); the predominant "moderate revisionist" view describes steady growth until the mid-19th century, only inter-rupted during the two decades after the revolution in 1789 (Crouzet, 2003). Growth was particularly strong between1815 and the 1840s, then slowed down during a depression around 1875-95, which was followed by another period ofrapid growth (Belle époque). On average, French GDP per capita grew as fast as its British counterpart over the period1820-1913, although French population grew at a markedly lower rate (Maddison, 2001).

6

gain access to technical knowledge (Horn, 2006). Finally, the state and provincial governmentsalso played an important role in French industrialization, by supporting scientific institutions, bybringing together entrepreneurs and scientists, but also by fostering adoption of machines and ex-pertise from abroad. These policies were put into practice at the national and local level by acommercial, industrial, and scientific elite (Chaussinand-Nogaret, 1985; Horn, 2006).

2.2 Enlightenment, Upper Tail Knowledge, and the Encyclopédie

The Enlightenment was a period of intellectual and cultural revolution in Western history, stretch-ing from the late 17th throughout the 18th century, stressing the importance of reason and science,as opposed to faith and tradition. The ‘Industrial Enlightenment’, which "bridges between theScientific and Industrial Revolution" (Mokyr, 2005a, p.22) was a fundamental part of it.17 The en-lightenment furthered growth in two ways: through the expansion of propositional knowledge withpractical applications, and through a reduction of access costs to existing knowledge. Despite itsefforts to popularize and spread knowledge, the Enlightenment never became a mass movement.While confined to a small elite, it still played a crucial role in fostering industrial development andeconomic growth.

The Encyclopédie

In the culturally vibrant atmosphere of 18th century France, Diderot and d’Alembert launched theambitious project of the Great Encyclopédie – the "most paradigmatic Enlightenment triumph"(Mokyr, 2005b, p.285). Following Lord Francis Bacon’s conceptual framework, Diderot’s objec-tive was to classify all domains of human knowledge in one single source, easily accessible toeverybody. The focus was on knowledge derived from empirical observation, as opposed to su-perstition.18 While mercantilistic ideas were still widespread, and artisans and guilds kept secrecyover knowledge, Diderot and the Philosophes believed that scientific knowledge should not be aprivate good, confined to a few people, but disseminated as widely as possible (Mokyr, 2005a).

The Encyclopédie went through several editions and reprints. The government initially refusedto allow official sales, and most copies went to customers outside France. Correspondingly, the first(the Encyclopédie by Diderot and d’Alembert) and the second (the Geneva reprint) editions soldtogether only 3,000 copies in France. Moreover, the first two editions were luxury items that did notpenetrate far beyond the restricted circle of courtiers, salon lions, and progressive parlementaires.

17The roots of the Industrial Enlightenment go back to the ideas of Lord Francis Bacon, who "aimed at expandingthe set of useful knowledge and applying natural philosophy to solve technological problems and bring about economicgrowth" (Mokyr, 2005b, p.326).

18The articles of the Encyclopédie often resulted provocative to authorities. For example, in the engraving of theTree of Knowledge, theology is considered a simple branch of Philosophy. This led to the imprisonment of the editorand the publishers; the Pope condemned the Encyclopédie, and it was suppressed by a royal decree (Vogt, 1982).

7

This changed radically with the Quarto (1777-1780) and the Octavo editions (1778-1782).19 Thesebecame wide-spread throughout the country and largely reached middle class budgets (Darnton,1973). Since our proxy for local knowledge elites is based on subscriptions to the Quarto, wediscuss this edition in more detail.20

The Quarto edition

The Quarto edition of the Encyclopédie is particularly useful as a proxy for local knowledge elitesfor several reasons. It represents the turning point when the Encyclopédie moved to a phase ofdiffusion of Enlightenment on a massive scale. The Quarto was designed to be affordable for mid-dle class readers. Correspondingly, its format was smaller than the luxurious Folio of Diderot, thequality of the paper poorer, and the price lower. The publishers described their price discrimina-tion strategy as follows: "The in-folio format will be for ‘grands seigneurs’ and libraries, while thein-quarto will be within the reach of men of letters and interested readers [‘amateurs’] whose for-tune is less considerable."21 This strategy proved extremely successful: the Quarto had the highestsales in France among all editions. The low cost of the Quarto rendered budget constraints lessimportant, so that scientific interest, rather than deep pockets, determined subscriptions.

Crucially for our study, one of the publishers, Duplain, secretly kept a list of subscriptions,which survived in the archives of the Société Tipographique de Neuchâtel (STN). This list containsthe name of booksellers (but not subscribers), their city, and the number of sets they purchased forretail among their local clients. Darnton (1973) provides this list, which comprises a total of8,011 subscriptions, out of which 7,081 were sold in France – in 118 cities.22 The list containsall subscriptions to the Quarto in France, so that we can safely assume that cities which are notlisted had zero subscribers. Subscriptions were not confined to major cities; instead, they weredistributed across the whole French territory (see Figure 1).

19The names Quarto and Octavo refer to the printing format: A Quarto sheet is folded twice, creating four leaves;an Octavo is folded three times, creating eight leaves. The central figure behind the Quarto edition was the Frenchentrepreneur Charles-Joseph Panckoucke (1736-1798), who had bought the plates of the Encyclopédie from the orig-inal publishers, together with the rights to future editions. Administrative obstacles in selling the Quarto appear tohave been minor. While the Encyclopédie was officially illegal until 1789, the government had relaxed its censorship.In addition, Panckoucke had good connections with the government, and did not hesitate to lobby and bribe publicofficials (Vogt, 1982).

20The Quarto edition comprised 36 volumes of text and 3 volumes of illustrative plates. Since these were typicallynot delivered in one chunk, readers of the Encyclopédie are commonly referred to as ‘subscribers’.

21Société Tipographique de Neuchâtel (the publisher) in a letter to Rudiger of Moscow, May 31, 1777; cited afterDarnton (1973, p.1349). The Quarto cost only one fifth of the first original folio. While unaffordable to lower socialclasses and workers, the lower price put it well within the reach of the provincial middle class. According to Darnton’(1973) calculations, a Quarto cost the equivalent of 10 weeks of income of a prosperous parish priest, and about 5weeks of income of a provincial bourgeois living off his rents.

22Lyon with 1,078 and Paris with 487 subscriptions are at the top of this list; at the opposite end of the spectrum,there are 22 towns with fewer than 5 subscriptions.

8

Subscriptions to the Quarto edition and local knowledge elites

Does a higher frequency of subscriptions at any one location reflect a broader interest in uppertail knowledge?23 Characteristics of individual subscribers are key to shed light on this question.While Duplain’s secret list does not allow for a systematic analysis because it only contains thenames of local booksellers, individual information exists in a few cases. For Besançon, a listof individual subscribers has survived. Darnton (1973, p.1350) summarizes these by professionand social status: 11% belonged to the first estate (clergy) and 39% to the second estate (nobil-ity); the remaining one half of the 137 subscribers belonged to lower social ranks, including thebourgeoisie.24 Professionals, merchants, and manufacturers account for 17% of the total (23 sub-scriptions). Thus, an important share of subscribers can be directly identified as economic agents(from lower ranks) involved in the French industrialization. Their share is likely a lower bound forthe importance of the encyclopedia in the business community, because many subscribers in theupper class (‘nobility’ – the largest single category) were also active businessmen (Horn, 2006).25

For example, Chaussinand-Nogaret (1985, p.87) argues thatOver a whole range of activities and enterprises nobles, either alone or in associationwith members of the greater business bourgeoisie, showed their dynamism, their taste forinvention and innovation, and their ability as economic leaders: ... their ability to directcapital..., to choose investments according to their productiveness and their modernity,and ... to transmute the forms of production into an industrial revolution.

It is, however, important to note that only a progressive subset of the nobility was involved in indus-trial activities.26 The same subset was also heavily engaged in the Enlightenment (Chaussinand-Nogaret, 1985, p.73). This underlines Mokyr’s (2005b) view that upper tail knowledge and indus-trialization were closely entwined.

In addition, many subscriptions went to high public officials and parlementaires – about 28%in Besançon (Darnton, 1973, p.1350). Enlightened elites in the provincial administration wereoften involved in fostering local industrialization. For example, in Rouen and Amiens, they es-

23For example, it is possible that wealthy people merely bought the encyclopedia to decorate their bookshelves.However, according to Darnton (1973, p.1352), if anything, the opposite was probably the case: "far more peoplemust have read the Encyclopédie than owned it, as would be common in an era when books were liberally loaned andwhen ‘cabinets litteraires’ were booming."

24This may be a lower bound for other locations. The second estate is probably over-represented in this sample –Besançon was a garrison town, and almost half of the subscriptions in this category went to noblemen in the army.

25This reflects the revisionist view that replaced earlier – often Marxist – interpretations of the nobility exemplifyingan aristocratic tyranny of arrogance and decadence. An impartial reading of the historical account shows that "noblesof the eighteenth century had been as modern and progressive as anyone." (Smith, 2006a, p.2).

26As Chaussinand-Nogaret (1985, p.90) puts it: "In the economic sphere, ...it is clear that the whole of nobility wasnot involved, but only the part that can be considered its natural elite, ... because of its ... openness to the progressivetendencies of the age."

9

tablished the Bureau of Encouragement that gathered businessmen, manufacturers, local savants,and provincial authorities in an effort to assist technological advance (Horn, 2006, p.81). Finally,the Encyclopédie – and especially the Quarto – also reached non-subscribers in the lower ranksof society, via indirect access (Roche, 1998). Organized lectures, symposia, and public experi-ments were booming in France during the Enlightenment, and public readings of the Encyclopédiewere organized by scientific societies, libraries, lodges, and coffeehouses (Darnton, 1979; Mokyr,2005b).27

In sum, the encyclopedia had a broad spectrum of readers from the knowledge elite that wereboth directly and indirectly involved in industrialization. This supports both our use of subscriberdensity as a proxy for local upper tail knowledge, and the hypothesis that this knowledge wascrucial in fostering economic growth. Next, we discuss a number of concrete examples for howscientific knowledge affected entrepreneurial activity and technological growth in France duringits industrialization.

2.3 Scientific Knowledge and Entrepreneurship

There are many prominent examples for the link between upper tail knowledge and entrepreneur-ship 18th and 19th century France. The father of the chemical revolution of the 18th century,Antoine Lavoisier (1743-1794), was educated in the spirit of enlightenment, and fascinated by theDictionnaire de Chymie (published in 1766). He worked on several applied problems such as therole of oxygen in combustion, street lighting, and he predicted the existence of silicon. AlexandreVandermonde (1735-1796), a mathematician attracted to machinery and technology, fostered thefirst major industrial application of Lavoisier’s chemistry in iron production (Mokyr, 2005b). Sim-ilarly, the chemist Claude Louis Berthollet (1748-1822) experimented with chlorine, discoveringnew methods for bleaching. His results were both published in scientific journals and applied inmost of the leading textile-manufacturing firms of France. These discoveries led the contempo-raneous observer Robert O’Reilly (an Irishman living in Paris) to declare in 1801: "a complete

revolution in the art of bleaching...we have finally arrived in an époque where science and indus-

trial arts, reinforcing each other, rapidly spill indefinite improvements" (cited after Musson andRobinson, 1969, p.253; our translation). Another example is the Duke d’Orléans, who set up a

27Certainly, reading or hearing about a new technology was not sufficient to be able to adopt and operate it. However,scientific publications and lectures made technological know-how available on a large scale, breaking the exclusivetransmission from master to apprentice (Mokyr, 2005b). The details needed for actual adoption of new technologieswere then typically found elsewhere, such as embodied in ‘imported’ British experts. For example, Fox (1984, p.143)describes the case of the British engineer Job Dixon, who came from Manchester to Cernay in southern Alsace in 1820and joined the firm Risler frères. This firm had been founded only two years earlier as the first machine-builder in theregion. It subsequently became the main supplier of the latest spinning and weaving machinery for the region, servingalso as a training ground for French engineers.

10

soda-making facility together with the chemist Nicolas Leblanc (1742-1806). He invested in sev-eral textile firms, adopting modern machinery from Britain, and introduced steam-engines intocotton spinning in France (Horn, 2006; Chaussinand-Nogaret, 1985).

In other cases, the same person or family was involved in both scientific research and industrialactivities. For instance, Jean-Antoine Chaptal, a famous chemist, successful entrepreneur, and po-litical figure considered science to be inseparable from technology, and the key to foster industrialdevelopment. He was well-connected in the French network of savants, which entertained an in-tensive exchange with international scientists such as James Watt, and stimulated the applicationof science to industry in France. Chaptal pursued this cause both as a public figure and as a privateentrepreneur. As a public official, he created favorable economic and bureaucratic conditions forentrepreneurs, for example by founding the Conseils d’Agriculture, des Arts et Commerce, and theSociété d’Encouragement pour l’Industrie National, where scientists, industrialists, and bureau-crats were brought together. He also subsidized promising artisans, engineers, and industrialists,and gave pubic lectures on chemistry and experimental physics. As a private entrepreneur, Chaptalbuilt the largest factory for chemical products in France (Horn and Jacob, 1998).

In many cases, entrepreneurial dynasties with scientific spirit came from the Protestant (Huguenot)minority. For example, the Koechlin and Dollfus families in Mulhouse were closely related byintermarriage and descendent directly from the famous mathematician Bernouilli. They ran pros-perous firms in cotton or wool spinning and founded the first cloth-printing firm in France. Somemembers also entered other industrial businesses, such as the manufacturing of textile machiner-ies, locomotives, and railroad equipment. Their dynasties kept marrying other scientific families(such as the Curies and the Friedels), and produced successful scientists themselves: for instance,Daniel Dollfus-Ausset (1797-1879) was a chemist who made major innovations in bleaching andsimultaneously ran his own textile firm. The Koechlin and Dollfus families were also co-foundersof the Société industrielle de Mulhouse, which promoted technological progress via conferencesand publications (Hau, 2012; Smith, 2006b).

Even in raw material production like silk growing, scientific knowledge played an importantrole. The silkworm is extremely sensitive to cold, heat, and drafts, which rendered its adoptionin France difficult. The entrepreneur Camille Beauvais ran a silk farm near Paris with the supportof distinguished chemist d’Arcet. Their methods raised worm productivity enormously, more thandoubling output per egg, and allowing for four harvests per year (Barbour and Blydenburgh, 1844,p.39). Beauvais trained young growers at his farm, who spread his methods throughout France andeventually to the United States. His work was promoted and advertised by scientific organizations,such as the Société d’Encouragement pour l’Industrie Nationale.

Our argument also applies contrariwise: the absence of advanced knowledge typically came

11

along with the use of backward technology. The French iron industry is one example; Dunham(1955, p.119) observes that "a typical ironmaster knew little or nothing about science... He was

as ignorant of planning, routing, and economics as of metallurgy, and carried on his business in

the manner of his fathers, with little knowledge of what went on outside his own district." Cor-respondingly, when puddling – a crucial process to produce high-grade bar iron – was inventedin England in 1784, the vast majority of French iron producers ignored it until well into the 19thcentury. However, there were some exceptions, and the evidence suggests that upper tail knowl-edge played a crucial role: a small minority of "the ablest metallurgists in France ... [brought] the

puddling process...successfully from England to France and introduced [it] almost simultaneously

at several widely separated establishments run by metallurgists of outstanding ability (Dunham,1955, p.128).

These examples illustrate that the effect of knowledge elites on local industrial developmentcould go via the dialogue between scientists and entrepreneurs, via scientifically savvy public of-ficials supporting entrepreneurship, and via members of the elite themselves operating businesses.

3 Model

In this section, we provide a simple model that connects the historical evidence discussed abovewith the empirical analysis that follows below. The model distinguishes between the skills ofworkers and upper tail knowledge of entrepreneurs. We present a mechanism where upper tailknowledge enables local entrepreneurs to keep up with technological progress at the frontier. Themodel delivers predictions for the role of upper tail knowledge vs. worker skills for income andeconomic growth before and after industrialization.

The model features n = 1, ..., N regions with given land endowment. In each region, there ismass L >> 1 of workers who supply one unit of labor inelastically in each period. Workers skillshn vary across regions. In addition, there are i ∈ [0, 1] entrepreneurs who produce intermediategoods in manufacturing. A share sn of entrepreneurs in region n disposes of upper tail (scientific)knowledge.28

There is no saving, so that all income is consumed in each period. In any given period, workersoptimally choose between working in two sectors: agriculture (A) and manufacturing (M ). Wekeep the model tractable by following Hansen and Prescott (2002) in assuming that agriculturaland (both) manufacturing goods are perfect substitutes.29 In addition, we assume that workers and

28To distinguish between their effects on development, we assume that hn and sn are independently distributedacross regions. This also reflects the empirical evidence presented in Figure A.4 in the appendix, which shows that ourhistorical proxies for the two types of human capital, literacy and subscriber density, are uncorrelated across Frenchregions.

29Thus, our results are driven only by the supply side. Introducing demand-side factors – such as increasing relative

12

entrepreneurs are immobile, operating within their region of origin. Thus, wages and employmentshares can differ across regions.

Sector-specific wages in each region depend on both types of knowledge. First, average workerskills affect the efficiency of production in both sectors, but to a lesser degree in agriculture. Sec-ond, highly skilled entrepreneurs can raise productivity in manufacturing. Our approach followsthe tradition of Nelson and Phelps (1966), who argued that the stock of human capital drives growthby affecting the ability to innovate or to adopt new technology. Following the historical evidencepresented above, we apply this argument to the upper tail of human capital: we assume that scien-tific knowledge is particularly useful when technological progress is rapid. For simplicity, we takethe aggregate technology frontier A as given. Following Hansen and Prescott (2002) and Kellyet al. (2014), we assume that A grows exogenously over time, and that this reflects the expansionof useful knowledge during the period of Industrial Enlightenment. All relevant cross-sectionalpredictions of our model can be derived in partial equilibrium, taking the price of output (in bothsectors) as given and normalizing it to P = 1.

3.1 Production

Each worker supplies one unit of labor and chooses a sector of employment at the beginning of aperiod. Technology in all sectors exhibits constant returns, so that the scale of production is notimportant. We denote total labor in sector j ∈ {A,M} in region n by Lj,n. In the following, wecharacterize the production technologies used by the two sectors. Agricultural output in region n

is given byYA,n = AAh

βAn XαA

n L1−αAA,n , (1)

where X is land endowment, αA is the share of land in production, and βA reflects the sensitivityof agricultural productivity with respect to worker skills.30 We assume that there are no propertyrights to land, so that wages in agriculture are given by the average product yA,n = YA,n/LA,n:31

wA,n = AAhβAn

(Xn

LA,n

)αA

= AAhβAn

(xn

lA,n

)αA

, (2)

where lA,n = LA,n/L is the agricultural labor share, and xn is land per worker in region n. Notethat agricultural wages increase if the labor share in agriculture declines, because this leaves moreland for each remaining peasant. Thus, growth in manufacturing can indirectly raise wages in

demand for manufacturing as income rises – would accelerate industrialization.30In growth models and development accounting, h typically multiplies L directly, reflecting the average impact of

schooling on productivity via Mincerian returns (c.f. Bils and Klenow, 2000). By using different βj for j ∈ {A,M},we allow these returns to vary across sectors, i.e., we allow for sector-specific returns to worker skills.

31This assumption does not affect our results, but it simplifies the analysis.

13

agriculture.Our modeling of the manufacturing sector builds on Acemoglu, Aghion, and Zilibotti (2006).

The setup embeds a role for entrepreneurial skills in the manufacturing production process; it alsohas the advantage that it reduces to a simple aggregate production function. The final manufac-turing good is produced under perfect competition by firms that use labor and a continuum ofintermediates as inputs. The technology exhibits constant returns, so that we can focus on aggre-gate output in manufacturing, produced by a representative firm in the final sector:

YM,n = ξ ·(∫ 1

0

AM,n(i)1−αM zn(i)

αMdi

)(hβMn LM,n

)1−αM (3)

where ξ is a constant, zn(i) is the flow of intermediate good i in final production, LM,n is totallabor in manufacturing, βM is the sensitivity of manufacturing production with respect to workerskills, and αM denotes the share of intermediates in final production. Intermediates are producedby entrepreneurs under monopolistic competition. Each entrepreneur i ∈ [0, 1] produces a specificintermediate i by transforming one unit of the final good into one unit of the intermediate. Thus,the marginal cost is identical for all entrepreneurs. However, the productivity with which interme-diates enter final production, AM,n(i), differs across entrepreneurs i.32 We study the evolution ofproductivity as a function of entrepreneurial skills below.

Solving the entrepreneurs’ optimization problem yields a simple expression for aggregate man-ufacturing output (see Appendix A.1 for detail):

YM,n = AM,nhβMn LM,n, with AM,n =

∫ 1

0

AM,n(i)di (4)

Thus, aggregate manufacturing productivity AM,n is a simple linear combination of individualentrepreneurial efficiencies i ∈ [0, 1]. Finally, the first order condition of (3) with respect to LM,n

implies that a share 1 − α of final output is paid to labor. Combining this with (4) thus yields thewage rate in manufacturing:

wM,n = (1− αM)AM,nhβMn (5)

Finally, we assume βM > βA, i.e., that manufacturing production is relatively more sensitive with

32Effectively, higher AM,n(i) raises the demand for intermediate i in final production, but it does not affect the unitcost of i. This approach (following Acemoglu et al., 2006) ensures tractability. It can be motivated, for example, byinterpreting AM,n(i) as the quality of intermediate i, so that more productive entrepreneurs produce higher qualityintermediates at the same marginal cost. Note, however, that productivity can still be interpreted as the standardquantity-related concept: in equilibrium, entrepreneurs with high AM,n(i) sell more and make higher profits, so thatour setup is akin to an alternative that loads productivity differences on marginal costs.

