INFORMATION TECHNOLOGY ANDECONOMIC CHANGE:THE IMPACT OF THE PRINTING PRESS Jeremiah DittmarJanuary 10, 2011AbstractThe printing press was the great innovation in early modern information technology, but economists have found no macroeconomic evidence of its impact. Thispaper exploits city-level data. Between 1500 and 1600, European cities whereprinting presses were established in the 1400s grew 60 percent faster than otherwise similar cities. Cities that adopted printing in the 1400s had no prior advantageand the association between adoption and subsequent growth was not due to printers choosing auspicious locations. These findings are supported by regressions thatexploit distance from Mainz, Germany – the birth place of printing – as an instrument for adoption.Keywords: Information Technology, Cities, Growth, History, Printing Press.JEL Classification: N13, N33, N93, O11, O18, O33 I thank Barry Eichengreen, Chad Jones, Christina Romer, Brad DeLong, Peggy Anderson, SureshNaidu, Davide Cantoni, Tien-Ann Shih, Robert Barro, Greg Clark, seminar participants at Berkeley,Stanford, Harvard, Yale, American, Northwestern, George Washington, the NBER Summer Institute,George Mason, and anonymous referees for their comments. Financial support from the EconomicHistory Association and the National Science Foundation (SES #1023380) is gratefully acknowledged.The errors are mine.1

1IntroductionThe movable type printing press was the great innovation in early modern informationtechnology. The first printing press was established in Mainz, Germany between 1446 and1450. Over the next fifty years the techology diffused across Europe. Between 1450 and1500, the price of books fell by two-thirds, transforming the ways ideas were disseminatedand the conditions of intellectual work. Historians suggest the printing press was one ofthe most revolutionary inventions in human history.1Economists have found no evidence of the technology’s impact in measures of aggregate productivity or per capita income – much as, until the mid-1990s, they foundno evidence of productivity gains associated with computer-based information technologies. A conventional explanation is that the economic effects of the printing press werelimited: whatever the advances, they occurred in a very small sector marked by modest price elasticities.2 However, that argument makes no attempt to gauge the positiveexternalities historians argue were associated with the diffusion of printing. Historicalresearch suggests that print media transformed the ways ideas were disseminated, promoted the accumulation of human capital, and played a key role in the evolution ofbusiness practices (Febvre and Martin 1958; Eistenstein 1979; Hoock 2008).This paper examines these spillovers by exploiting new, city-level data on the adoptionof the movable type printing press in 15th century Europe. It uses city-level data toexamine two principal questions: Was the new printing technology associated with citygrowth? And, if so, how large was the association? To explore these questions, thispaper compares cities where printers established presses to similar cities where they didnot. The paper uses OLS estimators to document the magnitude and the timing of theassociation between printing and city growth. It then employs historical evidence andinstrumental variable techniques to identify the impact of printing on city growth.The instrumental variable (IV) analysis is motivated by historical evidence. JohannesGutenberg established the first printing press in Mainz, Germany around 1450.3 At thattime only a small number of men in Mainz knew the secrets behind the technology.Between 1450 and 1500, the technology diffused in “concentric circles” (Barbier 2006,p. 192) as printers set out from Mainz to establish presses in other cities. Distancefrom Mainz was strongly and significantly associated with early adoption of the printing1For instance Roberts (1996), Braudel (1979c), and Gilmore (1952). On prices see van Zanden (2004)and Clark (2004).2Clark (2001, p. 60) argues that the macroeconomic impact was “unmeasurably small” for thesereasons.3For details of Gutenberg’s innovation and competing attempts to devise print media see Section 5.4.2

press, but not with city growth before the diffusion of the printing press or