14

respect to worker skills.33 This assumption matters for cross-sectional predictions in our model,but it does not affect growth. It is motivated by the historical observation that the predominantlysmaller-scale manufacturing practiced in France was typically performed by relatively skilled arti-sans. In the following, we study the evolution of productivity, where entrepreneurial skills play acentral role.

3.2 The Evolution of Productivity

The technological frontier at time t is given by At, and it grows at an exogenous rate γA,t.34 Theaggregate frontier affects the productivity of individual entrepreneurs i at locations n. This isreflected by the productivity process

AM,n,t(i) = ηAt + (1− η)(1 + τ(i) γA,t

)AM,n,t−1 (6)

where η ∈ (0, 1), and τ(i) reflects two types of entrepreneurs: τ(i) = 1 for those with uppertail (scientific) knowledge, and τ(i) = 0 for the remainder. AM,n,t−1 is aggregate manufacturingproductivity at location n in the previous period (described in more detail below). To interpret (6),consider first an entrepreneur with τ(i) = 0. In this case, η > 0 guarantees that at least someinnovation trickles through, and entrepreneurial productivity is the closer to the frontier the largeris η. We refer to this mechanism as (passive) catchup.

Next, consider highly skilled entrepreneurs with τ(i) = 1. These also experience catchup, butin addition they actively improve their productivity, by the rate γA,t relative to the initial local pro-ductivity AM,n,t−1 . We refer to this process as ‘knowledge effect’ – highly skilled entrepreneursimprove local technology by keeping up with technical progress at the frontier. This reflects severaldimensions of the historical evidence discussed in Section 2. First, more scientifically savvy en-trepreneurs were more likely to know about the existence of new technologies, which reduces theirsearch costs and raises the likelihood of adoption.35 Second, they could operate modern technol-ogy more efficiently because of a better understanding of the underlying processes. Third, scien-

33See Dunham (1955, pp. 183–186) for a discussion of the role of worker skills during industrialization in France.Within manufacturing, βM may have decreased during the first industrialization: switching from artisan manufacturingto large-scale standardized factory production resulted in lower demand for skilled workers, culminating in the Ludditemovement in England (O’Rourke et al., 2008). However, large-scale production was very limited in France prior to1850 (Dunham, 1955; Nye, 1987). Thus, on average βM > βA is likely to hold throughout our period of analysis.

34Since we study the cross-sectional effects of upper tail knowledge, we take the evolution of At as given. Wethus abstract from the feedback mechanism in Unified Growth Theory whereby human capital drives aggregate tech-nological progress (Galor, 2011). At the local level, however, our approach allows for upper tail skills to accelerateproductivity growth.

35As Mokyr (2000, p.30) put it: "Of course I do not argue that one could learn a craft just from reading an ency-clopedia article (though some of the articles in the Encyclopédie read much like cookbook entries). But ... once thereader knew what was known, he or she could look for details elsewhere."

15

tific knowledge made further innovative improvements more likely.36 Importantly, the ‘knowledgeeffect’ is the stronger the higher is γA,t. This reflects the argument by Nelson and Phelps (1966)that knowledge – here in the form of its upper tail – is particularly useful in periods of rapidtechnological change.

We now turn to the evolution of the aggregate manufacturing productivity at location n, AM,n,t.This term corresponds to the average entrepreneurial productivities, as given by (4). Recall that ateach location n, there is a share sn of highly skilled entrepreneurs. Thus, integrating (6) over allentrepreneurs i ∈ [0, 1] yields:

AM,n,t = ηAt + (1− η)AM,n,t−1

(1 + sn · γA,t

)(7)

This equation illustrates three forces that drive manufacturing productivity at location n: First,passive catchup with the frontier that depends on η. Second, the ‘knowledge effect’, which islarger for regions with a higher proportion of highly skilled entrepreneurs (sn), and larger whentechnological progress γA,t is rapid. Third, there is also a spillover effect of entrepreneurs withupper tail knowledge: they raise AM,n,t, which is then reflected as AM,n,t−1 in (6) in the followingperiod. Note that this benefits both entrepreneurs with and without scientific knowledge. Our setupalso ensures that AM,n,t ≤ At, which holds with equality in regions with sn = 1. In other words,a region where all entrepreneurs have scientific knowledge will always be at the technologicalfrontier.

Finally, we specify the productivity process in agriculture. We assume that upper tail knowl-edge is not important in this sector.37 However, we assume that some technologies from the frontier‘trickle through’ to agriculture, as well.38 For simplicity, we model this process in the same fashionas for manufacturing, so that agricultural productivity in region n evolves according to

AA,n,t = ηAt + (1− η)AA,n,t−1 (8)

Note that this equation simply corresponds to (7) with sn = 0. Thus, in regions without upper

36In our setup, innovation can be interpreted as entrepreneurs with τ(i) = 1 improving their technology, inspired bynew discoveries at the frontier. This is in line with Mokyr’s (2005a) argument that technological progress often camein the form of micro-inventions by implementation of broader technological concepts.

37Compared to manufacturing, agriculture saw much less innovation that required advanced knowledge to beadopted. This pattern is clearly borne out by innovations exhibited at world fairs: Among the 6,377 British exhibits in1851, only 261 (or 4.1%) were agricultural machinery. At the other end of the spectrum, modern manufacturing sec-tors made up the large majority of innovations: textiles alone accounted for more than 26%, and engines and scientificinstruments for another 15% (Moser, 2012, Table 3).

38This reflects the historical evidence that agricultural productivity also grew significantly during the industrialrevolution (Crafts, 1985).

16

tail knowledge, productivity in agriculture and manufacturing are the same. This delivers a usefulbenchmark case for our analysis.

3.3 Equilibrium and Predictions

In the following, we analyze how worker skills and upper tail knowledge affect income, growth,and the sectoral allocation of labor. Importantly, we assume that (exogenous) technological progressat the frontier, γA,t, is initially slow and then accelerates. Growth in total factor productivity (TFP)was minuscule prior to industrialization – probably in the range of 0.1% per year (Galor, 2005)– and it then accelerated to approximately 1% in the mid-19th century (Crafts and Harley, 1992;Antràs and Voth, 2003). With low γA,t, equation (7) implies that upper tail knowledge does not haveimportant effects on regional productivity; it only becomes important when technology advancesmore quickly. This difference is crucial for our predictions before versus during industrialization.

Within each region n, labor mobility ensures that wages in agriculture and manufacturing areequalized: wA,n = wM,n = wn. Using (2) and (5), this yields the employment share in agriculture:

lA,n =

(AA

(1− αM)AM

hβA−βMn

) 1αA

xn (9)

More land-abundant regions (higher xn) have higher employment shares in agriculture. Since weassume that manufacturing production occurs in cities, the urbanization rate is given by lM,n =

1− lA,M . In addition, equation (5) implies that wages grow at the same rate as local manufacturingproductivity AM,n.39 The growth rate is thus given by

γw,n,t = η

(At

AM,n,t−1

− 1

)+ (1− η) sn · γA,t (10)

where the first term reflects the catchup effect, and the second term, the ‘knowledge effect’ asdiscussed above.

We now present three predictions of the model. The first two analyze the cross-sectional effectof knowledge elites for the cases of relatively low γA (before the Industrial Revolution), and forhigh γA (in the industrial period). The third prediction highlights the role of worker skills. Wediscuss the intuition behind each prediction in the text.

Prediction 1. Pre-industrialization: If the technological frontier expands slowly (low γA), laborshares in manufacturing, wages, and economic growth are only weakly affected by local upper tailknowledge.

39Total income in region n also comprises entrepreneurial profits given by αM (1− αM )YM,n (see Appendix A.1).Since profits are directly proportional to wages, we focus on the latter when discussing the model predictions.

17

Intuitively, if technological progress is slow, entrepreneurs with upper tail knowledge enjoyonly a tiny productivity advantage (or none at all, if γA = 0). Thus, (7) implies that productivityis similar or identical in regions with high and low sn. Consequently, wages and labor shares –given by (5) and (9) – are also similar in the cross-section. The same is true for income growth,given by (10). The left panel of Figure 3 provides an illustration of Prediction 1. Under reasonableparameter choices, the percentage of entrepreneurs with scientific knowledge has only minusculeeffects on development.40

Next, we turn to the industrialization period, when technological knowledge grew rapidly. Thefollowing prediction shows that, despite production knowledge being non-rival and available to allregions, it can have differential effects on economic development.

Prediction 2. During and after industrialization: As the technological frontier expands more rapidly(high γA), a larger local knowledge elite leads to higher wages, higher manufacturing employment,and faster economic growth.

The intuition for this prediction follows the same logic as above, but now with rapid techno-logical growth at the frontier, so that upper tail knowledge has sizeable effects on regional produc-tivity. The right panel of Figure 3 illustrates the prediction in the simple calibrated version of ourmodel: both wages and manufacturing employment now grow hand-in-hand with the local densityof scientific knowledge.

Finally, we describe the effect of worker skills on income and employment.

Prediction 3. Effect of worker skills: Higher average worker skills hn in region n lead to higheremployment shares in manufacturing and higher regional wages, but not to faster growth. Thisholds irrespective of the rate of technological progress at the frontier.

Figure 4 illustrates this prediction.41 Regional wages in both sectors grow in worker skills,as given by (2) and (5). In addition, worker skills are more important in manufacturing than

40The model calibration serves mainly illustrative purposes – we do not intend to precisely predict actual magnitudesof effects. We choose βA = 0.3, βM = 0.5, and αA = 0.6 (αM is a free parameter, chosen implicitly with xn). Forthe catchup parameter, we use η = 0.02, which implies that regions with sn = 0 lag about 35 years behind the frontier.We simulate the model for 250 periods, corresponding to 1600-1850. Over the first 100 periods, we use γA = 0.1%;thereafter, γA rises by 0.015% each year, so that it reaches 2.35% after 250 periods. This is consistent with thecombined contribution of TFP and physical capital to growth during the mid-19th century (Antràs and Voth, 2003)(since we abstract from the latter in the model, we have to load its effect on TFP growth if we want to match historicaldata). Our calibration of γA ensures that wages follow the trend of p.c. income shown in appendix Figure A.1. Wenormalize AM = 1 in 1700 and choose land xn such that the labor share in (overall) manufacturing is approximately10% – this corresponds to the urbanization rate in France in 1700 (total city population from Bairoch, Batou, andChèvre (1988) divided by French population from McEvedy and Jones (1978)). In Figure 3, we keep average workerskills constant at hn = 1.1 and use the range of sn ∈ [0, 0.2]. The latter reflects the historical accounts that only asmall share of French entrepreneurs were progressive and possessed upper tail knowledge. Changing either hn or theinterval for sn does not affect our qualitative predictions.

41We use the parameters described in footnote 40 and keep scientific knowledge constant at the benchmark level

18

in agriculture (βM > βA). Thus, following equation (9), higher hn leads to a concentration ofemployment in manufacturing – and thus in cities. Since these effects are independent of scientificknowledge, we can plot the pre-industrial and industrial periods together. Finally, wage growth asgiven by (10) is independent of worker skills. Intuitively, worker skills affect how productively agiven technology is operated, but not which technology is used.

Summing up, as compared to the previous theoretical literature, our model provides a moredifferentiated view on the role of human capital during industrialization. Distinguishing betweenworker skills and upper tail (scientific) knowledge allows us to derive predictions that differ impor-tantly for the two types of human capital. To test the model’s predictions, we collect a rich datasetcovering industrialization in France.

4 Data

In this section we describe our data, beginning with our main variable – encyclopedia subscriberdensity. We then turn to a set of city-level and regional variables that reflect French develop-ment before, during, and after industrialization. We also analyze whether subscriber density variessystematically with other local characteristics.

4.1 Main Explanatory Variables: Subscriptions and Scientific Societies

Our main dataset consists of 193 French cities for which Bairoch et al. (1988) report population in1750. Among these, 85 have recorded subscriptions to the encyclopedia.42 According to Duplain’slist (see Section 2), there were 7,081 subscriptions to the Encyclopédie’s Quarto Edition in France.Altogether, we match 6,944 subscriptions (or 98% of the total) to the cities in our sample. SinceDuplain’s list covers the universe of subscriptions, we can safely assume that the 108 cities coveredby Bairoch et al. (1988), but not mentioned in Duplain’s list, had no subscribers. In the following,we use Subsn to denote overall subscribers in city n. Since all subscriptions to the Quarto inDuplain’s list were sold at the same price, including shipment (Darnton, 1979, p.264), we arguablymeasure local demand rather than supply. This differentiates our empirical approach from previousstudies that build on local printing activity (c.f. Baten and van Zanden, 2008; Dittmar, 2011).

Because larger cities will mechanically tend to have more subscribers, we normalize sub-

sn = 0. The range hn ∈ (1.0, 1.3) on the horizontal axis is chosen as follows: the average return to schooling ina cross-section of countries is about 0.1 (Bils and Klenow, 2000). In our sample, literacy in 1686 varies between 0and 60% across French departments, and we assume that literacy is equivalent to 5 years of schooling. With this, weobtain an upper bound of exp(0.1 · 0.6 · 5) ≈ 1.3, and a lower bound of exp(0) = 1. Changing these values does notalter our qualitative results.

42In total there are 118 cities with subscriptions to the Encyclopédie listed in Darnton (1973); 12 of these are notreported in Bairoch et al. (1988), and the remaining 21 can be matched to Bairoch et al. (1988), but population dataare not available in 1750.

19

scriptions by population in 1750. Subscriptions per capita (among cities with above-zero en-tries) varied substantially, from 0.5 per 1,000 in Strasbourg to 16.3 in Valence; Paris belongedto the lower tercile of this distribution, with 0.85 subscriptions per 1,000. To reduce the influ-ence of extreme values, we use log-subscriber density as our baseline variable: lnSubDensn =

ln(Subsn/pop1750n + 1), where pop1750n is city population in 1750.43

We also use an earlier indicator for upper tail knowledge: a dummy for cities hosting a sci-entific society before 1750, as well as the number of ordinary members of scientific societies atthe city level. These data are from McClellan (1985). Data on the number of Huguenots at thedepartment level in 1670 are from Mours (1958). Famous Scientists in 1000-1887 are from the In-

dex Bio-Bibliographicus Notorum Hominum (IBN), as coded by de la Croix and Licandro (2012).To proxy for average worker skills, we use department-level data on male literacy in 1686 and1786 from Furet and Ozouf (1977). Literacy rates reflect the percentage of men able to sign theirwedding certificate. For 1837, department-level schooling data are available from Murphy (2010),computed as the ratio of students to school-age population (5 to 15 years).

4.2 Outcome Variables

Our first outcome variable is city growth – a widely used proxy for economic development (Ace-moglu, Johnson, and Robinson, 2005; Dittmar, 2011). We use the panel of city population fromBairoch et al. (1988), which includes cities that reached (at least once) 5,000 inhabitants between1000 and 1800. Bairoch et al. (1988) report city size for every 100 years until 1700, and for every50 years thereafter until 1850.

To test the model predictions on income levels, we need a proxy that is observed before andafter industrialization. Following a rich literature in economic history, we use soldier height (c.f.Steckel, 1983; Brinkman, Drukker, and Slot, 1988; Komlos and Baten, 1998).44 We obtain con-script height before 1750 at the department level from almost 30,000 individual records collectedby Komlos (2005). These reflect conscriptions over the first half of the 18th century. We filterout time- and age-specific patterns in recruitment by using the residual variation in height aftercontrolling for decade fixed effects and conscript age – see Appendix C.3 for more detail. As ourfirst proxy for income after industrialization, we use department level soldier height from Aron,Dumont, and Le Roy Ladurie (1972) for the period 1819-1826. In addition, we use disposableincome in 1864 from Delefortrie and Morice (1959).

43Adding the number 1 ensures that the measure is also defined for cities with zero subscriptions. Appendix C.2provides further detail and distribution plots.

44Cross-sectional analyses typically document a strong positive correlation between height and per capita income(see Steckel, 2008, for a recent survey of the literature and empirical evidence). In longitudinal studies, the relationshipis less clear, since it can also be affected by income inequality, volatility, and food prices (Komlos, 1998). We thusonly exploit the variation in height across French regions, but not within regions over time.

20

Next, we use wages in industry and agriculture (measured in 1852) from Goreaux (1956),employment shares in industry and agriculture (in 1876) from Service de la Statistique Général deFrance (1878), and wages in the textile and the food industry (in 1864) from Delefortrie and Morice(1959). Finally, we perform a detailed within-sector analysis, using local wages and value addedper worker as a proxy for productivity. The underlying data are from Chanut, Heffer, Mairesse,and Postel-Vinay (2000), who cleaned and digitalized a survey of 14,238 firms over the period1839-1847. The data were collected by the Statistique de la France at the arrondissement (county)level, and categorize firms into 16 industrial sectors.

4.3 Control Variables

In the following, we briefly describe our control variables. Appendix C.4 provides more detaileddescriptions and sources. Our baseline set of controls includes various geographic characteristicsof cities, such as dummies for cities with ports on the Atlantic Ocean and on the MediterraneanSea, as well as for cities located on navigable rivers. Following Dittmar (2011), we also include adummy for cities that had a university before 1750, a printing press between 1450 and 1500, andthe log number of editions printed before 1501. To control for cultural and language differences,we construct a dummy for cities located in non-French speaking departments.

We also control for a number of potential confounding factors: First, encyclopedia subscrip-tions may have been high where book sales in general were high – thus not reflecting upper tailknowledge but a broader interest in reading. We obtain book sales to each French city over theperiod 1769-1794 from the FBTEE (2012) project that reconstructed the book trade of the impor-tant Swiss publishing house STN (which also published the Encyclopédie). This source covers thesales of more than 400,000 copies, for which the location of the buyer (private or book stores, withthe vast majority located in France) is available.45 A second potential concern is that the nobilitymay have both purchased a disproportionate amount of encyclopedias and funded industrialization.We obtain the number of noble families in each department from the Almanach de Saxe Gotha, themost important classification of European royalty and nobility. Altogether, our sample containsmore than 1,000 noble families in 1750, in 88 French departments. Third, encyclopedia sales mayhave been high where industrialization was already on the way in the mid-17th century. To accountfor this possibility, we use data on proto-industrialization in France, as used by Abramson and Boix(2013). These allow us to compute the number of mines, forges, iron trading locations, and textilemanufactures prior to 1500 for each department. Finally, another potential concern is related to the‘Reign of Terror’ in 1792-94, when alleged counter-revolutionaries were murdered on a massive

45This data is available at http://chop.leeds.ac.uk/stn/. The STN book sales are directly comparable to our data onencyclopedia subscriptions: both occurred during the same period, are shipped from one place (Switzerland and Lyon)towards all of France, and both reflect local demand for books (sales) rather than supply (printing).

21

scale. If these were concentrated in less progressive areas (i.e., with low subscriptions), executionsmay explain slower economic progress there. To control for this possibility, we process data onapproximately 13,000 executions, assigning them to each department.

In order to guarantee consistency with our main explanatory variable, we calculate the localdensity of scientific society members, total book sales, noble families in 1750, proto-industriallocations, and executions during the ‘Reign of Terror’ in the same way as for subscriptions:ln(1 + x/popn,1750). Finally, Appendix C.5 describes how we aggregate city-level variables tothe arrondissement and department level.

4.4 Balancedness

Do other town characteristics vary systematically with encyclopedia subscriptions? In Table 1 weregress our main explanatory variable lnSubDens on a variety of controls (one-by-one, includinga dummy for Paris). We begin with our baseline controls in the first two columns. Column 1 usesall cities, while column 2 uses only those with above-zero subscriptions. Few control variablesshow a consistent pattern. City size is significantly positively correlated with lnSubDens in col1, but significantly negatively in col 2.46 Seaports are essentially uncorrelated with subscriberdensity. The coefficient on navigable rivers is significant in col 1, but switches signs and becomesinsignificant in col 2. The correlation with university and printing press dummies, as well as withbooks printed before 1500, are all positive (and significant in col 1), as one should expect sincethey reflect local advanced knowledge. In addition, cities in non-French speaking areas have asmaller proportion of subscribers, which is also to be expected given that the encyclopedia waspublished in French.47

Next, columns 3-5 in Table 1 report the coefficients of regressing subscriber density on ourproxies for average worker skills, as well as a variety of potentially confounding variables.48 Liter-acy rates in both 1686 and 1786, as well as schooling rates in 1837, are not significantly correlatedwith subscriber density in any specification. On the other hand, overall (STN) book purchases arestrongly and positively associated with encyclopedia subscriptions. This suggests that locationswith a greater interest in reading also host more people with scientific interests. In other words, theupper tail of the knowledge distribution is thicker where the distribution in general is shifted to the

46Below, we confirm that our results hold in both the full sample and the Subsn > 0 sample. Note that thisimplicitly addresses potential unobserved factors that are associated with city size, since these should also change theway in which they affect our results between the two samples.

47There were a number of regions in 18C France that spoke different languages such as Alsacien and Basque. Thecorresponding six departments comprise 24 cities in our sample, out of which 6 had above-zero subscriptions.

48Since most of these variables are observed at the department level, we compute subscriber density at the depart-ment level in the same way as for cities (see Appendix C.5). The department-level data comprise 88 observationsin the cross-section, and 66 out of these had above-zero subscriptions. Col 3 reports coefficients for all departmentswithout controls, col 4 adds controls, and col 5 restricts the sample to departments with above-zero subscriptions.

22

right. In our empirical analysis below we show that subscriber density strongly dominates generalbook purchases as an explanatory variable for development. Next, the correlation between sub-scriber density and proto-industrialization is small and insignificant. This makes reverse causalityunlikely – the concern that encyclopedia sales may have been higher in areas with early industrialactivity. The local density of noble families is positively associated with subscriber density, witha significant coefficient in col 5. This is in line with nobles being prominent subscribers to theencyclopedia. Finally, the density of ‘counter-revolutionaries’ killed during the Reign of Terroris positively, but not significantly correlated with subscriber density. Appendix C.6 provides fur-ther tests of balancedness, comparing cities with and without subscriptions, as well as those withabove- and below-median subscriber density. These additional tests confirm the pattern describedabove: the few city characteristics that vary systematically with subscriber density are those thatone should expect if subscriptions reflect the size of the local knowledge elite.

5 Empirical Results

In this section we present our empirical results. We test the predictions of the model by estimatingequations of the form

yn = β · Sn + γhn + δXn + εn , (11)

where Sn denotes different measures for the size of the knowledge elite in location n, i.e., thedensity of encyclopedia subscribers and members of scientific societies; hn denotes proxies foraverage human capital, such as literacy and schooling; Xn is a vector of control variables, and εn

represents the error term. We use a variety of outcome variables yn, depending on the propositionthat we analyze. Propositions ??-?? imply the following expected coefficients: First, when usingcity growth as an outcome, we expect β = 0 and γ = 0 prior to 1750, as well as β > 0 and asmall negative γ after 1750. Second, for income proxies such as soldier height, we again expectβ = 0 prior to 1750 and β > 0 afterwards, but γ > 0 in both periods. Third, the same is true whenwe use measures for manufacturing activity (e.g., wages and employment) – with an additionaldimension: after 1750, we only expect β > 0 for modern manufacturing, but β = 0 for traditionalmanufacturing.

5.1 City Growth

City growth is a widely used proxy for economic development (DeLong and Shleifer, 1993; Ace-moglu et al., 2005; Dittmar, 2011). Following this approach, we use the outcome variable gpopnt,the log difference in city population between two periods t and t − 1 (mostly 100-year intervals).Figure 5 shows the distribution of gpopnt for different subscriber densities. During industrializa-tion in France, cities with encyclopedia subscriptions grew substantially faster: the city growth

23

distribution is markedly shifted to the right. In addition, this shift is more pronounced for citieswith above-median subscriptions per capita. In the following, we analyze this pattern in moredetail and account for possible confounding factors, such as initial city size.

Cities with vs. without Subscriptions: Matching Estimation

Between 1750 and 1850, cities with above-zero subscriptions to the encyclopedia grew at ap-proximately double the rate as compared to those without subscribers (0.51 vs. 0.26 log points).However, merely comparing the two subsets is problematic, because larger cities are more likelyto have at least one subscriber. As a first pass at this issue, we use propensity score matching,comparing cities with and without subscriptions of very similar size. The results are reported inTable 2.49 Panel A shows results for a variety of specifications for the period 1750-1850. In col1, we use the full sample and match by initial population; col 2 excludes the 10% smallest andlargest cities in 1750. In cols 3 and 4 we introduce geographic latitude and longitude as additionalmatching variables.50 We thus compare nearby cities with similar population size, accounting forregional unobservables. The results are stable throughout: French cities with subscriptions grewapproximately 0.15 log points faster than those of comparable size without subscriptions. Thisdifference is statistically and economically significant: average city growth over this period was0.37 log points.

In Panel B of Table 2, we repeat the analysis for the pre-industrial period. The difference ingrowth is now substantially smaller and statistically insignificant, and for 1400-1500, the coeffi-cient is even negative and significant. The matching results support our model prediction that aknowledge elite – as proxied by encyclopedia subscriptions – fostered economic growth during,but not before industrialization.

Subscriber Density

We now turn to our main explanatory variable, subscriber density lnSubDensn, using OLS re-gressions. This offers several advantages over the previous matching exercise: it exploits the fullvariation in subscriber density (instead of only a dummy), we can examine the coefficient on con-trols to see how they affect city growth, and we can use population-based weights to reduce the

49Following Abadie, Drukker, Herr, and Imbens (2004), we use the three nearest neighbors. Our results are robustto alternative numbers of neighbors. We define ‘treatment’ as cities with above-zero subscriptions and report averagetreatment of the treated (ATT) effects. Finally, we exclude the top and bottom 1-percentile of city growth rates for eachrespective period. This avoids that extreme outliers due to population changes of very small cities (for example, from1,000 to 4,000 or vice-versa) drive our results. In the OLS analysis below, we address this issue by using populationweights; in propensity score matching, weighting by city size is not feasible.

50The average population difference between matched cities with and without subscriptions is 6,500 inhabitants incol 1, and 500 inhabitants in col 2. When matching by geographic location (col 3), matched cities are on average lessthan 30 miles apart.

24

noise in growth rates due to population changes in small cities – there is more reliable informationin a city growing from 100,000 to 200,000 inhabitants, than in one growing from 1,000 to 2,000.

Table 3, Panel A, presents our main OLS results for city growth. Subscriber density is stronglyand positively associated with city growth over the period 1750-1850 (columns 1-4). Atlantic andMediterranean ports have a similar effect; the former is in line with Acemoglu et al. (2005). Thenegative coefficient on initial population (after controlling for other characteristics) provides someevidence for conditional convergence (Barro and Sala-i-Martin, 1992). We also include a dummyfor Paris – the capital and by far the largest city. In addition, we control for the effect of locallanguages: because the Encyclopédie was written in French, subscriptions were lower in areaswith different languages: 0.50 per 1,000 inhabitants, versus 1.83 in French-speaking cities. BothParis and cities in non-French speaking areas grew faster than average between 1750 and 1850.51

Finally, we include a dummy for cities that had a printing press in 1500 and control for the lognumber of editions printed by that date. This replicates the specification in Dittmar (2011), whoshows that an early adoption of printing had strong positive effects on long-term city growth inEurope. Within France, the pattern is less robust – the dummy for cities hosting a printing presschanges signs depending on the period. In sum, including a rich set of control variables doesnot affect the size or statistical significance of the coefficient on lnSubDens. A one standarddeviation increase in subscriber density is associated with a city growth rate that is higher by 8.2-16.9 percentage points (0.17-0.35 standard deviations).

Columns 5-8 repeat the analysis for the pre-industrial period. The coefficients on lnSubDens

are now small and insignificant. In addition, where sample size and specification are comparable,the difference between the post- and pre-1750 coefficients is strongly significant: comparing thecoefficients on lnSubDens in cols 3 and 5, the 95% confidence interval does not overlap. In PanelB of Table 3 we repeat all city growth regressions for the subset of cities with above-zero subscrip-tions. The results closely resemble those of Panel A: subscriber density is strongly and significantlyassociated with city growth after 1750, but not before – in fact, in Panel B the coefficients for thetwo periods prior to industrialization are now negative (and insignificant).

Literacy and additional controls

We now turn to the role of average worker skills (measured by literacy) for city growth. Table4 reports the results; it also introduces additional controls – the potential confounding factors

51The latter is mostly due to 6 Alsacien-speaking cities in the Rhine area. This area saw rapid growth during theindustrialization. When including a separate Rhine-area dummy, the coefficient of non-French speaking falls to lessthan one half its original size, while the coefficient on lnSubDensn is unchanged.

25

discussed in Section 4.3.52 The coefficients on literacy are negative throughout, and statisticallysignificant in some specifications. According to the point estimates, the effect of literacy is alsoquantitatively small: a one standard deviation increase in literacy is associated with a decline incity growth by 0.11 standard deviations; for lnSubDens, the point estimates imply an increase byapproximately 0.33 standard deviations.

Among the additional controls, only proto-industrialization is positively and significantly as-sociated with city growth (col 2 in Table 4). This confirms the findings in Abramson and Boix(2013). Nevertheless, controlling for proto-industrial activity does not alter the coefficient on sub-scriber density – in fact, the two measures are (weakly) negatively correlated (see Table 1). Thestandardized beta coefficient on proto-industrial activity is 0.17, and thus half the size of the sub-scriber density effect. Nobility density is not associated with city growth (col 3). This is in linewith the historical evidence discussed in Section 2.2 that only a progressive subset of the nobilitywas involved in industrial activity. Next, while overall book purchases are strongly correlated withlnSubDens (see Table 1), they do not affect city growth (col 4). The coefficient on executionsduring the reign of terror (cols 5) is also small and insignificant. Finally, column 6 includes allpotential confounding factors together, confirming the previous results. Importantly, in all regres-sions the coefficients on lnSubDens remain highly significant and quantitatively similar to ourbaseline results in Table 3. In sum, the results in Table 4 strongly support our model predictionsthat there is a crucial difference between average worker skills and upper tail knowledge.

Alternative specifications

So far, we have shown that the association between subscriber density and city growth was strongafter 1750, and weak before that date. Next, we analyze systematically whether the observed in-crease in the coefficient was statistically significant. Table 5 replicates the specification from Nunnand Qian (2011), using log population as dependent variable in a panel setting. This specificationalso includes city fixed effects, absorbing all unobserved city characteristics that do not vary overtime. We find that the interaction of lnSubDens with a post-1750 dummy is highly significant andpositive, with a magnitude that is very similar to the previous growth regressions. This finding isrobust to including interactions of the post-1750 dummy with our baseline controls (col 2), as wellas with our additional controls (col 3). The baseline result also holds in a balanced sample thatonly includes cities where population is observed in every sample year between 1500 and 1850(col 4). Finally, columns 5 and 6 report the results for placebo ‘treatments’, when moving theindustrialization cutoff to 1600 or 1700. Both yield small and insignificant coefficients.

52Standard errors are now clustered at the department level, since this is the geographical unit at which literacy andmost confounding factors are observed.

26

In Table 6 we show that our results also hold for the two sub-periods 1750-1800 (col 1) and1800-1850 (col 2). The coefficients on lnSubDens are highly significant and of a similar mag-nitude.53 Splitting the period of French industrialization into pre- and post-1800 also allows foran additional check: in col 3, we control for growth in the earlier period. This variable shouldcapture unobserved (and persistent) drivers of city growth. Including it does not change the coef-ficient on lnSubDens in the later period. This further alleviates the concern that unobserved citycharacteristics may be behind our results.

We perform a number of additional checks in Appendix C.7. First, we run our city growthregressions for the period 1750-1850, using four different sub-samples, each including those citiesfor which population data are available in the year 1400, 1500, 1600, and 1700 respectively. Thecoefficients on lnSubDens are very similar to our baseline results and are always significant at the1% level. Next, our results are almost identical when pooling growth over the period 1400-1750and comparing it with growth in 1750-1850 (as visualized in Figure 2). In addition, we use twoalternative measures of subscriber density: one that is not log-based, and another one that allowsfor variation in subscriber density across cities without subscription, assigning lower densities tolarger cities. All our results continue to hold. Finally, in line with the model predictions, we findno clear relationship between early literacy (recorded in 1686) and city growth prior to 1750.

Scientific societies

Since data on encyclopedia subscriptions are from the second half of the 18th century (1777-80),local adoption of industrial technology may already have been on the way when subscriptionsoccurred. Thus, reverse causality is a concern, because successful early industrial entrepreneursmay have been among the subscribers. In the following, we address this issue by using an earlierproxy for knowledge elites: the presence and membership of scientific societies in each city before1750 from McClellan (1985). Altogether, there are 22 cities with scientific societies foundedbefore 1750. This alternative variable is strongly associated with our main measure: cities hostinga scientific society were over three times more likely to also have encyclopedia subscriptions, andthey had almost four times more subscribers per capita (see Table A.6 in the Appendix).

Table 7 repeats our city growth regressions, using early scientific societies as explanatory vari-able. In Panel A we use propensity score matching, comparing cities of similar size and geographiclocation that hosted a scientific society prior to 1750 with cities that did not. Over the industrial-ization period (cols 1-2), French cities hosting a scientific society grew 0.2 log points faster than

53The raw correlation of city growth over the two periods is small: 0.07. One possible explanation is a structuralbreak after the French Revolution. Nevertheless, this did not change the way in which knowledge elites affected citygrowth – that is, even if the cross-sectional pattern of growth changed, cities with higher subscriber density grew fasterin both periods.

27

those without; the growth differential is statistically significant and is robust to adding geographiclatitude and longitude as matching variables. It is also very similar in magnitude to the results inTable 2. Throughout the pre-industrial period (cols 3-6), the growth differential is insignificant. InPanel B, we presents OLS results, using the local density of members, lnMembDens, as explana-tory variable.54 Membership density in scientific societies is strongly and positively associatedwith city growth over the period 1750-1850 (columns 1-2), but not before (columns 3-6). Theresults based on early scientific societies thus confirm our main results.

5.2 Income and Industrialization

We now turn to the effect of subscriber density on income and industrial employment in the 19thcentury, using a variety of outcome variables. The corresponding data is available at the departmentlevel, which we use as our main unit of analysis in the following.

Soldier Height

We begin by using a common proxy for income – soldier height. This variable is available inFrance before and after 1750, allowing us to test the income effects that our model predicts forworker skills and knowledge elites before and after the onset of industrialization. Columns 1-4 inTable 8 show that conscript height in 1819-26 is strongly positively associated with both subscriberdensity and literacy. This holds even when we control for historical height (cols 3-4). The pointestimates imply that a one standard deviation increase in lnSubDens is associated with an increasein average soldier height by 0.3cm (or 0.25 standard deviations). Columns 5 and 6 show thatsoldier height prior to 1750 is also positively associated with literacy, but not with lnSubDens.55

The results on soldier height are strong evidence in favor of Propositions ??-??.

Disposable Income, Employment Shares and Wages

In column 1 of Table 9 we show that encyclopedia subscriptions in the mid-18th century predictdisposable income in 1864, i.e., about a century later. The point estimate implies that a one stan-dard deviation increase in lnSubDens is associated with 6.5 percent (0.29 standard deviations)higher income. To proxy for average worker skills, we can now use department-level schoolingrates, which is available for 1837. This variable is positively and significantly associated with in-come. Next, we use employment shares in 1876 as dependent variables. In line with Proposition??, subscriber density predicts lower employment shares in agriculture (col 2) and higher industrial

54This variable is calculated in the same way as lnSubDens (see Section 4.1). The two measures are stronglycorrelated, with a coefficient of 0.313 (p-value 0.0001).

55Table A.9 in the appendix reports further robustness checks, showing that all results hold when regressing con-script height separately on lnSubDens and Literacy, and when weighting regressions by the number of soldiers forwhich height is observed in each department.

28

employment (col 3). The same pattern holds for schooling rates, confirming Proposition ??.In columns 4-7, we test our model’s predictions for wages after the onset of industrialization,

using department-level data from the mid-19th century. Column 4 shows that subscriber density isnot associated with wages in agriculture – the coefficients are small and insignificant; however, itis a positive and significant predictor of industrial wages (col 5). The industry wages in columns5 reflect combined traditional and modern manufacturing. In columns 6-7, we separate the twoby using wages in the textile (modern) and food (traditional) industries. We find that the effect oflnSubDens is positive and significant for the textile sector (and larger than the previous coefficientfor industry overall); and it is minuscule and insignificant for the food processing sector. Thisunderlines the differential effect of upper tail knowledge – in line with our model, it is important formodern industrial technology but not for traditional modes of production. Finally, the coefficienton schooling is positive and strongly significant throughout columns 4-7.56

All regressions in Table 9 control for population density in order to account for differences inurban vs. rural areas. As expected, industrial employment is larger in more densely populatedareas.

5.3 Innovativeness and the Role of Knowledge Elites

Our results suggest that knowledge elites are strongly associated with higher firm productivity inmodern manufacturing, but not in traditional manufacturing. In the following, we analyze onemechanism that can explain this pattern: upper tail knowledge was important for the adoption,diffusion, and further improvement of innovative industrial technology (Mokyr, 2005a,b). Forentrepreneurs and scientists in France, staying up-to-date with the latest British inventions ensureda competitive advantage, and scientific publications were an important source of this information(Horn, 2006). In the following, we implement a three-step argument. We first use British patentdata to split industrial sectors into modern and traditional. We then show that the dominant shareof technologies described and illustrated in the encyclopedia was indeed modern.57 Finally, weshow that subscriber density predicts firm productivity in modern, but not in traditional sectors.

From English Patents to Plates in the Encyclopedia

Nuvolari and Tartari (2011) provide data on the share of "inventive output" of 21 British industrialsectors for the period 1617-1841. This measure is based on reference-weighted patents, adjustedfor the sector-specific frequency of patenting rates and citations. For example, textiles have the

56In Table A.10 in the appendix we provide additional evidence in support of our model: the density of proto-industrial activity (i.e., prior to industrialization) is strongly associated with literacy, but not with subscriber density.

57We do not argue that the encyclopedia was the only publication that illustrated recent innovations. However, itssubscribers were also more likely to read other scientific publications, and to be involved in innovation. This underlinesthe character of our subscriber density measure as a proxy for local knowledge elites.

29

highest score, accounting for 16.6% of total inventive output; and pottery, bricks and stones are atthe lower end with a share of 1.4%.

We then use the British patent data to analyze whether ‘modern’ sectors were disproportion-ately represented in the encyclopedia. The encyclopedia used plates to illustrate crafts, processes,and inventions. We obtain detailed information on 2,575 illustrative plates accompanying 326encyclopedia entries (see Appendix C.9 for sources and further detail). Thus, on average eachentry was accompanied by 8 illustrative plates. About half of these describe manufacturing tech-nologies, and entries include examples such as cloth cutting and figuring or machines to evacuate

water from a mine. We match each entry to one of the 21 British industrial sectors, which allowsus to split entries and plates into ‘modern’ and ‘old’ technologies, corresponding to above- andbelow-median share of total inventive output. The share of inventive output was more than fivetimes higher in the modern category (8.4% vs. 1.6%), and more than two thirds of all plates ded-icated to manufacturing described ‘modern’ technologies (see Table A.11 in the appendix). Thus,there is strong evidence that the encyclopedia indeed spread predominantly knowledge on modern,innovative industrial technology.

Knowledge Elites and Productivity in Modern vs. Traditional Manufacturing

We now analyze the relationship between subscriber density and firm productivity (proxied bywages) in ‘modern’ versus ‘old’ manufacturing sectors. This analysis builds on data from a Frenchindustrial survey of more than 14,000 firms in 1839-47, which reports the sector and firm locationby arrondissement.58 We run the following regression separately for ‘modern’ and ‘old’ sectors:

ln (wagejn) = β · lnSubDensn + γ · schoolingn + δXn + ϕ · sizejn + αj + εjn , (12)

where wagejn is the average male wage paid by firms operating in sector j in arrondissement n,and the vector Xn includes our baseline controls used above, as well as total population and theurbanization rate in order to control for agglomeration effects. In addition, we control for sectorfixed effects (αj), and for average firm size (sizejn) to capture scale effects.

If upper tail knowledge affected development by raising the productivity in innovative technol-ogy, we expect the coefficient β to be larger when running (12) for ‘modern’ as opposed to ‘old’sectors. Table 10 presents compelling evidence in support of this hypothesis. Subscriber densityis a strong positive predictor of firm productivity in the ‘modern’ sectors (col 1). This holds afteradding our baseline controls (col 3), and also when including additional controls and sector fixedeffects (cols 5 and 7). The last specification exploits only within-sector variation across arrondisse-

58French arrondissements are comparable to US counties – there were altogether 356 arrondissements in 86 depart-ments in the mid-19th century. Appendix C.10 describes the firm survey in more detail and shows how we matchFrench to British sectors. It also lists the resulting consistent 8 sectors and their share of ‘inventive output’.

30

ments; the point estimate can be interpreted as follows: suppose that we "move" a representativefirm in a given ‘modern’ sector from an arrondissements without subscriptions (the lower half ofsubscriber density) to one in the 90th percentile of subscriber density. Then its output per workerwould increase by 13 percent. For ‘old’ sectors, on the other hand, the coefficients on subscriberdensity are small and almost always insignificant (odd columns).59 In line with our model, school-ing is positively associated with firm wages in both ‘old’ and ‘modern’ manufacturing.

Finally, we analyze the effect of encyclopedia subscriptions on wages within individual sectors,running regression (12) separately for different sectors j. Table 11 reports the results, rankingsectors by the size of the coefficient on lnSubDens. The top four sectors are all ‘modern’, andthe bottom four are all ‘old’. In particular, the coefficient on subscriber density is large withinsectors that saw rapid innovation during industrialization, such as textiles or transportation (thefirst steamboat was built in France in 1783). For the least innovative sectors (leather, mining,ceramics and glass), subscriber density is only weakly related to wages. Column 3 shows that thenumber of observations and the R2 are similar for ‘modern’ and ‘old’ sectors. Thus, overall fit orsmall samples do not drive the differences in coefficients. Table 11 also revisits the concern thatunobserved local wealth may drive our results, i.e., that the rich could afford the encyclopedia andalso had the financial means to invest in industrial machines. We use two proxies for an industry’sdependence on up-front investment: the number of steam engines (col 4), and the number of otherengines such as wind and water mills (col 5). Interestingly, both measures for up-front investmenttend to be higher in sectors where the effect of encyclopedia subscriptions is weaker (such asmetal and leather). This makes financial abundance an unlikely confounding factor.60 In sum, ouranalysis suggests that knowledge elites supported industrialization by raising the local productivityin innovative modern technology.

5.4 Discussion: Interpretation and Limitations of Results

Our empirical analysis follows the common approach to regress growth rates on initial levels ofhuman capital (Barro, 2001). It thus also shares the common limitation that skills are not assignedexogenously to different locations, making causal inference difficult. In the following, we interpretour results accordingly.

We have documented a striking pattern: encyclopedia subscriptions are strongly associated

59Note that this difference is not due to noisier estimates – the standard errors are very similar to those for ‘modern’sectors, and the number of observations is actually larger for ‘old’ sectors. In Table A.13 we show that these resultsare also robust to including regional fixed effects.

60Verley (1985, p.103-104) observes that the metal industry was particularly capital intensive and often operatedby the rich nobility, while textile production occurred at a smaller scale and required much less capital. Our findingthat subscriber density is particularly strongly associated with productivity in the latter is thus compatible with aknowledge-based explanation, but not with an investment-based (confounding) one.

31

with economic growth and income after 1750, but not before the onset of industrialization inFrance. Our interpretation is not that the encyclopedia turned its readers into entrepreneurs, orthat it caused local upper tail knowledge. Instead, we use subscriber density as an indicator forthe local presence of knowledge elites. This culture of knowledge was likely locally persistent andexisted prior to the encyclopedia: we documented that pre-1750 membership in scientific societiesis a strong predictor of encyclopedia subscriptions. Our analysis is thus similar to a difference-in-difference (DID) approach where the ‘treatment’ (upper tail knowledge) is a latent, persistent localcharacteristic whose effect is turned on only after the emergence of modern technology rendersupper tail knowledge economically ‘useful’ (Mokyr, 2005b).61 This limits the set of potential con-founding factors – to drive our results, they would have to be correlated with subscriber density,affect growth via channels unrelated to upper tail knowledge, and be ‘switched on’ around 1750.

We have analyzed alternative explanations that may fit this pattern – institutions, nobility, and‘Reign of Terror’ – and concluded that they are unlikely to account for our results.62 While wecannot fully exclude the possibility of yet other unobservables, we believe that the most reasonableinterpretation of our results is the one that relies on overwhelming support from the historicalliterature: knowledge elites played a crucial role for economic growth once industrial technologybecame economically viable in the 18th century. The underlying mechanism is not confined toinventors or scientists actively improving technology, but it also comprises lower access costs(e.g., via information networks in the knowledge elite) and higher efficiency at adopting complexmodern techniques (Mokyr, 2005b). Thus, our interpretation emphasizes a broad concept of uppertail knowledge – but one that is clearly distinct from ordinary worker skills.

Average worker skills also show a stable local pattern over time – the department-level literacyrate in 1686 has a correlation coefficients of 0.84 with literacy in 1786, and of 0.65 with school-ing in 1837. We confirm the model prediction that these measures are positively associated withdevelopment both before and after industrialization began. While endogeneity is a concern forthese cross-sectional results, it can hardly explain our finding that literacy was not, or if anythingnegatively, associated with city growth during industrialization.

In sum, by predicting when the two types of human capital will and will not affect economic de-velopment, our model allows us to run a number of implicit checks. These are all confirmed by the

61In DID terminology: by comparing locations with high (‘treated’) and low (‘control’) subscriber density, weaccount for unobserved country-wide changes during industrialization; and by comparing the relationship betweensubscriber density and economic outcomes before and after 1750, we control for unobserved local characteristics.

62We also demonstrated that subscriber density was unrelated to most observable local characteristics at the eve ofFrench industrialization, and that it was not associated with economic performance prior to 1750. In addition, 18Csubscriber density does not predict mid-19C productivity across the board, but only where we should expect it: withininnovative modern sectors.

32

data. The empirical results thus strongly support our theoretical setup that differentiates betweenaverage and upper tail skills, and its prediction that only the latter was crucial for industrialization.

6 Conclusion

An ample literature has highlighted the importance of human capital for economic developmentin the modern world. However, its role during the Industrial Revolution is typically found to beminor. Thus, a crucial driver of modern growth appears to be unrelated to the onset of growth itself,and thus to the greatest structural break in economic history that led to a vast dispersion of percapita income (Galor, 2011). Previous studies have typically used literacy or elementary educationto measure human capital at the eve of the Industrial Revolution. We show that distinguishingbetween average worker skills and upper tail knowledge is crucial, reinstating the role of humancapital for industrialization.

We extend a workhorse model of the industrial revolution by splitting manufacturing into atraditional and a modern sector. This allows us to introduce two assumptions that are motivated byample historical evidence: i) worker skills were needed in both sectors, and if anything, somewhatmore so in the traditional one, ii) upper tail knowledge was only important in the modern sector.Since technological progress during industrialization was concentrated in the modern sector, themodel predicts that upper tail knowledge was crucial for economic take-off. On the other hand,spatial differences in average worker skills can explain income dispersion in the cross-section,but are unrelated to growth. We present strong empirical support for these predictions, usingliteracy and elementary schooling to proxy for average worker skills, and the novel measure of18C encyclopedia subscriptions to capture local knowledge elites. Importantly, we do not arguethat average worker skills were altogether unimportant; we show that they were strongly correlatedwith income levels before and after industrialization, but not with growth. In contrast, subscriberdensity is not associated with income or industrial activity prior to 1750, but is a strong predictorof growth thereafter, and of income levels and industrial activity in the 19th century. In line withour theoretical prediction, subscriber density is a particularly strong predictor of firm productivityin innovative modern sectors.

Our results have important implication for economic development: while improvements in ba-sic schooling raise wages, greater worker skills alone are not sufficient to trigger a transition tosustained industrial growth. Instead, upper tail skills – even if confined to a small entrepreneurialelite – are crucial, fostering growth via the innovation and diffusion of modern technology. Inthis respect, our findings resemble those in today’s economies, where the existence of a socialclass with high education is crucial for development (Acemoglu, Hassan, and Robinson, 2011),entrepreneurial skills matter beyond those of workers (Gennaioli et al., 2013), and scientific edu-

33

cation is key (Hanushek and Kimko, 2000).

ReferencesAbadie, A., D. Drukker, J. L. Herr, and G. W. Imbens (2004). Implementing Matching Estimators for

Average Treatment Effects in Stata. Stata Journal 4, 290–311.

Abramson, S. F. and C. Boix (2013). The Roots of the Industrial Revolution: Political Institutions or(Socially Embedded) Know-How? APSA 2013 Annual Meeting Paper.

Acemoglu, D., P. Aghion, and F. Zilibotti (2006, 03). Distance to Frontier, Selection, and Economic Growth.Journal of the European Economic Association 4(1), 37–74.

Acemoglu, D., F. A. Gallego, and J. A. Robinson (2014). Institutions, Human Capital and Development.Annual Review of Economics. forthcoming.

Acemoglu, D., T. A. Hassan, and J. A. Robinson (2011). Social Structure and Development: A Legacy ofthe Holocaust in Russia. Quarterly Journal of Economics 126(2), 895–946.

Acemoglu, D., S. Johnson, and J. A. Robinson (2005). The Rise of Europe: Atlantic Trade, InstitutionalChange and Economic Growth. American Economic Review 95(3), 546–579.

Allen, R. C. (2003). Progress and Poverty in Early Modern Europe. Economic History Review 56(3),403–443.

Antràs, P. and H.-J. Voth (2003). Factor Prices and Productivity Growth during the British Industrial Revo-lution. Explorations in Economic History 38, 52–77.

Aron, J. P., P. Dumont, and E. Le Roy Ladurie (1972). Anthropologie du Conscrit Français. Paris: MoutonDe Gruyter.

Bairoch, P., J. Batou, and P. Chèvre (1988). La Population des villes Europeenees de 800 à 1850: Banquede Données et Analyse Sommaire des Résultats. Geneva: Centre d’histoire economique Internationale del’Université de Genève, Libraire Droz.

Ballot, C. (1923). L’Introduction du Machinisme dans l’Industrie Française. F. Rieder and cie, Paris.

Barbour, I. and S. Blydenburgh (1844). The Silk Culture in the United States. Greeley & McElrath.

Barro, R. J. (1991). Economic Growth in a Cross Section of Countries. Quarterly Journal of Eco-nomics 106(2), 407–43.

Barro, R. J. (2001). Human Capital and Growth. American Economic Review Papers and Proceedings 91(2),12–17.

Barro, R. J. and X. Sala-i-Martin (1992). Convergence. Journal of Political Economy 100(2), 223–251.

Baten, J. and J. van Zanden (2008). Book Production and the Onset of Modern Economic Growth. Journalof Economic Growth 13(3), 217–235.

Becker, S. O., E. Hornung, and L. Woessmann (2011). Education and Catch-up in the Industrial Revolution.American Economic Journal: Macroeconomics 3(3), 92–126.

Benhabib, J. and M. M. Spiegel (1994). The Role of Human Capital in Economic Development: Evidencefrom Aggregate Cross-Country Data. Journal of Monetary Economics 34(2), 143–173.

34

Bils, M. and P. J. Klenow (2000). Does Schooling Cause Growth? American Economic Review 90(5),1160–1183.

Bloom, N. and J. V. Reenen (2007). Measuring and Explaining Management Practices Across Firms andCountries. Quarterly Journal of Economics 122(4), 1351–1408.

Braudel, F. (1982). Civilization and Capitalism, 15th–18th Century: The perspective of the world. Univer-sity of California Press.

Brinkman, H. J., J. Drukker, and B. Slot (1988). Height and Income: A New Method for the Estimation ofHistorical National Income Series. Explorations in Economic History 25(3), 227–264.

Cantoni, D. and N. Yuchtman (2014). Medieval Universities, Legal Institutions, and the Commercial Revo-lution. Quarterly Journal of Economics. forthcoming.

Carus-Wilson, E. M. (1966). The Woollen Industry. Cambridge Economic History of Europe. Volume II.Cambridge University Press.

Caselli, F. and W. J. Coleman (2006). The World Technology Frontier. American Economic Review 96(3),499–522.

Chanut, J.-M., J. Heffer, J. Mairesse, and G. Postel-Vinay (2000). L’Industrie franccaise au milieu du 19esiécle: les enquêtes de la statistique générale de la France. Editions de l’école des hautes études ensciences sociales, Paris.

Chaptal, M. (1819). De l’Industrie Français, Volume 2. Paris.

Chaussinand-Nogaret, G. (1985). The French Nobility in the Eighteenth Century: From Feudalism to En-lightenment. Cambridge University Press.

Ciccone, A. and E. Papaioannou (2009). Human Capital, the Structure of Production, and Growth. Reviewof Economics and Statistics 91(1), 66–82.

Clair, C. (1976). A History of European Printing. London: Academic.

Clark, G. (2005). The Condition of the Working Class in England, 1209-2003. Journal of Political Econ-omy 113(6), 1307–1340.

Cohen, D. and M. Soto (2007). Growth and human capital: good data, good results. Journal of EconomicGrowth 12(1), 51–76.

Crafts, N. F. R. (1985). British Economic Growth During the Industrial Revolution. Oxford: OxfordUniversity Press.

Crafts, N. F. R. (1996). The First Industrial Revolution: A Guided Tour for Growth Economists. AmericanEconomic Review Papers and Proceedings 86(2), 197–201.

Crafts, N. F. R. and K. Harley (1992). Output Growth and the British Industrial Revolution: A restatementof the Crafts-Harley View. Economic History Review 45, 703–730.

Crouzet, F. (2003). The Historiography of French Economic Growth in the Nineteenth Century. EconomicHistory Review 62(2), 215–242.

Darby, H. C. and H. Fullard (1970). The New Cambridge Modern History, Volume 14. Cambridge: Cam-bridge University Press.

Darnton, R. (1973). The Encyclopedie Wars of Prerevolutionary France. The American Historical Re-view 78(5), 1331–1352.

35

Darnton, R. (1979). The business of enlightenment: A publishing history of the Encyclopédie, 1775–1800.Cambridge, MA: Belknap-Harvard University Press.

de la Croix, D. and O. Licandro (2012). The Longevity of Famous People from Hammurabi to Einstein.CORE Discussion Paper 2012/52.

de Tocqueville, A. (1856). The Old Regime and the French Revolution. Garden City, N.Y.: Doubleday.Reprinted in 1955.

Delefortrie, N. and J. Morice (1959). Les Revenus departmentaux en 1864 et en 1954. Paris, France: Colin.

DeLong, J. B. and A. Shleifer (1993). Princes and Merchants: City Growth before the Industrial Revolution.Journal of Law and Economics 36, 671–702.

Dittmar, J. E. (2011). Information Technology and Economic Change: The Impact of the Printing Press.Quarterly Journal of Economics 126, 1133–1172.

Dunham, A. L. (1955). The Industrial Revolution in France, 1815–1848. Exposition Press, New York.

FBTEE (2012). The French Book Trade in Enlightenemnt Europe Database.

Febvre, L. and H.-J. Martin (1958). L’Apparition du Livre. Paris: Albin Michel.

Flora, P., F. Kraus, and W. Pfenning (1983). State, Economy, and Society in Western Europe 1815-1975:The Growth of Industrial Societies and Capitalist Economies, Volume 1. St. James Press, Chicago.

Fox, R. (1984). Presidential Address: Science, Industry, and the Social Order in Mulhouse, 1798-1871. TheBritish Journal for the History of Science 17(2), 127–168.

Furet, F. and J. Ozouf (1977). Lire et écrire. L’alphabétisation des Français de Calvin à Judes Ferry. Minuit,Paris.

Galor, O. (2005). From Stagnation to Growth: Unified Growth Theory. In P. Aghion and S. Durlauf (Eds.),Handbook of Economic Growth, Volume 1, Chapter 4, pp. 171–293. Elsevier.

Galor, O. (2011). Unified Growth Theory. Princeton University Press.

Gennaioli, N., R. La Porta, F. Lopez-de-Silanes, and A. Shleifer (2013). Human Capital and RegionalDevelopment. Quarterly Journal of Economics 128(1), 105–164.

Glaeser, E. L., R. L. Porta, F. L. de Silanes, and A. Shleifer (2004). Do Institutions Cause Growth? Journalof Economic Growth 9(3), 271–303.

Goreaux, L. M. (1956). Les migrations agricoles en France depuis un siécle et leur relation avec certainsfacteurs économiques. Études et Conjoncture 11, 327–376.

Graham, H. (2011). Humble Petitioners? Seigneurial Courts, Royal Justice and the Role of Litigants in theEighteenth Century. French History & Civilization 4, 12–21.

Hall, R. E. and C. I. Jones (1999). Why Do Some Countries Produce So Much More Output Per WorkerThan Others? Quarterly Journal of Economics 114(1), 83–116.

Hamscher, A. N. (2012). The Royal Financial Administration and the Prosecution of Crime in France,1670–1789. Newark: University of Delaware Press.

Hansen, G. and E. Prescott (2002). Malthus to Solow. American Economic Review 92(4), 1205–1217.

Hanushek, E. A. and D. D. Kimko (2000). Schooling, Labor-Force Quality, and the Growth of Nations.American Economic Review 90(5), 1184–1208.

36

Hanushek, E. A. and L. Woessman (2012). Do Better Schools Lead to More Growth? Cognitive Skills,Economic Outcomes, and Causation. Journal of Economic Growth 17(4), 267–321.

Hanushek, E. A. and L. Woessmann (2008). The Role of Cognitive Skills in Economic Development.Journal of Economic Literature 46(3), 607–668.

Hau, M. (2012). The Invention of Enterprise: Entrepreneurship from Ancient Mesopotamia to ModernTimes, Chapter Entrepreneurship in France, pp. 305–330. Princeton University Press.

Horn, J. (2006). The Path Not Taken: French Industrialization in the Age of Revolution, 1750-1830. Cam-bridge MA: MIT Press.

Horn, J. and M. C. Jacob (1998). Jean-Antoine Chaptal and the Cultural Roots of French Industrialization.Technology and Culture 39(4), 671–698.

Hornung, E. (2014). Immigration and the Diffusion of Technology: The Huguenot Diaspora in Prussia.American Economic Review 104(1), 84–122.

ISTC (2008). Incunabula Short-Title Catalogue. London: British Library.

Jedin, H., K. S. Latourette, and J. Martin (1970). Atlas Zur Kirchengeschichte. Freiburg: Herder.

Kelly, M., J. Mokyr, and C. Ó Grada (2014). Precocious Albion: A New Interpretation of the BritishIndustrial Revolution. Annual Reviews of Economics 6. forthcoming.

Kessler, A. D. (2010). Marginalization and Myth: The Corporatist Roots of France’s Forgotten ElectiveJudiciary. American Journal of Comparative Law 58, 679–720.

Komlos, J. (1998). Shrinking in a Growing Economy? The Mystery of Physical Stature during the IndustrialRevolution. Journal of Economic History 58(3), 779–802.

Komlos, J. (2005). Height of French Soldiers, 1716–1784. [Computer file] ICPSR04363-v1.

Komlos, J. and J. Baten (1998). The Biological Standard of Living in Comparative Perspective. FranzSteiner, Stuttgart.

Krueger, A. B. and M. Lindahl (2001). Education for Growth: Why and for Whom? Journal of EconomicLiterature 39(4), 1101–1136.

Livingstone, D. N. (2003). Putting Science in its Place: Geographies of Scientific Knowledge. ChicagoUniversity Press.

Maddison, A. (2001). The World Economy. A Millennial Perspective. Paris: OECD.

Maddison, A. (2007). Historical Statistics. University of Groningen. Webpage:http://www.ggdc.net/maddison/.

Mankiw, N. G., D. Romer, and D. N. Weil (1992). A Contribution to the Empirics of Economic Growth.Quarterly Journal of Economics 107(2), 407–37.

Marczewski, J. (1961). Some Aspects of the Economic Growth of France 1660-1958. Economic Develop-ment and Cultural Change 9(3), 369–386.

McClellan, J. E. (1985). Science Reorganized: Scientific Societies in the Eighteenth Century. New York:Columbia University Press.

McEvedy, C. and R. Jones (1978). Atlas of World Population History, Facts on File. New York.

Mitch, D. (1993). The Role of Human Capital in the First Industrial Revolution. In J. Mokyr (Ed.), TheBritish Industrial Revolution: An Economic Perspective, pp. 267–307. Boulder: Westview.

37

Mokyr, J. (1990). The Lever of Riches. Oxford: Oxford University Press.

Mokyr, J. (2000). Knowledge, Technology, and Economic Growth During the Industrial Revolution. InB. V. Ark and G. Kuper (Eds.), Technology and Productivity Growth. Hague: Kluwert. Available athttp://faculty.wcas.northwestern.edu/~jmokyr/Groningen.pdf.

Mokyr, J. (2002). The Gifts of Athena: Historical Origins of the Knowledge Economy. Princeton, NJ:Princeton University Press.

Mokyr, J. (2005a). Long-Term Economic Growth and the History of Technology. Volume 1 of Handbookof Economic Growth, pp. 1113–80. Elsevier, Amsterdam.

Mokyr, J. (2005b). The Intellectual Origin of Modern Economic Growth. Journal of Economic His-tory 65(2), 285–351.

Mokyr, J. and H.-J. Voth (2009). Understanding Growth in Early Modern Europe. In S. Broadberry andK. O’Rourke (Eds.), The Cambridge Economic History of Europe. Cambridge: Cambridge UniversityPress.

Moser, P. (2005). How Do Patent Laws Influence Innovation? Evidence from Nineteenth-Century World’sFairs. American Economic Review 95(4), 1214–1236.

Moser, P. (2012). Innovation without Patents: Evidence from World’s Fairs. Journal of Law and Eco-nomics 55(1), 43–74.

Mours, S. (1958). Les Églises Réformées en France. Tableaux et Cartes. Librairie Protestante.

Mousnier, R. (1979). The Institutions of France under the Absolute Monarchy, 1598–1789. Chicago:University of Chicago Press.

Muessig, U. (2012). Judges and Judging in the History of the Common Law and Civil Law, Chapter Su-perior courts in early-modern France, England and the Holy Roman Empire, pp. 209–233. CambridgeUniversity Press.

Murphy, T. E. (2010). Old Habits Die Hard (Sometimes): What can Département Heterogeneity Tell usabout the French Fertility Decline? IGIER Working Paper 364.

Musson, A. E. and E. Robinson (1969). Science and Technology in the Industrial Revolution. ManchesterUniversity Press.

Nelson, R. R. and E. S. Phelps (1966). Investment in Humans, Technological Diffusion, and EconomicGrowth. American Economic Review 56(2), 69–75.

North, D. C. and R. P. Thomas (1973). The Rise of the Western World: A New Economic History. CambridgeUK: Cambridge University Press.

Nunn, N. and N. Qian (2011). The Potato’s Contribution to Population and Urbanization: Evidence from anHistorical Experiment. Quarterly Journal of Economics 126(2), 593–650.

Nuvolari, A. and V. Tartari (2011). Bennet Woodcroft and the value of English patents, 1617–1841. Explo-rations in Economic History 48, 97–115.

Nye, J. V. (1987). Firm Size and Economic Backwardness: A New Look at the French IndustrializationDebate. The Journal of Economic History 47(3), 649–669.

O’Rourke, K. H., A. S. Rahman, and A. M. Taylor (2008). Luddites and the Demographic Transition. NBERworking paper 14484.

38

O’Rourke, K. H. and J. G. Williamson (1995). Education, Globalization and Catch-up: Scandinavia in theSwedish Mirror. Scandinavian Economic History Review 43(3), 287–309.

Roche, D. (1998). France in the Enlightenment. Harvard University Press, Cambridge, MA.

Roehl, R. (1976). French Industrialization: A Reconsideration. Explorations in Economic History 13,223–281.

Rostow, W. W. (1975). How it All Began: Origins of the Modern Economy. Methuen and Co., London.

Scoville, W. C. (1953). The Huguenots in the French Economy, 1650-1750. Quarterly Journal of Eco-nomics 67(3), 423–444.

Service de la Statistique Général de France (1878). Annuaire Statistique de la France. Paris, ImprimerieNationale.

Smith, J. M. (2006a). The French Nobility in the Eighteenth Century: Reassessments and New Approaches.University Park, Pa.: Pennsylvania State University Press.

Smith, M. S. (2006b). The Emergence of Modern Business Enterprise in France, 1800–1930. HarvardUniversity Press.

Sprandel, R. (1968). Das Eisengewerbe im Mittelalter. A. Hiersemann.

Steckel, R. H. (1983). Height and Per Capita Income. Historical Methods 16, 1–7.

Steckel, R. H. (2008). Biological Measures of the Standard of Living. Journal of Economic Perspec-tives 22(1), 129–152.

Taylor, A. M. (1999). Sources of Convergence in the Late Nineteenth Century. European Economic Re-view 43(9), 1621–1645.

Tilly, C. (1990). Coercion, Capital, and European States, 990–1990. London: Blackwell.

Verley, P. (1985). La Révolution Industrielle. MA Editions, Paris.

Vogt, W. P. (1982). Paideia for Profit. History of Education Quarterly 22(1), 89–98.

Voigtländer, N. and H.-J. Voth (2012). Persecution Perpetuated: The Medieval Origins of Anti-SemiticViolence in Nazi Germany. Quarterly Journal of Economics 127(3), 1339–1392.

39

FIGURES

Subscriber density

no data[0 − 0.2](0.2 − 1.0](1.0 − 3.5](3.5 − 15.2]

Subs. per 1,000

Literacy in 1786

no data[0 − 0.3](0.3 − 0.5](0.5 − 0.7](0.7 − 0.9]

Literacy 1786

Figure 1: Encyclopedia subscriber density and literacy rates

Notes: The left panel shows the spatial distribution of encyclopedia subscribers per 1,000 city inhabitantsin the second half of the 18th century. The right panel shows the distribution of literacy rates (percentage ofmales signing their marriage certificate) across French departments in 1786. Both variables are describedin detail in Section 4.1. Figure A.4 in the appendix plots the two variables against each other, show thatthey are not correlated.

40

1400 – 1750

−.9

−.6

−.3

0.3

.6.9

City

gro

wth

(re

sidu

al)

−1 −.5 0 .5 1 1.5Subscriber density (residual)

coef = .037, (robust) se = .034, t = 1.10

1750 – 1850

−.9

−.6

−.3

0.3

.6.9

City

gro

wth

(re

sidu

al)

−1 −.5 0 .5 1 1.5Subscriber density (residual)

coef = .218, (robust) se = .051, t = 4.28

Figure 2: Encyclopedia subscriptions and city growth – before and after 1750

Notes: The figure plots average annual city growth in France against encyclopedia subscriber density(lnSubDens), after controlling for our baseline controls (those listed in the left panel of Table 1). The leftpanel uses the pre-industrial period 1400-1750. The right panel examines the same cities over the periodof French industrialization, 1750-1850. Average annual city growth is more than twice as high in the latterperiod, from 0.23% to 0.48%.

Pre-industrial (1750)

0 0.05 0.1 0.15 0.20

0.05

0.1

0.15

0.2

0.25

0.3

Urb

. rat

e / r

elat

ive

wag

e gr

owth

Share of entrepreneus with scientific knowledge (sn)

0 0.05 0.1 0.15 0.21

1.05

1.1

1.15

1.2

1.25

1.3

rela

tive

wag

e

Urbanization rate

relative wage

relative wage growth

Industrial period (1850)

0 0.05 0.1 0.15 0.20

0.05

0.1

0.15

0.2

0.25

0.3

Urb

. rat

e / r

elat

ive

wag

e gr

owth

Share of entrepreneus with scientific knowledge (sn)

0 0.05 0.1 0.15 0.21

1.05

1.1

1.15

1.2

1.25

1.3

rela

tive

wag

eUrbanization rate

relative wage growth

relative wage

Figure 3: Scientific knowledge and economic development

Notes: The figure illustrates how the share of entrepreneurs with scientific knowledge in region n, sn,affects urbanization, wages, and economic growth. The left panel refers to the pre-industrial period (illus-trating model Prediction 1), and the right panel illustrates Prediction 2, referring to the industrial period.The urbanization rate corresponds to the labor share in manufacturing. Wages (right axis) are reportedrelative to regions without scientific knowledge (sn = 0). Relative wage growth (left axis) is measured asannual percentage growth in region n, net of growth in regions with sn = 0.

41

1 1.05 1.1 1.15 1.2 1.25 1.30

0.05

0.1

0.15

0.2

0.25

0.3

Urb

. rat

e / r

elat

ive

wag

e gr

owth

Worker skills hn

1 1.05 1.1 1.15 1.2 1.25 1.31

1.05

1.1

1.15

1.2

1.25

1.3

rela

tive

wag

e

relative wage growth

Urbanization rate

relative wage

Figure 4: The role of worker skills

Notes: The figure illustrates model Prediction 3, showing how worker skills in region n, hn, affect ur-banization, wages, and economic growth. Since the effect of worker skills does not change over time, thefigure illustrates both the pre-industrial period and the industrial period. See Figure 3 for a description ofthe three depicted variables.

0.2

.4.6

.81

Ker

nel d

ensi

ty

−.5 0 .5 1 1.5City growth 1750−1850

No subscriptionsSubscriptions p.c. > 0, below−medianSubscriptions p.c. > 0, above−median

Figure 5: Subscriptions and city growth, 1750-1850

Notes: The figure shows the Kernel density of city growth over theperiod 1750-1850 for three subsets of French cities: 108 cities withoutencyclopedia subscriptions, as well as 43 (42) cities with above-zerosubscriptions and below-median (above-median) subscriber density.

42

TABLES

Table 1: Correlations with subscriber density (lnSubDens)

Baseline Controls Worker skills and additional controls

(1) (2) (3) (4)‡ (5)‡

Cities included: All Subs>0 Departments included: All All Subs>0

Population in 1750 0.374∗∗∗ -0.234∗∗ Literacy 1686 0.551 0.905 0.026(0.073) (0.110) (0.593) (0.650) (0.777)

Atlantic Port 0.081 -0.222 Literacy 1786 0.290 0.358 0.054(0.207) (0.213) (0.345) (0.323) (0.383)

Mediterranean Port 0.022 -0.129 School Rate 1837 0.363 0.313 0.200(0.276) (0.223) (0.350) (0.335) (0.412)

Navigable River 0.422∗∗ -0.167 lnSTNBooksDens 0.232∗∗∗ 0.212∗∗∗ 0.163∗∗∗

(0.202) (0.210) (0.048) (0.060) (0.061)

Non French-Speaking -0.376∗∗ -0.719∗∗ Pays d’Eléction -0.037 0.006 -0.238(0.146) (0.297) (0.185) (0.204) (0.249)

lnProtoIndDens -0.027 -0.033 -0.546Early Knowledge Controls (0.830) (0.829) (0.846)

University 1.030∗∗∗ 0.316 lnNoblesDens 0.233 0.253 0.736∗∗∗

(0.186) (0.194) (0.278) (0.301) (0.249)

Printing Press 0.712∗∗∗ 0.203 lnExecuteDens 0.272∗∗ 0.210 0.247(0.170) (0.182) (0.117) (0.138) (0.148)

ln(Books Printed 1500) 0.171∗∗∗ 0.039(0.061) (0.066)

Observations 193 85 75/88 75/88 59/64

Notes: The table shows the coefficients of individual regression of subscriber density to the Quarto edition of theEncyclopédie, lnSubDens, on a variety of city characteristics. lnSubDens is computed as described in Section 4.1.Population in 1750 measures urban population (in thousands) for the cities in our sample. Atlantic Port, Mediter-ranean Port and Navigable River are dummies taking value 1 for cities with ports on the Atlantic Ocean or on theMediterranean Sea or located on a navigable river. Non French Speaking is a dummy for six French departmentswho spoke a language other than French. University is a dummy taking value 1 for cities which hosted a Universitybefore 1750. Printing Press is a dummy taking value 1 for cities where a printing press was established before 1500.Ln(Books Printed 1500) represents the log number of editions printed before 1501. Additional controls comprise thefollowing variables (see Section 4.3 for sources and detail): Literacy in 1686 and 1786 measure the percentage ofmen signing their wedding certificate in the respective year. SchoolRate in 1837 measures the ratio of students toschool-age population (5 to 15 years) in 1836-37. lnSTNBooksDens represents the density of book sales to Francefrom the Swiss publishing house Société Typographique de Neuchâtel (STN). Pays d’élection are regions where theFrench king exerted particularly strong control. lnProtoIndDens is an index of proto-industrialization in France thatincludes the number of mines, forges, iron trading locations, and textile manufactures before 1500. lnNoblesDensreflects the density of noble families in each French department. lnExecuteDens measures the density of peopleexecuted during the reign of Terror. In columns 1-2, all regressions are run at the city level, in columns 3-5 at thedepartment level. Robust standard errors in parentheses. * p<0.1, ** p<0.05, *** p<0.01.‡Includes baseline controls.

43

Table 2: Matching estimation by city size and location

Dependent variable: log city growth over the indicated period

(1) (2) (3) (4)

PANEL A: Period 1750-1850City size percentiles incl.: All 10-90 pct All 10-90 pct

ISubs>0 0.146∗∗ 0.155∗∗ 0.267∗∗∗ 0.163∗∗

(0.07) (0.07) (0.07) (0.07)

Matching variablesPopulation X X X XLocation X X

Observations 177 154 167 144

PANEL B: Pre-17501700-1750 1600-1700 1500-1600 1400-1500

ISubs>0 0.087 0.034 0.181 -0.567∗∗∗

(0.06) (0.17) (0.17) (0.21)

Matching variablesPopulation X X X XLocation X X X X

Observations 129 58 43 37

Notes: All regressions are run by propensity score matching at the city level, excluding the 1% of log city growthoutliers, and using the three nearest neighbors when there are more than 50 observations and the two nearest neighborsotherwise. Treatment variables is the indicator variable ISubs>0, which takes on value 1 if a city had above-zerosubscriptions to the encyclopedia (see the notes to Table 1 and Section 4.1 for detail). Columns 1 and 2 in Panel Ause city population as matching variable. Columns 3 and 4, as well as Panel B, add geographic longitude and latitude(location) as matching variables. Standard errors in parentheses. * p<0.1, ** p<0.05, *** p<0.01.

44

Table 3: Encyclopedia subscriptions and city growth, before and after 1750

Dependent variable: log city growth over the indicated periodPeriod 1750-1850 Pre-1750

(1) (2) (3) (4) (5) (6) (7) (8)

[unweighted] 1700-1750 1600-1700 1500-1600 1400-1500

PANEL A: All citieslnSubDens 0.100∗∗ 0.171∗∗∗ 0.169∗∗∗ 0.204∗∗∗ 0.008 0.060 0.056 0.115

(0.039) (0.036) (0.033) (0.036) (0.037) (0.119) (0.087) (0.167)

lnPopinitial 0.055∗∗∗ -0.085∗∗ -0.089∗ -0.156∗∗∗ -0.058 -0.456∗∗ -0.376∗∗∗ -0.287∗∗

(0.014) (0.041) (0.048) (0.051) (0.040) (0.186) (0.087) (0.134)

Atlantic Port 0.221∗∗∗ 0.242∗∗ 0.349∗∗ 0.087 -0.124 0.372∗∗ 0.256(0.082) (0.094) (0.162) (0.101) (0.239) (0.145) (0.272)

Mediterranean Port 0.779∗∗∗ 0.794∗∗∗ 0.752∗∗∗ -0.203∗∗ 0.784∗ 0.309∗ 0.716∗∗

(0.076) (0.091) (0.142) (0.094) (0.398) (0.161) (0.277)

Navigable River 0.095 0.068 0.134∗ 0.001 0.222 -0.017 0.221(0.068) (0.072) (0.069) (0.076) (0.191) (0.131) (0.291)

Paris 0.575∗∗∗ 0.610∗∗∗ 0.760∗∗∗ -0.020 0.638 1.237∗∗∗ 0.402(0.136) (0.132) (0.171) (0.135) (0.478) (0.321) (0.582)

Non French Speaking 0.337∗∗∗ 0.330∗∗∗ 0.428∗∗∗ 0.100 -0.438 0.078 0.156(0.089) (0.097) (0.145) (0.129) (0.376) (0.272) (0.373)

University -0.063 -0.123 0.122∗ 0.299∗ -0.108 -0.049(0.067) (0.084) (0.069) (0.173) (0.117) (0.271)

Printing Press in 1500 0.093 0.188∗ -0.078 -0.628∗∗ 0.448∗∗ -0.063(0.094) (0.098) (0.083) (0.272) (0.182) (0.351)

ln(Books Printed 1500) -0.001 0.006 0.029∗ 0.162∗∗ -0.040 0.020(0.020) (0.025) (0.017) (0.065) (0.040) (0.067)

R2 0.12 0.36 0.36 0.27 0.17 0.55 0.53 0.31Observations 193 193 193 193 148 56 45 39

PANEL B: Only cities with positive subscriptionslnSubDens 0.090 0.135∗∗∗ 0.117∗∗ 0.086∗ -0.069 -0.061 0.018 0.175

(0.063) (0.051) (0.045) (0.044) (0.053) (0.129) (0.094) (0.213)

Controls X X X X X X X XR2 0.08 0.46 0.48 0.38 0.32 0.64 0.64 0.30Observations 85 85 85 85 76 44 36 32

Notes: All regressions are run at the city level and are weighted (except for Column 4) by initial population of therespective period. The dependent variable in the different columns is log city population growth over the periodindicated in the header. The main period of French industrialization is 1750-1850; city growth in earlier periods isused as a placebo. ‘Controls’ are the same as in the corresponding columns of Panel A. For details on lnSubDensand the other explanatory variables see the notes to Table 1, Section 4 and Appendix C.4. Robust standard errors inparentheses. * p<0.1, ** p<0.05, *** p<0.01.

45

Table 4: Literacy and additional controls

Dependent variable: log city growth, 1750-1850

(1) (2) (3) (4) (5) (6) (7)

lnSubDens 0.180∗∗∗ 0.198∗∗∗ 0.187∗∗∗ 0.176∗∗∗ 0.176∗∗∗ 0.179∗∗∗ 0.187∗∗∗

(0.040) (0.042) (0.040) (0.038) (0.041) (0.040) (0.042)Literacy 1786 -0.209 -0.156 -0.190 -0.276∗∗ -0.208 -0.185 -0.192

(0.142) (0.135) (0.143) (0.133) (0.144) (0.144) (0.138)lnSTNBooksDens -0.025 -0.020

(0.021) (0.021)Pays d’Eléction -0.076 -0.043

(0.065) (0.069)lnProtoIndDens 1.107∗∗∗ 0.952∗∗

(0.363) (0.391)lnNoblesDens 0.085 0.129

(0.135) (0.119)lnExecuteDens 0.037 0.030

(0.037) (0.039)ln(Pop 1750) -0.075∗ -0.053 -0.086∗ -0.067 -0.061 -0.072 -0.033

(0.043) (0.041) (0.045) (0.040) (0.055) (0.044) (0.048)Baseline Controls X X X X X X XR2 0.38 0.39 0.39 0.40 0.38 0.39 0.41Observations 166 166 164 166 166 166 164

Notes: All regressions are run at the city level and are weighted by city population in 1750. The dependent variable islog city population growth in 1750-1850. ‘Baseline Controls’ include all variables listed in Table 1 (left panel). Fordetails on lnSubDens and the other explanatory variables see notes to Table 1, Section 4 and Appendix C.4. Robuststandard errors in parentheses (clustered at the department level). * p<0.1, ** p<0.05, *** p<0.01.

46

Table 5: Panel regressions with city population, 1500-1850

Dependent variable: log city population

(1) (2) (3) (4) (5) (6)

Baseline: treatment post-1750 Placebo periods y[balanced] 1600 1700

lnSubDensxPost1750 0.106∗∗∗ 0.136∗∗∗ 0.109∗∗ 0.164∗∗ 0.146∗∗∗ 0.158∗∗∗

(0.029) (0.031) (0.042) (0.070) (0.029) (0.033)lnSubDensxPosty -0.039 -0.043

(0.077) (0.044)Baseline Controls X X X X XAdditional Controls XCity FE X X X X X XTime Period FE X X X X X XR2 0.86 0.87 0.88 0.85 0.88 0.87Observations 846 846 722 270 846 846

Notes: All regressions are run at the city level. The dependent variable is the log of city population in the years 1500,1600, 1700, 1750, 1800, and 1850. The Post indicator variable takes value zero for the periods 1500-1750, and onefor the periods 1800 and 1850. ‘Baseline Controls’ include all variables listed in Table 1 (left panel); ‘AdditionalControls’ are those used in in Table 4. For details on lnSubDens and controls see the notes to Table 1, Section 4 andAppendix C.4. Robust standard errors in parentheses (clustered at the department level in col 3). * p<0.1, ** p<0.05,*** p<0.01.

Table 6: City growth over the sub-periods 1750-1800 and 1800-1850

Dependent variable: log city growth over the indicated period

(1) (2) (3)Period 1750-1800 1800-1850 1800-1850

Control for prior growth

lnSubDens 0.108∗∗∗ 0.058∗∗ 0.073∗∗∗

(0.035) (0.026) (0.027)

Growth 1750-1800 -0.141∗

(0.082)

Baseline Controls X X XR2 0.29 0.41 0.42Observations 192 192 192

Notes: All regressions are run at the city level over the period indicated in the table header. ‘Baseline Controls’include all variables listed in Table 1 (left panel), as well as log population in 1750. Col 4 uses predicted city growthfrom the regression in col 1. For details on lnSubDens and controls see the notes to Table 1, Section 4 and AppendixC.4. Robust standard errors in parentheses. * p<0.1, ** p<0.05, *** p<0.01.

47

Table 7: Scientific societies and city growth

Dependent variable: log city growth over the indicated period

(1) (2) (3) (4) (5) (6)

1750-1850 1750-1850 1700-1750 1600-1700 1500-1600 1400-1500

PANEL A: Matching EstimationIScient.Scociety>0 0.204∗∗ 0.193∗∗ 0.028 0.167 0.195 -0.324

(0.098) (0.095) (0.083) (0.151) (0.137) (0.220)

Matching variablesPopulation X X X X X XLocation X X X X XObservations 185 175 136 58 44 41

PANEL B: OLS EstimationlnMembDens 0.171∗ 0.287∗∗∗ 0.069 0.058 0.211 -0.270

(0.088) (0.080) (0.105) (0.190) (0.179) (0.474)Baseline Controls X X X X XR2 0.10 0.32 0.19 0.56 0.53 0.32Observations 185 185 140 52 41 37

Notes: All regressions are run at the city level. The dependent variable in the different columns is log city populationgrowth over the period indicated in the header. In Panel A, all regressions are run by propensity score matching,excluding the 1% of log city growth outliers, and using the three nearest neighbors when there are more than 50observations and the two nearest neighbors otherwise. Treatment variable is the indicator IScient.Society>0, whichtakes on value 1 if a city hosted a scientific society before 1750 (see Section 4 for details). Column 1 uses citypopulation as matching variable. Columns 2–6 add geographic longitude and latitude (location) as matching variables.In Panel B, all regressions are run by OLS and are weighted by initial population of the respective period. ‘BaselineControls’ include all variables listed in Table 1 (left panel). In Column 1, only Pop 1750 is included as control. Fordetails on lnMembDens and controls see the notes to Table 1, Section 4 and Appendix C.4. Standard errors inparentheses.* p<0.1, ** p<0.05, *** p<0.01.

48

Table 8: Soldiers height before and after 1750

Dependent variable: Soldier height in cm

(1) (2) (3) (4) (5) (6)

Period 1819-1826 Pre-1750

lnSubDens 0.416∗∗∗ 0.450∗∗∗ 0.403∗∗∗ 0.428∗∗∗ 0.116 0.120(0.136) (0.129) (0.139) (0.131) (0.115) (0.116)

Literacy 2.805∗∗∗ 3.056∗∗∗ 2.463∗∗∗ 2.810∗∗∗ 1.043∗ 0.982∗

(0.398) (0.360) (0.459) (0.409) (0.525) (0.549)Height pre-1750 0.303 0.245

(0.193) (0.171)Baseline Controls X X XR2 0.42 0.60 0.46 0.56 0.06 0.16Observations 77 77 77 77 74 74

Notes: All regressions are run at the French department level and include a dummy for Paris (Department Seine). Thedependent variable in cols 1-4 is soldier height as reported by Aron et al. (1972). In Columns 5-6, the dependentvariable is average soldier height recorded over the period 1716-49 and collected by Komlos (2005). To account forvariation in height and soldier age within this period, we control for age, age squared, and birth decade (see AppendixC.3 for detail). We exclude departments with data for less than 20 soldiers (Table A.9 reports results when usingall departments and weighting by the number of soldiers). ‘Baseline Controls’ include all variables listed in Table 1(left panel); we use Literacy in 1786 in Colums 1-4 and Literacy in 1686 in Columns 5-6. For details on lnSubDens,Literacy and controls see the notes to Table 1, Section 4 and Appendix C.4. Original city-level variables are aggregatedto the department level as described in Appendix C.5. Robust standard errors in parentheses. * p<0.1, ** p<0.05, ***p<0.01.

Table 9: Disposable income, employment shares and wages

(1) (2) (3) (4) (5) (6) (7)

Dep. var. ln(disposable Employment shares Wages

income) Agric. Industry Agric. Industry Textile Food Proc.

lnSubDens 0.073∗∗ -0.030∗∗ 0.017∗ 0.039 0.054∗∗∗ 0.067∗∗ 0.008(0.031) (0.015) (0.010) (0.026) (0.017) (0.028) (0.031)

School Rate 1837 0.208∗ -0.179∗∗∗ 0.104∗∗∗ 0.411∗∗∗ 0.200∗∗∗ 0.325∗∗∗ 0.355∗∗∗

(0.118) (0.056) (0.037) (0.084) (0.056) (0.106) (0.094)Population Density 1876 -0.024 -0.199∗∗∗ 0.131∗∗∗ 0.018 0.036 0.184∗∗∗ 0.105

(0.044) (0.032) (0.028) (0.047) (0.038) (0.039) (0.075)Baseline Controls X X X X X X XR2 0.33 0.67 0.55 0.51 0.49 0.42 0.32Observations 84 85 85 79 79 83 84

Notes: All regressions are run at the French department level and include a dummy for Paris (Department Seine).‘Baseline Controls’ include all variables listed in Table 1 (left panel). For details on lnSubDens and controls see thenotes to Table 1, Section 4 and Appendix C.4. Original city-level variables are aggregated to the department level asdescribed in Appendix C.5. Robust standard errors in parentheses. * p<0.1, ** p<0.05, *** p<0.01.

49

Table 10: Subscriber density and average local firm productivity in 1837

Dep. Var.: log wages (arrondissement average, for firms in ‘modern’ and ‘old’ sectors)

(1) (2) (3) (4) (5) (6) (7) (8)

Sector modern old modern old modern old modern old

lnSubDens 0.097∗∗∗ 0.032∗∗ 0.092∗∗∗ 0.033∗∗ 0.091∗∗∗ 0.027∗ 0.086∗∗∗ 0.025(0.013) (0.013) (0.015) (0.014) (0.019) (0.015) (0.019) (0.015)

School Rate 1837 0.252∗∗∗ 0.238∗∗∗ 0.236∗∗∗ 0.256∗∗∗ 0.230∗∗∗ 0.222∗∗∗ 0.223∗∗∗ 0.205∗∗∗

(0.076) (0.066) (0.072) (0.068) (0.083) (0.063) (0.077) (0.066)Establishment Size -0.023∗∗ 0.054∗∗∗ -0.021∗ 0.053∗∗∗ -0.019∗ 0.054∗∗∗ 0.010 0.044∗∗∗

(0.009) (0.008) (0.011) (0.009) (0.011) (0.010) (0.011) (0.011)Pop. Controls X X X X X X X XBaseline Controls X X X X X XAdditional Controls X X X XSector FE X XR2 0.22 0.22 0.27 0.27 0.30 0.28 0.43 0.32Observations 633 794 428 541 386 493 386 493

Notes: All regressions are run at the French arrondissement level and include a dummy for Paris (Department Seine).The dependent variable is the log of average male wages across all firms in a sector j in arrondissement n (see SectionC.10 for detail). ‘Baseline Controls’ include all variables listed in Table 1 (left panel). ‘Additional Controls’ are thoseused in in Table 4. For details on lnSubDens and controls see the notes to Table 1, Section 4 and Appendix C.4.Original city-level variables are aggregated to the arrondissement level as described in Appendix C.5. Standard errors(clustered at the department level) in parentheses. * p<0.1, ** p<0.05, *** p<0.01.

50

Table 11: Subscriber density and firm productivity within individual industries

(1) (2) (3) (4) (5)

Sector Name Sector Coefficient R2 Engines per 1,000type lnSubDens Obs. Steam Others

Transportation Equipment modern 0.114∗∗∗ 0.63 2 9(0.026) 39

Printing Technology, and modern 0.103∗∗∗ 0.20 1 26Scientific Instruments (0.021) 221

Textile and Clothing modern 0.067∗∗∗ 0.28 3 8(0.018) 298

Furniture and Lighting modern 0.056∗ 0.54 13 1(0.033) 75

Metal and Metal Products old 0.042∗ 0.16 6 34(0.024) 273

Leather old 0.041∗ 0.19 3 46(0.022) 165

Mining old 0.038∗∗ 0.31 3 15(0.018) 188

Ceramics and Glass old 0.003 0.27 4 5(0.020) 168

Notes: For each sector, column 1 specifies whether the sector belongs to the ‘modern’ or ‘old’ manufacturing classifi-cation (see Section 5.3 for detail). Sectors are ranked by the size of the coefficient of lnSubDens, reported in column2 – where the dependent variable is the logarithm of male wages at the arrondissement level. For each regression,column 3 reports the R2 and the number of observations. In all regressions, we control for ‘Size and Pop. Controls’,as in Table 10 (cols 1-2). For details on lnSubDens, and controls see the notes to Table 1, Section 4, and AppendixC.4 and C.10. Standard errors (clustered at the department level) in parentheses. * p<0.1, ** p<0.05, *** p<0.01.Columns 4-5 show the average number of steam engine and of other engines per 1,000 workers (see Appendix C.10for detail).

51

Table 12: Subscriber density and scientists

(1) (2) (3) (4) (5) (6)

‘Famous’ Scientists 1851 Innovation Exhibits

lnSubDens 0.044∗∗∗ 0.048∗∗∗ 0.033∗∗ 0.034∗∗ 0.034∗∗ 0.037∗

(0.014) (0.013) (0.015) (0.015) (0.016) (0.022)Baseline Controls X X X XAdditional Controls X XR2 0.13 0.15 0.18 0.18 0.33 0.39Observations 193 166 164 193 165 163

Notes: All regressions are run at the city level and include dummies for Paris and for non-French speaking depart-ments. The dependent variable in cols 1-3 are ‘famous’ scientists per capita. These are people listed in the IndexBio-Bibliographicus Notorum Hominum whose profession is related to science, mathematics, chemistry, or physics.These data are from de la Croix and Licandro (2012). In total, there 574 ‘famous’ scientists listed for France over theperiod 1000–1887. The dependent variable in cols 4-6 are innovations per capita exhibited at the London world fair in1851. These data are from Moser (2005). For details on lnSubDens and controls see the notes to Table 1 and Section4. Robust standard errors in parentheses. * p<0.1, ** p<0.05, *** p<0.01

Table 13: City growth and historical determinants of upper tail knowledge

Huguenots and subscriber density vs. literacy

(1) (2) (3) (4) (5) (6)

Dep. Var.: SubDens Literacy Log City Growth1750-1850 1700-1750

lnHuguenotsDens 0.145∗∗∗ 0.151∗∗∗ 0.016 0.053∗∗∗ 0.023 -0.013(0.034) (0.040) (0.016) (0.019) (0.022) (0.022)

Literacy 1786 0.291 0.010 -0.047(0.251) (0.153) (0.155)

lnSubDens 0.198∗∗∗

(0.052)Baseline Controls X X X XR2 0.10 0.25 0.04 0.18 0.26 0.11Observations 163 148 240 148 148 116

Notes: All regressions are run at the city level and include dummies for Paris and for non-French speaking departments.For details on lnSubDens, Literacy and controls see the notes to Table 1 and Section 4. Standard errors (clustered atthe department level) in parentheses. * p<0.1, ** p<0.05, *** p<0.01‡ We use Literacy in 1786 in Colums 1, 2 and 4, and Literacy in 1686 in Column 6.

52

Online Appendix

Human Capital and Industrialization:Evidence from the Age of Enlightenment

Mara P. Squicciarini Nico VoigtländerKULeuven and FWO UCLA and NBER

A Model: Technical Results and Proofs

A.1 Aggregate Manufacturing Production Function

Demand for intermediate product i in region n follows from (3) and is given by:

zn(i) = ξAM,n(i)hβMn LM,n

(1

p(i)

) 11−αM

(A.1)

Since each entrepreneur i faces a constant marginal cost of one unit of the final good, his totalcosts are given by 1 · zn(i). Consequently, entrepreneurs choose their price p(i) so as to maximizeprofits πn(i) = (p(i)− 1) · zn(i), subject to (A.1). This yields:

p(i) =1

αM

, ∀i (A.2)

Using this in (3) and choosing the constant ξ ≡ α− 2·αM

1−αMM implies that total manufacturing output in

region n is given by

YM,n = AM,nhβMn LM,n, with AM,n ≡

∫ 1

0

AM,n(i)di (A.3)

Consequently, aggregate manufacturing productivity is a simple linear combination of individualentrepreneurial efficiencies. Note from (3) that a share 1 − αM of total output is paid to labor:wM,nLM,n = (1− αM)YM,n, so that the wage rate in region n is given by

wM,n = (1− αM)AM,nhβMn (A.4)

This completes the set of equations used in the main analysis. For completeness, we also presentthe total profits by entrepreneurs, as well as output net of intermediate inputs. Total profits are

Appendix p.1

given by Πn =∫ 1

0πn(i)di, with πn(i) given above. Substituting (A.1) and (A.2), we obtain:

Πn = αM(1− αM)YM,n (A.5)

Finally, note that the measure of total output given by (A.3) still includes the part that is used forintermediate production. Net output is given by Y net

M,n = YM,n −∫ 1

0z(i)di. Using (A.1) and (A.2)

this yields:Y netM,n =

(1− α2

M

)YM,n (A.6)

It is straightforward to verify that total wage payments and profits add up to Y netM,n.

A.2 The Evolution of Productivity in Manufacturing

Aggregate productivity follows from (A.3), by integrating equation (6) from the paper over allentrepreneurs i ∈ [0, 1]. Without loss of generality, we assume that entrepreneurs i ≤ sn havescientific knowledge, and the remainder does not. This yields:

AM,n,t =

∫ sn

0

AM,n,t(τ(i) = 1)di+

∫ 1

sn

AM,n,t(τ(i) = 0)di (A.7)

Substituting for AM,n,t(i) from (6) then implies:

AM,n,t = ηAt + (1− η)AM,n,t−1

(1 + sn · γA,t

)(A.8)

Appendix p.2

B French Institutions and Centralization

As we argued in the introduction, the fact that we use variation within France avoids many of theproblems common to cross-country studies. In particular, since France was a highly centralized(and, until 1789, absolutist) state, regional variation in institutions or the rule of law is unlikely toaffect our results. In the following, we provide a more detailed historical argument to support thisclaim. Braudel (1982) argues that

[B]y the thirteenth century, France was already the major political unit of the conti-nent...having all the requisite ancient and modern characteristics of a state: the charis-matic aura, the judicial, administrative and above all financial institutions, without whichthe political unit would have been completely inert. (Braudel, 1982, p.323)

The judicial system was also highly centralized, as illustrated in Figure ++. The king withhis conseil was ultimately at the top of the royal judicial system. At the next level, there wereabout 15 regional and provincial parliaments and councils. Below these followed approximately400 local royal courts – called bailliages in northern France and sénéchaussées in southern France.These were responsible for appeals for non-nobles and ecclesiastics, and they were also the firstinstance for disputes concerning nobles. Finally, the bottom of the pyramid of royal jurisdictionswas represented by about a thousand lower courts (Hamscher, 2012, p.11), whose names differedacross regions (e.g., prévôtés, vicomtés, or châtellenies).

The judges of the royal justice system were generally fidéles to the king (Mousnier, 1979,p.354), received the same training, and were closely monitored. The same is true for those of thelower local courts – the seigneurial jurisdictions.1 Seigneurial courts were ultimately subject to theking’s justice, and their sentences could be appealed in royal tribunals (Hamscher, 2012). Graham(2011) observes that

[J]udges, lawyers and court officials received the same sorts of training during the eigh-teenth century, whether they ended up working in royal jurisdictions or seigneurial ones.Officials in the royal courts were reasonably effective in monitoring their seigneurial coun-terparts precisely because of their shared expertise and a concern to defend their mutualinterests. (Graham, 2011, p.16)

Similarly, Muessig (2012, p.212) concludes that "by the 14th century at the latest, royal jurisdiction

dominated, and the feudal justice seigneuriale became subordinate to royal jurisdiction." All thispoints to the conclusion that the institutional setting was homogenous among the different Frenchregions. In Graham’s (2011, p.16) words: "by the eighteenth century, there was in essence just one

1The seigneurial courts were appointed by lords, or seigneurs, and exercised a civil and criminal jurisdiction withinthe limit of the seigneurie.

Appendix p.3

system of justice in France: that of the king."

During the French Revolution, the judiciary system was largely reformed, and almost all an-

cien regime courts were abolished. The revolutionary program initially planned to institute a newsystem relying mostly on informal mediation, rather than on formal law. However, after the Reignof Terror, this distaste for legal formality faded away, and lawyers and formal courts were re-instituted. A crucial change occurred during the Napoleonic period when key reforms were im-plemented, establishing the institutions that still characterize today’s French legal system.2 At thebottom of the private judicial system were the tribunaux de premiere instance, which had gen-eral jurisdiction over all civil and criminal matters. The next level was represented by the cours

d’appel, which handled appeals from the tribunaux de premiere instance. Finally, at the top of thejudicial system was the cour de cassation (Kessler, 2010).

Most important for our study, if France was centralized before the Revolution, it became evenmore so thereafter. Among the early observers, Alexis de Tocqueville emphasized that the Frenchgovernment was already "highly centralized" before 1789, and that the French Revolution resultedin "far fewer changes" than is usually supposed (Tocqueville (p. ++ix, 20, 32-41) de Tocqueville,1856). Tilly confirms this view, arguing that "revolutionaries installed one of the first systems of

direct rule ever to take shape in a large state" (Tilly, 1990, p. ++107-10).

C Data and Additional Empirical Results

C.1 Growth of per Capita Income in France, 1650-1900

Figure A.1 shows GDP per capita in France over the period 1650 to 1900. Overall, the dataconfirm Roehl’s (1976) assessment that there is no clear ‘take-off’ point. Nevertheless, growthbecame steady in the mid-18th century and only slowed down markedly during the decade afterthe French Revolution.

C.2 Subscriber Density: Distribution, Correlations, and Alternative Definitions

Figure A.2 shows the distribution of our main explanatory variable, subscriber density, for allcities (Panel A) and for cities with positive subscriptions (Panel B). In the following, we introducetwo alternative definitions. First, the left panel of Figure A.3 shows subscriptions per 1,000 cityinhabitants without taking logs. Comparing this with our main measure illustrates that taking logstightens the distribution, restricting the extent to which extreme values affect our results.

For cities without subscriptions, our baseline measure of subscriber density has a shortcoming:by assigning zero density throughout, it does not take into account city size. For example, the zero

2An important difference with theancien regime was the introduction of a distinct public judiciary system, whereprivate individuals could challenge state actions.

Appendix p.4

Figure A.1: Trend in per-capita GDP in France, 1670-1890

Notes: The figure shows the evolution of French GDP per capita over the 1670-1890 period. GDP per capita iscomputed using data on gross output (at current prices) and price deflators from Marczewski (1961), combined withpopulation data from McEvedy and Jones (1978). The unit of measurement is Francs (in 1905-1913 prices); incomein 1820 corresponds to approximately 1,100 International (1990) dollar, according to Maddison (2007)

subscriptions for 22,000 inhabitants in Arles (in the South of France) arguably reflect a thinnerdensity of scientific knowledge than Subs = 0 in the town of Saverne (in the North-East) with1,000 inhabitants in 1750. To exploit this additional information, we define an alternative variableas follows: lnSubDens2 n = ln [(Subsn + 1)/pop1750n ], where pop1750n is city population in 1750.This introduces variation across cities with zero subscriptions: lnSubDens2n is the smaller thelarger is city population.3 The resulting distribution is presented in the right panel of Figure A.3.Clearly, there is now more variation in the left part. Below in Appendix C.7, we show that the twoalternative measures yield very similar results compared with our main measure lnSubDens.

Figure A.4 shows the correlation between subscriber density and literacy – in 1686 (left panel)and in 1786 (right panel). The two variables are uncorrelated.

3This formulation can also be motivated by the distribution of scientific knowledge in equation (??) in the model:Suppose that individuals decide to subscribe to the encyclopedia if they belong to the far upper tail of the knowledgedistribution, say beyond some point S∗. Then, more populous regions n with zero subscriptions have (most likely) athinner upper tail than less populous ones with Subsn = 0.

Appendix p.5

All cities

0.5

11.

52

2.5

Den

sity

0 1 2 3lnSubDens

Only cities with positive subscriptions

0.2

.4.6

Den

sity

0 1 2 3lnSubDens

Figure A.2: Baseline measure of subscriber density: lnSubDensn

Notes: The figure shows the distribution of our main explanatory variable, subscriber density, defined aslnSubDensn = ln(Subsn/pop

1750n + 1). The right panel plots the histogram for all cities and the left

panel plots the histogram only for cities with positive subscriptions.

SubDensn

0.1

.2.3

.4.5

Den

sity

0 5 10 15SubDens

lnSubDens2n

0.2

.4.6

Den

sity

−4 −2 0 2 4lnSubDens2

Figure A.3: Subscriber density: with and without variation across towns with Subs = 0

Notes: The left panel shows subscriber density defined as SubDensn = Subsn/pop1750n . In this setup,

all cities with zero subscriptions have the value SubDensn = 0. The right panel plots the histogram forlnSubDensn = ln

[(Subsn + 1)/pop1750n

], which allows for variation between cities with zero subscrip-

tions, depending on population size.

Appendix p.6

Literacy Rate in 1686 and SubDens

05

1015

Enc

yclo

pedi

a S

ubsc

riptio

ns (

per

1,00

0)

0 .2 .4 .6Literacy rate in 1686

Literacy Rate in 1786 and SubDens

05

1015

Enc

yclo

pedi

a S

ubsc

riptio

ns (

per

1,00

0)

0 .2 .4 .6 .8 1Literacy rate in 1786

Figure A.4: Correlation between Literacy and SubDens

Notes: The figure shows the correlation between subscriber density and literacy in 1686 (left panel)and between subscriber density and literacy in 1786 (right panel). Subscriber density is defined asSubDensn = Subsn/pop

1750n .

Appendix p.7

C.3 Soldier Height

Data on soldier height before 1750 are from Komlos (2005), that recorded height data and city ofbirth for more than 38,000 soldiers from 1716 to 1784 – these include conscripts born between1650 and 1770. In many cases, names of birthplaces are not reported, unrecognizable, or ambigu-ous. Among those that can be clearly identified, we managed to match soldiers to all towns andcities listed in Bairoch, Batou, and Chèvre (1988). In addition, for those cities not listed in Bairochet al. (1988), we match those with more than 10 soldiers to the corresponding department. Overall,this yields observations on about 22,000 soldiers in all 90 departments, and approximately 19,000of these were drafted prior to 1750. From these data, we compute the average height at the depart-ment level before 1750 (i.e., we drop all drafts that occurred in 1750-84). Since average heightmay have changed over the period 1716-49, we filter out birth cohort fixed effects by decade. Wealso address the possibility that the age at recruitment may have differed across regions, whichin turn may affect conscript height because body growth continues into the early 20s. We filterout age-specific patterns by controlling for age and age2. Table A.8 reports the results of theseregressions.

C.4 Control Variables: Sources and Descriptions

Following Dittmar (2011), we obtain a dummy for cities that had a university before 1750 basedon data from Jedin, Latourette, and Martin (1970) and Darby and Fullard (1970). Our dummy for aprinting press between 1450 and 1500 is constructed based on (Febvre and Martin, 1958) and Clair(1976). The number of editions printed before 1501 is from the Incunabula Short-Title Catalogue(ISTC, 2008). We construct a dummy for cities located in non-French speaking departments usinglinguistic data from http://www.lexilogos.com/france_carte_dialectes.htm. There are three maingroups of romance languages in France: langue d’oc, langue d’oil (the official French), and languefrancoprovencal. We consider all three "French". By this definition, the following dialects are"non-French": Alsacien, Basque, Breton, Catalan, and Corsican. Finally, we follow Abramson andBoix (2013) to construct data on proto-industrialization in France, based on the original sourcesCarus-Wilson (1966) and Sprandel (1968).

Local Density of the Nobility

Data on the number of noble families is provided by the Almanach de Saxe Gotha.4 More specifi-cally, we use data on marquises. Entries also contain information on the departments of origin ofthese families, as well as the dates of creation and (if applicable) extinction of the dynasty. Foreach department, we compute the total number of marquis families existing in 1750. This number

4Available at http://en.wikipedia.org/wiki/List_of_French_marquisates#cite_note-1.

Appendix p.8

varies substantially, from zero in Corsica and less than 4 in the departments of Haut-Rhin, Vosges,Meuse, and Moselle (all located in the North East) to 32 and 39 noble families in the departmentsof Seine et Marne, and Seine et Oise.

Executions During the ‘Reign of Terror’

Historical evidence suggests that encyclopedia readers often belonged to the progressive élite priorto the French Revolution. Thus, subscriber density may be lower in predominantly reactionaryregions. These, in turn, may have suffered disproportionately during the ‘Reign of Terror’ (1792-94), when alleged counter-revolutionaries were executed on a massive scale. This gives rise to theconcern that the ‘Reign of Terror’ may have hampered growth in reactionary areas. To control forthis possibility, we geocode data on approximately 13,000 individuals that were executed duringthe ‘Reign of Terror’. These data are available at http://les.guillotines.free.fr. For each executed,we coded information on the department of residence. On average, there are about 150 executionsper department, with the smallest numbers in the Alps area and in Corse, and the highest in thedepartments of Maine et Loire, Loire Atlantique and Vendée (all in the Pays de la Loire region -together they count about 30% of all executions), and in the Rhone (where Lyon is located) andSeine (where Paris is located) departments. We calculate the local ‘density’ of executions followingthe convention for subscriber density: lnExecuteDens = ln(1 + executionsn/pop

1750n ).

C.5 From city level to arrondissement and department level data

In our arrondissement (Tables 10, 11, and A.13) and department level regressions (Tables 8, ??, ??,A.9, and A.10), we aggregate city subscriptions to the arrondissement and department level. Morespecifically, we compute SubDensarr and SubDensdep as the average subscriptions per capitaacross all cities in a given arrondissement (department), where population corresponds to 1750city inhabitants from Bairoch et al. (1988). We then derive lnSubDensarr = ln(1+ SubDensarr)

and lnSubDensdep = ln(1 + SubDensdep).5 Thus, arrondissements (and departments) withoutsubscriptions receive a value of zero.

In line with our city level regressions, we use the baseline controls listed in Table 1 (left panel),and we aggregate them at the arrondissement (department) level. This procedure yields dummiesfor arrondissements (departments) having at least one Atlantic port, a Mediterranean port, a nav-igable river and a University, and a dummy for non-French speaking departments. Moreover, wealso control for the potential confounding factors described in Section 4.3: the density of STNbooks traded in France is calculated in the same way as lnSubDensdep, while the density of proto-industrial activities, the density of nobles in 1750, and the density of executed under the ‘Reign

5Throughout the paper and appendix, we use lnSubDens to refer to subscriber density at all levels of aggregation.

Appendix p.9

of Terror’ are already defined at the department level (see Section 4 for detail). Finally, in thedepartment level regressions, we control for population density in 1876. This is computed usingpopulation data in 1876 and data on the total department surface (in hectares) from Service de laStatistique Général de France (1878).

C.6 Balancedness of the Sample

In the main text, we analyzed the correlation of subscriber density with other city characteristics.Here, we focus on the extensive margin, comparing cities with and without subscriptions. In TableA.1, columns 1-3 report the mean and standard deviations of subscriber density and our baselinecontrols, and column 4 shows the t-test for the difference in means between cities with and withoutsubscriptions. On average, cities with subscriptions are significantly larger, with 26,000 vs. 6,300inhabitants. This difference is not surprising, however, given that larger cities have many morepotential subscribers. While there is no significant difference with respect to location at Atlantic orMediterranean ports (but only on location on navigable rivers), cities with subscriptions are morelikely to have a university prior to 1750, as well as printing press and a higher number of booksprinted in 1500. These differences are statistically significant. Finally, cities with subscriptionsalso tend to have higher levels of literacy in 1786 – despite the difference amounts to 7 percentagepoints, it is not statistically significant. Because of its strong correlation with city population,the dummy for subscriptions that we used above captures many other city characteristics that arealso related to size. Next, we restrict attention to cities with above-zero subscriptions and splitthe sample into those with above- and below-median subscriptions per capita. Columns 5-8 ofTable A.1 show the results.6 For cities with above- and below-median subscriptions per capita,population is almost identical. The same is true for all other controls, with a few exceptions thatreflect local advanced knowledge: 34% and 39% of cities with above-median subscriptions arehosting a university and a printing press respectively, compared to 14% and 26% of cities withbelow-median subscriptions – the difference, however, is not significant for the "Printing Press"variable. Interestingly, cities with below-median subscriptions tend to have higher levels of literacyin 1786. The difference amounts to 7 percentage points, but it is not statistically significant.

6We also exclude Paris, which with its 570,000 inhabitants was five times larger in 1750 than the second-biggestcity, Lyon. When including Paris cities with below-median subscriptions actually have a larger population on average,reversing the pattern from columns 1-3.

Appendix p.10

Table A.1: Subscriptions and control variables – descriptive statistics

All cities Cities with subs>0 (Paris excl.)(1) (2) (3) (4) (5) (6) (7) (8)

All Subs>0 No Subs t-test All Above Below t-testmedian median

Subscriptions per 1,000 1.73 3.94 3.97 6.81 1.27 10.26(3.14) (3.71) (3.72) (3.40) (0.99)

Population in 1750 15 26.05 6.31 3.28 19.57 19.07 20.05 -0.24(42.57) (62.52) (3.05) (18.68) (21.45) (15.85)

Atlantic Port 0.06 0.08 0.05 1.03 0.08 0.07 0.09 -0.33(0.24) (0.27) (0.21) (0.28) (0.26) (0.29)

Mediterranean Port 0.03 0.04 0.03 0.3 0.04 0.02 0.05 -0.54(0.17) (0.18) (0.16) (0.19) (0.16) (0.21)

Navigable River 0.09 0.16 0.03 3.41 0.15 0.12 0.19 -0.81(0.28) (0.37) (0.16) (0.36) (0.33) (0.39)

University 0.11 0.25 0.01 5.53 0.24 0.34 0.14 2.21(0.32) (0.43) (0.10) (0.43) (0.48) (0.35)

Printing Press 0.18 0.33 0.06 5.01 0.32 0.39 0.26 1.32(0.39) (0.47) (0.25) (0.47) (0.49) (0.44)

ln(Books Printed 1500) 0.48 0.87 0.17 3.82 0.78 0.93 0.64 0.87(1.31) (1.74) (0.7) (1.55) (1.62) (1.49)

Literacy 1786 0.43 0.46 0.39 1.73 0.45 0.42 0.49 -1.32(0.25) (0.25) (0.25) (0.25) (0.24) (0.25)

Observations (Min/Max) 166/193 82/85 84/108 81/84 40/41 41/43

Notes: This table compares the mean and the standard deviation of our variable of interest and of our baseline con-trols (as listed in Table 1, left panel). In columns 1-4 we use the full sample, looking at all cities (col 1) and thendistinguishing between cities with (col 2) and without subscriptions (col 3). In columns 5-8 we restrict the sampleto cities with a positive subscriptions (col5), and we distinguish between cities with above-(col 6) and below-mediansubscriptions (col 7). Columns 4 and 8 present the t-test for the difference in means between cities with and withoutsubscriptions and cities with above- and below- median subscriptions. For details on encyclopedia subscriptions andcontrols see the notes to Table 1, Section 4 in the paper and Appendix C.4.

Appendix p.11

C.7 Additional Results for City Growth

We now perform a series of robustness checks and show that our baseline results on city growth(Table 3) are not driven by sample composition or by a particular specification of our variable ofinterest.

We begin with Figure A.5, which shows that the strong correlation between city growth andsubscriber density in the full sample is not driven by outliers. Since it also uses average annualgrowth rates (rather than log city growth), the figure is directly comparable to Figure 2 in the paper.However, it also uses cities for which growth cannot be computed over the period 1400-1750 dueto missing data.

−1.

2−

.8−

.40

.4.8

1.2

City

gro

wth

(re

sidu

al)

−1 0 1 2 3Subscriber density (residual)

coef = .186, (robust) se = .034, t = 5.41

Figure A.5: Encyclopedia subscriptions and city growth 1750-1850 – full sample

Notes: The figure plots average annual city growth in France against subscriber density (lnSubDens), after controllingfor our baseline controls (those used in the left panel of Table 1 in the paper). We exclude three outliers: all smallcities (of 2,000 inhabitants) whose size drops to less than 1/3 of the original size. Including them gives a coefficientof 0.205 (t=5.70).

In Table A.2, we run our growth regression for the 1750-1850 period, using four differentsub-samples, each including those cities for which population data are available in the year 1400,1500, 1600, and 1700 respectively. The coefficients of lnSubDens are very similar in magnitudeto those of Table 3 and are always significant at the 1% level.

In Tables A.3 and A.4, we use the alternative specifications of subscriber density described inAppendix C.2: SubDensn and lnSubDens2n. The first measure still assigns a value of zero toall cities without subscriptions, but is not log-based. In contrast, the second measure introduces

Appendix p.12

Table A.2: Restricting the sample to cities with available data for the pre-1750 period

Dependent variable: log city growth, 1750-1850

(1) (2) (3) (4)

Data on city size in 1700 1600 1500 1400

lnSubDens 0.154∗∗∗ 0.139∗∗∗ 0.157∗∗∗ 0.193∗∗∗

(0.036) (0.042) (0.037) (0.052)

Baseline Controls X X X XR2 0.39 0.57 0.55 0.57Observations 148 58 62 50Notes: All regressions are run at the city level and use cities where data oncity size from Bairoch et al. (1988) are available over for the year indicatedin the header. ‘Baseline Controls’ include all variables listed in Table 1in the paper (left panel), as well as log population in 1750. For details onlnSubDens and controls see the notes to Table 1, Section 4 and AppendixC.4. Robust standard errors in parentheses. * p<0.1, ** p<0.05, ***p<0.01.

variation across cities with zero subscriptions and is thus smaller for larger cities with Subs = 0. Inboth cases, our results continue to hold: subscriber density is positively and significantly associatedwith city growth over the 1750-1850 period, but not before.

In Table A.5, we check whether potential confounding factors are driving our results. We per-form the same analysis as in Table 4, but now restrict the sample to cities with positive subscrip-tions. In all cases, lnSubDens remains a strong predictor of city growth over the industrializationperiod, while the coefficient of literacy is negative and significant in almost all cases, confirmingthe findings in Table 4 in the paper.

Scientific Societies

In Table A.6 we examine the relationship between early scientific societies (founded prior to 1750)and encyclopedia subscriptions. Column 1 shows that 91% of cities with an early scientific societyalso had subscribers to the encyclopedia, as compared to 26% of all other cities reported in Bairochet al. (1988). Column 2 confirms this pattern: cities with scientific societies had on average 4.54subscribers per 1,000 inhabitants, while cities without scientific societies had 1.23.

Appendix p.13

Table A.3: Using alternative subscriber density SubDens

Dependent variable: log city growth over the indicated periodPeriod 1750-1850 Pre-1750

(1) (2) (3) (4) (5) (6) (7) (8)

[unweighted] 1700-1750 1600-1700 1500-1600 1400-1500

PANEL A: All citiesSubDens 0.021∗∗ 0.040∗∗∗ 0.040∗∗∗ 0.048∗∗∗ 0.002 0.012 0.002 0.039

(0.009) (0.009) (0.008) (0.008) (0.009) (0.023) (0.018) (0.035)Controls X X X X X X X XR2 0.10 0.35 0.35 0.26 0.17 0.55 0.52 0.33Observations 193 193 193 193 148 56 45 39

PANEL B: Only cities with positive subscriptionsSubDens 0.016 0.032∗∗∗ 0.028∗∗∗ 0.022∗∗ -0.008 -0.001 -0.003 0.054

(0.012) (0.010) (0.009) (0.009) (0.011) (0.025) (0.019) (0.044)Controls X X X X X X X XR2 0.06 0.47 0.48 0.39 0.30 0.64 0.64 0.32Observations 85 85 85 85 76 44 36 32

Notes: All regressions are run at the city level and are weighted (except for Column 4) by initial population of therespective period. The dependent variable in all columns is log city population growth over the period indicated in theheader. The main period of French industrialization is 1750-1850; city growth in earlier periods is used as a placebo.‘Controls’ are the same as in the corresponding columns of Table 3 in the paper. For details on controls see the notes toTable 1, Section 4 in the paper and Appendix C.4. The main explanatory variable SubDens2 is explained in AppendixC.2. Robust standard errors in parentheses. * p<0.1, ** p<0.05, *** p<0.01.

Appendix p.14

Table A.4: Using alternative subscriber density lnSubDens2

Dependent variable: log city growth over the indicated periodPeriod 1750-1850 Pre-1750

(1) (2) (3) (4) (5) (6) (7) (8)

[unweighted] 1700-1750 1600-1700 1500-1600 1400-1500

PANEL A: All citieslnSubDens2 0.061∗∗∗ 0.088∗∗∗ 0.086∗∗∗ 0.107∗∗∗ 0.003 0.052 0.033 0.040

(0.022) (0.018) (0.017) (0.019) (0.021) (0.063) (0.047) (0.085)Controls X X X X X X X XR2 0.13 0.36 0.37 0.28 0.17 0.56 0.53 0.31Observations 193 193 193 193 148 56 45 39

PANEL B: Only cities with positive subscriptionslnSubDens2 0.067 0.081∗∗ 0.071∗∗ 0.049∗ -0.054 -0.049 0.009 0.096

(0.042) (0.032) (0.028) (0.025) (0.034) (0.085) (0.063) (0.136)Controls X X X X X X X XR2 0.11 0.47 0.48 0.37 0.34 0.64 0.64 0.30Observations 85 85 85 85 76 44 36 32

Notes: All regressions are run at the city level and are weighted (except for Column 4) by initial population of therespective period. The dependent variable in all columns is log city population growth over the period indicated in theheader. The main period of French industrialization is 1750-1850; city growth in earlier periods is used as a placebo.‘Controls’ are the same as in the corresponding columns of Table 3 in the paper. For details on controls see the notes toTable 1, Section 4 in the paper and Appendix C.4. The main explanatory variable SubDens2 is explained in AppendixC.2. Robust standard errors in parentheses. * p<0.1, ** p<0.05, *** p<0.01.

Appendix p.15

Table A.5: Controlling for potentially confounding factors (only cities with Subsn > 0)

Dependent variable: log city growth, 1750-1850

(1) (2) (3) (4) (5) (6) (7)

lnSubDens 0.125∗∗ 0.150∗∗∗ 0.130∗∗ 0.126∗∗ 0.127∗∗ 0.126∗∗ 0.144∗∗∗

(0.051) (0.054) (0.050) (0.049) (0.051) (0.050) (0.050)Literacy 1786 -0.315∗ -0.245 -0.296∗ -0.371∗∗ -0.321∗ -0.293 -0.270

(0.167) (0.160) (0.166) (0.164) (0.161) (0.182) (0.176)lnSTNBooksDens -0.026 -0.015

(0.023) (0.022)Pays d’Eléction -0.107 -0.065

(0.083) (0.094)lnProtoIndDens 1.003 0.629

(0.617) (0.668)lnNoblesDens -0.202 -0.158

(0.212) (0.196)lnExecuteDens 0.022 0.033

(0.035) (0.036)ln(Pop 1750) -0.125∗∗∗ -0.094∗∗ -0.142∗∗∗ -0.114∗∗∗ -0.165∗∗ -0.123∗∗ -0.139∗∗

(0.044) (0.045) (0.045) (0.039) (0.068) (0.047) (0.059)Baseline Controls X X X X X X XR2 0.51 0.52 0.53 0.53 0.52 0.52 0.55Observations 82 82 81 82 82 82 81

Notes: All regressions are run at the city level and are weighted by city population in 1750. The dependentvariable is log city population growth in 1750-1850. ‘Baseline Controls’ include all variables listed inTable 1 in the paper (left panel). For details on lnSubDens and the other explanatory variables see notesto Table 1 , Section 4 in the paper and Appendix C.4. Robust standard errors in parentheses (clustered atthe department level). * p<0.1, ** p<0.05, *** p<0.01.

Appendix p.16

Table A.6: Subscriptions and Scientific Societies

(1) (2)

Scientific Society Subs. Dummy Subs pc

Yes 0.91 4.54

No 0.26 1.23

Notes: This table analyzes the relationship between earlyscientific societies and encyclopedia subscriptions. Al-together, there were 22 French cities with scientific soci-eties founded before 1750. Column 1 shows the percent-age of cities with subscribers to the encyclopedia, andColumn 2 shows subscriptions per 1,000 inhabitants, forcities with and without scientific societies. For detailson scientific societies and encyclopedia subscriptions seeSection 4.1 in the paper.

Appendix p.17

City Growth and Literacy

Finally, we shed light on the relationship between literacy and city growth prior to 1750. Forthis analysis, we use the earlier literacy rates in 1686 (which are highly correlated with those in1786, with a correlation coefficient of 0.84). Coefficients are clustered at the department level –the geographical unit of observation for literacy. Table A.7 shows that literacy is positively (andmarginally significantly) associated with growth in 1700-50, and weakly negatively for 1600-1700.Subscriber density, on the other hand, is not associated with growth prior to 1750, confirming ourprevious results. One possible interpretation for the weak positive effect of literacy in the first halfof the 18th century is that it may have supported pre-industrial development due to its positiveeffect on traditional manufacturing.

Table A.7: City growth and literacy in the pre-1750 period

Dependent variable: log city growth over the indicated period

(1) (2) (3) (4) (5) (6)

1700-1750 1600-1700

lnSubDens 0.021 0.004 0.024 0.080(0.028) (0.039) (0.117) (0.118)

Literacy 1686 0.334∗ 0.358∗ 0.327∗ -0.408 -0.362 -0.430(0.197) (0.192) (0.190) (0.397) (0.376) (0.413)

lnPopinitial -0.018 -0.018 -0.043 0.118∗∗ 0.121∗∗ -0.306(0.014) (0.014) (0.052) (0.049) (0.053) (0.183)

Baseline Controls X XR2 0.04 0.05 0.22 0.09 0.09 0.57Observations 126 126 126 56 50 50

Notes: All regressions are run at the city level and are weighted by initial population of therespective period. The dependent variable in the different columns is log city population growthover the period indicated in the header. ‘Baseline Controls’ include all variables listed in Table1 in the paper (left panel). For details on lnSubDens and the other explanatory variables seenotes to Table 1, Section 4 in the paper and Appendix C.4. Standard errors (clustered at thedepartment level) in parentheses. * p<0.1, ** p<0.05, *** p<0.01.

Appendix p.18

Table A.8: Controlling for the effect of age and birth decade on soldier height, 1660-1740

Dependent variable: Soldiers height in cm

Age 0.440∗∗∗

(0.021)

Age2 -0.006∗∗∗

(0.000)

Birth Decade Dummy Yes

R2 0.07Observations 29292

Notes: This regression is run by OLS. Birth Decade Dum-mies include dummies for all decades from 1660 to 1740.Robust standard errors in parentheses. * p<0.1, ** p<0.05,*** p<0.01.

C.8 Additional Results for Income and Industrialization

We now show further robustness checks on our income and industrialization regressions. Table A.9reports results for average conscript height in the pre-1750 period. We perform a similar analysisto Table 8 (col. 5-6), but we now separately regress conscript height on lnSubDens (col. 1, 3) andLiteracy (col. 2, 4). Then, rather than excluding departments with data for less than 20 soldiers, weweight regressions by the number of soldiers for which height is available in each department (col.5-6). All our results still hold: soldier height prior to 1750 is positively associated with literacy, butnot with lnSubDens. In Table A.10, we provide additional evidence for the relationship betweenworker skills and manufacturing: the dependent variable is the number (col. 1-3) and the density(col 4) of proto-industrial activity (i.e., traditional manufacturing). The regression coefficientsconfirm that in the pre-industrial period, manufacturing is strongly associated with literacy, but notwith knowledge elites.

C.9 British Inventions and Encyclopedia Plates

Data on Encyclopedia plates are from the Encyclopedia of Diderot and d’Alembert: collaborative

translation project7 and from the ARTFL Encyclopédie project8. For each volume, they list allarticles and plates of the Encyclopédie of Diderot and d’Alembert. Altogether, there are 2575plates, accompanying 326 entries. They contain reference numbers and characters to match thevarious parts of the figure with the text and the legend. For all 326 entries, we computed the exactnumber of accompanying plates, and – where possible – we matched each entry with one of the

7Available at http://quod.lib.umich.edu/d/did/index.html8Available at http://encyclopedie.uchicago.edu/

Appendix p.19

Table A.9: Soldiers height before 1750

Dep. var.: Soldiers height in cm (controlling for age and birth decade)

(1) (2) (3) (4) (5) (6)

Weighted

Literacy 1686 1.166∗∗ 1.126∗∗ 1.007∗∗ 0.989∗∗

(0.539) (0.557) (0.433) (0.441)lnSubDens 0.065 0.073 0.049 0.062

(0.107) (0.108) (0.112) (0.111)Baseline Controls X X XR2 0.05 0.00 0.14 0.09 0.08 0.13Observations 76 84 76 84 75 75

Notes: All regressions are run at the French department level and include a dummy for Paris(Department Seine). The dependent variable is average soldier height recorded over the period1716-49 and collected by Komlos (2005). To account for variation in height and soldier agewithin this period, we control for age, age squared, and birth decade (see Appendix A.8). InColumns 1-4, we include only departments with data for at least 20 soldiers, while in Colums5-6 we use all observations and we weight regressions by the number of soldiers per department.‘Baseline Controls’ include all variables listed in Table 1 in the paper (left panel). For detailson Literacy, lnSubDens and controls see the notes to Table 1, Section 4 in the paper andAppendix C.4. Original city-level variables are aggregated to the department level as describedin Appendix C.5. Robust standard errors in parentheses. * p<0.1, ** p<0.05, *** p<0.01.

Appendix p.20

Table A.10: Proto-Industrialization and the two measures for human capital

(1) (2) (3)

Dependent variable: Proto-Industrialization ln(Proto-Ind.

Index Index)

lnSubDens 0.069 0.114 -0.006(0.201) (0.209) (0.012)

Literacy 1686 2.914∗ 2.834∗ 0.116(1.463) (1.582) (0.098)

Baseline Controls X XR2 0.06 0.12 0.07Observations 75 75 73

Notes: All regressions are run at the French department level and include adummy for Paris (Department Seine). The dependent variable in cols 1-3 is the in-dex of proto-industrial activities (computed as the total number of mines, forges,iron trading locations, and textile manufactures) per department. The dependentvariable in col 4 is the local density (in logarithm) of proto-industrial activities.‘Baseline Controls’ include all variables listed in Table 1 in the paper (left panel).For details on Literacy, lnSubDens and controls see the notes to Table 1, Sec-tion 4 in the paper and Appendix C.4. Original city-level variables are aggregatedto the department level as described in Appendix C.5. Robust standard errors inparentheses.* p<0.1, ** p<0.05, *** p<0.01.

Appendix p.21

Table A.11: Modern vs. old sectors: Share of inventive output and encyclopedia plates

(1) (2)

Share Invent. Share PlatesOutput Encyclopedia

Modern 0.084 0.67

Old 0.016 0.33

Notes: This table distinguishes between ‘modern’and ‘old’ manufacturing sectors, based on the me-dian share of total "inventive output" from Nuvolariand Tartari (2011) (see Section 5.3 in the paper fordetail). Column 1 shows the average share of to-tal "inventive output" for both manufacturing tech-nologies. Column 2 compares the share of platesdescribing ‘modern’ and ‘old’ manufacturing tech-nologies in all encyclopedia plates dedicated tomanufacturing (see Appendix C.9 for sources anddetail).

21 British Industrial sector from Nuvolari and Tartari (2011). In total, 156 entries accompaniedby 1272 plates describe manufacturing technologies. Among them, 103 entries (and 849 plates)are dedicated to the ‘modern’ sector. The number of plates per entry varies substantially. Forinstance, among all plates describing manufacturing, 30 entries have only one plate, while theentries "Surgery", "Clock Making", and "Turner and Turning Lathe" have 39, 51 and 87 platesrespectively.

C.10 Firm Data and Results

Firm level data are from Chanut, Heffer, Mairesse, and Postel-Vinay (2000), who cleaned and dig-italized a survey of more than 14,000 firms, originally conducted by the Statistique de la France

over the period 1839-1847. The data are collected at the arrondissement (county) level, and catego-rize firms into 13 main manufacturing sectors.9 Merging the 13 French and the 21 British sectors,we obtain 8 consistent sectors.10 We then compute their innovation index as the weighted average"share of inventive output" from the British patent data.11 Using this index, we classify French

9Typically, arrondissements map one-to-one into cities in our sample; only 2 arrondissements include more thanone city with observed population in 1750 and above-zero subscriptions. The departments of Corsica, Savoie, HauteSavoie and Territoire de Belfort are not included in the survey.

10Some categories overlap, so that a consistent match results in 8 sectors. Whenever there is more than one Britishsector corresponding to a French sector, we compute a weighted average – where the weights are the number of patentsin the British sector.

11We use the average, rather than the sum of "inventive output" shares because otherwise aggregating many sectorswould mechanically raise their innovation index.

Appendix p.22

sectors into ‘modern’ and ‘old’, based on above- vs. below-median innovation index. Table A.12lists the resulting 8 sectors together with their innovation index. The dataset has about 630 (800)sector-arrondissement observations for modern (old) sectors.

In our analysis, we use male wages as dependent variable. These represent the average dailywages for men (recorded in centimes). As controls, we use establishment size, computed using dataon the number of workers and the number of establishments for each firm. We also include ‘Sizeand Population Controls’: we control for the average firm size in a given sector and arrondissement,using data from Chanut et al. (2000) (see Section 5.3 and Section C.10 for detail), and for the log ofthe total population, as well as for the urbanization rate in 1831. The urbanization rate is computedusing data on total and urban population in 1831. These data are from Service de la StatistiqueGénéral de France (1878) and are recorded at the department level. In Table 11, we use informationon the number of steam engines and other engines (which include water, wind and animal engines),as proxies for an industry’s dependence on energy-related up-front investment. These figures arereported in Chanut et al. (2000).

Finally, in Table A.13 we perform the same analysis as in Table 10, but we now include regionfixed effects. Our results continue to hold: the coefficient on lnSubDens in ‘modern’ sectorsis highly significant and much larger than in ‘old’ sectors. The coefficient on schooling insteadbecomes smaller in size and is insignificant for both ‘modern’ and ‘old’ sectors. This is becauseschooling is defined at the department level, so that region fixed effect absorb much of the variation.

Appendix p.23

Table A.12: Classifaction of individual industrial sectors into ‘modern’ and ‘old’

(1) (2)

Sector Name Innovation SectorIndex Type

Textile and Clothing 0.145 modern

Printing Technology, and 0.094 modernScientific Instruments

Furniture and Lighting 0.045 modern

Transportation Equipment 0.040 modern

Metal and Metal Products 0.039 old

Leather 0.018 old

Mining 0.017 old

Ceramics and Glass 0.012 old

Notes: For each sector, column 1 reports the innovation in-dex, obtained using data from Nuvolari and Tartari (2011);column 2 classifies sectors into ‘modern’ or ‘old’ manufac-turing, based on the median of the share of the innovationindex. For details on the innovation index and on the Frenchindustrial survey see Section 5.3 in the paper and AppendixC.10.

Appendix p.24

Table A.13: Subscriber density and firm wages in 1837, incl. region fixed effects

Dep. var.: log wages (arrondissement average, for firms in ‘modern’ and ‘old’ sectors)

(1) (2) (3) (4) (5) (6) (7) (8)

modern old modern old modern old modern old

lnSubDens 0.102∗∗∗ 0.028∗ 0.091∗∗∗ 0.025 0.099∗∗∗ 0.029 0.095∗∗∗ 0.026(0.014) (0.014) (0.018) (0.015) (0.020) (0.019) (0.021) (0.019)

School Rate 1837 0.119 -0.012 0.067 0.040 0.081 0.066 0.090 0.053(0.120) (0.073) (0.111) (0.076) (0.119) (0.091) (0.104) (0.104)

Establishment Size -0.025∗∗ 0.058∗∗∗ -0.024∗∗ 0.057∗∗∗ -0.025∗∗ 0.054∗∗∗ 0.005 0.043∗∗∗

(0.009) (0.008) (0.012) (0.010) (0.012) (0.010) (0.010) (0.010)Pop. Controls X X X X X X X XBaseline Controls X X X X X XAdditional Controls X X X XSector FE X XRegion FE X X X X X X X XR2 0.30 0.32 0.34 0.36 0.36 0.36 0.48 0.40Observations 633 794 428 541 386 493 386 493

Notes: Region fixed effects are for 21 official regions in France. All regressions are run at the French arrondissementlevel and include a dummy for Paris (Department Seine). The dependent variable is the log of average male wagesacross all firms in sector j in arrondissement n (see Section C.10 for detail). ‘Baseline Controls’ include all variableslisted in Table 1 in the paper (left panel). ‘Additional Controls’ are those used in Table 4 in the paper. For detailson lnSubDens and controls see the notes to Table 1, Section 4 in the paper and Appendix C.4. Original city-levelvariables are aggregated to the arrondissement level as described in Appendix C.5. Standard errors (clustered at thedepartment level) in parentheses. * p<0.1, ** p<0.05, *** p<0.01.

Appendix p.25

++For density variables shall we say per capita (log subscriptions per capita)? density of (sub-scribers density)? Shall we mention that it’s always over pop in 1750?++

Appendix p.26

Vari

able

Nam

eVa

riab

leD

escr

iptio

nSo

urce

Exp

lana

tory

Var

iabl

es

I Subs

dum

my

equa

lto

1fo

rciti

esw

itha

posi

tive

num

bero

fsub

scri

ptio

nsD

arnt

on(1

973)

lnSu

bDen

slo

gof

subs

crip

tions

perc

apita

Dar

nton

(197

3)

I Scien

t.Society

dum

my

equa

lto

1fo

rciti

esho

stin

ga

scie

ntifi

cso

ciet

ybe

fore

1750

McC

lella

n(1

985)

lnM

embD

ens

log

ofpe

rcap

itam

embe

rsof

scie

ntifi

cso

ciet

ies

befo

re17

50M

cCle

llan

(198

5)

lnH

ugue

nots

Den

slo

gof

hugu

enot

spe

rcap

ita++

in16

70++

(dep

t.le

vel)

Mou

rs(1

958)

lnSc

ient

ists

Den

slo

gof

fam

ous

scie

ntis

tspe

rcap

ita++

1000

-188

7++

dela

Cro

ixan

dL

ican

dro

(201

2)

lnIn

vent

ions

Den

slo

gof

inve

ntio

ns(e

xhib

ited

atth

eC

ryst

alPa

lace

Exh

ibiti

onM

oser

(200

5)an

dat

the

Cen

tenn

ialE

xhib

ition

in18

76)p

erca

pita

Lite

racy

perc

enta

geof

men

able

tosi

gnth

eirw

eddi

ngce

rtifi

cate

in16

86or

1786

Fure

tand

Ozo

uf(1

977)

Scho

olR

ate

1837

ratio

ofst

uden

tsto

scho

ol-a

gepo

pula

tion

Mur

phy

(201

0)

Out

com

eV

aria

bles

log

city

grow

thlo

gof

city

popu

latio

ngr

owth

Bai

roch

etal

.(19

88)

Sold

ierH

eigh

tpre

-175

0so

ldie

rhei

ghti

ncm

inth

e17

16-1

749

peri

od(d

eptl

evel

aver

age)

Kom

los

(200

5)

Sold

ierH

eigh

t181

9-18

26so

ldie

rhei

ghti

ncm

(dep

t.le

vel)

Aro

n,D

umon

t,an

dL

eR

oyL

adur

ie(1

972)

ln(d

ispo

sabl

ein

com

e)lo

gdi

spos

able

inco

me

in18

64(d

ept.

leve

l)D

elef

ortr

iean

dM

oric

e(1

959)

Em

ploy

men

tSha

res

(agr

ic)

empl

oym

ents

hare

sin

agri

cultu

rein

1876

(dep

t.le

vel)

Serv

ice

dela

Stat

istiq

ueG

énér

alde

Fran

ce(1

878)

Em

ploy

men

tSha

res

(ind

ustr

y)em

ploy

men

tsha

res

inin

dust

ryin

1876

(dep

t.le

vel)

Serv

ice

dela

Stat

istiq

ueG

énér

alde

Fran

ce(1

878)

Wag

esA

gric

wag

esin

agri

cultu

rein

1852

(dep

t.le

vel)

Gor

eaux

(195

6)

Wag

esIn

dust

ryw

ages

inin

dust

ryin

1852

(dep

t.le

vel)

Gor

eaux

(195

6)

Wag

esTe

xtile

wag

esin

text

ilein

1864

(dep

t.le

vel)

Del

efor

trie

and

Mor

ice

(195

9)

Wag

esFo

odPr

oc.

wag

esin

food

proc

essi

ngin

1864

(dep

t.le

vel)

Del

efor

trie

and

Mor

ice

(195

9)

Wag

esm

oder

n/ol

dw

ages

inm

oder

nan

dol

dse

ctor

sin

1839

-184

7(a

rron

d.le

vel)

Cha

nute

tal.

(200

0)

VariableN

ame

VariableD

escriptionSource

ControlV

ariables

lnPop

initia

llog

citypopulation

Bairoch

etal.(1988)

Atlantic

Portdum

my

equalto1

forcitieslocated

onan

atlanticport

Dittm

ar(2011)

Mediterranean

Portdum

my

equalto1

forcitieslocated

ona

mediterranean

portD

ittmar(2011)

Navigable

River

dumm

yequalto

1forcities

locatedon

anavigable

riverD

ittmar(2011)

University

dumm

yequalto

1forcities

hostinga

universitybefore

1750Jedin

etal.(1970);Darby

andFullard

(1970)

Parisdum

my

equalto1

forParisorforthe

Seinedepartm

ent

Non

FrenchSpeaking

dumm

yequalto

1forcities

locatedin

non-Frenchspeaking

departments

http://ww

w.lexilogos.com

/france_carte_dialectes.htm

Printingpress

in1500

dumm

yequalto

1forcities

who

hada

printingpress

Febvreand

Martin

(1958);Clair(1976)

ln(Books

Printed1500)

logofeditions

printedbefore

1500IST

C(2008)

Establishm

entSizelog

ofnumberofw

orkersperestablishm

entin1839-1847

(arrond.level)C

hanutetal.(2000)

PopulationD

ensitypopulation

in1876

overtotalsurface(dept.level)

Servicede

laStatistique

Généralde

France(1878)

Confounding

Factors

lnSTNB

ooksDens

logofST

Nbook

salespercapita

FBT

EE

(2012)

Paysd’E

léctiondum

my

equalto1

forcitieslocated

ina

payd’éléction

http://fr.wikipedia.org/w

iki/Généralité_(France)

lnProtoIndD

enslog

ofproto-industrialactivitiespercapita

(dept.level)C

arus-Wilson

(1966);Sprandel(1968)

lnNoblesD

enslog

ofnoblesfam

iliespercapita

(dept.level)http://en.w

ikipedia.org/wiki/L

ist_of_French_marquisates#cite_note-1

lnExecuteD

enslog

ofexecutedduring

thereign

ofterrorpercapita(dept.level)

http://les.guillotines.free.fr

Others

ShareInventive

Output

measure

basedon

reference-weighted

patents,adjustedforthe

sector-specificfrequency

Nuvolariand

Tartari(2011)ofpatenting

ratesand

citations

SharePlates

Encyclopedia

shareofE

ncyclopediaplates

relatedto

modern

sectorsovertotalm

anufacturingplates

http://quod.lib.umich.edu/d/did/index.htm

lhttp://encyclopedie.uchicago.edu/

ReferencesAbramson, S. F. and C. Boix (2013). The Roots of the Industrial Revolution: Political Institutions or

(Socially Embedded) Know-How? APSA 2013 Annual Meeting Paper.

Aron, J. P., P. Dumont, and E. Le Roy Ladurie (1972). Anthropologie du Conscrit Français. Paris: MoutonDe Gruyter.

Bairoch, P., J. Batou, and P. Chèvre (1988). La Population des villes Europeenees de 800 à 1850: Banquede Données et Analyse Sommaire des Résultats. Geneva: Centre d’histoire economique Internationale del’Université de Genève, Libraire Droz.

Braudel, F. (1982). Civilization and Capitalism, 15th–18th Century: The perspective of the world. Univer-sity of California Press.

Carus-Wilson, E. M. (1966). The Woollen Industry. Cambridge Economic History of Europe. Volume II.Cambridge University Press.

Chanut, J.-M., J. Heffer, J. Mairesse, and G. Postel-Vinay (2000). L’Industrie franccaise au milieu du 19esiécle: les enquêtes de la statistique générale de la France. Editions de l’école des hautes études ensciences sociales, Paris.

Clair, C. (1976). A History of European Printing. London: Academic.

Darby, H. C. and H. Fullard (1970). The New Cambridge Modern History, Volume 14. Cambridge: Cam-bridge University Press.

Darnton, R. (1973). The Encyclopedie Wars of Prerevolutionary France. The American Historical Re-view 78(5), 1331–1352.

de la Croix, D. and O. Licandro (2012). The Longevity of Famous People from Hammurabi to Einstein.CORE Discussion Paper 2012/52.

de Tocqueville, A. (1856). The Old Regime and the French Revolution. Garden City, N.Y.: Doubleday.Reprinted in 1955.

Delefortrie, N. and J. Morice (1959). Les Revenus departmentaux en 1864 et en 1954. Paris, France: Colin.

Dittmar, J. E. (2011). Information Technology and Economic Change: The Impact of the Printing Press.Quarterly Journal of Economics 126, 1133–1172.

FBTEE (2012). The French Book Trade in Enlightenemnt Europe Database.

Febvre, L. and H.-J. Martin (1958). L’Apparition du Livre. Paris: Albin Michel.

Furet, F. and J. Ozouf (1977). Lire et écrire. L’alphabétisation des Français de Calvin à Judes Ferry. Minuit,Paris.

Goreaux, L. M. (1956). Les migrations agricoles en France depuis un siécle et leur relation avec certainsfacteurs économiques. Études et Conjoncture 11, 327–376.

Graham, H. (2011). Humble Petitioners? Seigneurial Courts, Royal Justice and the Role of Litigants in theEighteenth Century. French History & Civilization 4, 12–21.

Hamscher, A. N. (2012). The Royal Financial Administration and the Prosecution of Crime in France,1670–1789. Newark: University of Delaware Press.

ISTC (2008). Incunabula Short-Title Catalogue. London: British Library.

Jedin, H., K. S. Latourette, and J. Martin (1970). Atlas Zur Kirchengeschichte. Freiburg: Herder.

Appendix p.29

Kessler, A. D. (2010). Marginalization and Myth: The Corporatist Roots of France’s Forgotten ElectiveJudiciary. American Journal of Comparative Law 58, 679–720.

Komlos, J. (2005). Height of French Soldiers, 1716–1784. [Computer file] ICPSR04363-v1.

Maddison, A. (2007). Historical Statistics. University of Groningen. Webpage:http://www.ggdc.net/maddison/.

Marczewski, J. (1961). Some Aspects of the Economic Growth of France 1660-1958. Economic Develop-ment and Cultural Change 9(3), 369–386.

McClellan, J. E. (1985). Science Reorganized: Scientific Societies in the Eighteenth Century. New York:Columbia University Press.

McEvedy, C. and R. Jones (1978). Atlas of World Population History, Facts on File. New York.

Moser, P. (2005). How Do Patent Laws Influence Innovation? Evidence from Nineteenth-Century World’sFairs. American Economic Review 95(4), 1214–1236.

Mours, S. (1958). Les Églises Réformées en France. Tableaux et Cartes. Librairie Protestante.

Mousnier, R. (1979). The Institutions of France under the Absolute Monarchy, 1598–1789. Chicago:University of Chicago Press.

Muessig, U. (2012). Judges and Judging in the History of the Common Law and Civil Law, Chapter Su-perior courts in early-modern France, England and the Holy Roman Empire, pp. 209–233. CambridgeUniversity Press.

Murphy, T. E. (2010). Old Habits Die Hard (Sometimes): What can Département Heterogeneity Tell usabout the French Fertility Decline? IGIER Working Paper 364.

Nuvolari, A. and V. Tartari (2011). Bennet Woodcroft and the value of English patents, 1617–1841. Explo-rations in Economic History 48, 97–115.

Roehl, R. (1976). French Industrialization: A Reconsideration. Explorations in Economic History 13,223–281.

Service de la Statistique Général de France (1878). Annuaire Statistique de la France. Paris, ImprimerieNationale.

Sprandel, R. (1968). Das Eisengewerbe im Mittelalter. A. Hiersemann.

Tilly, C. (1990). Coercion, Capital, and European States, 990–1990. London: Blackwell.

Appendix p.30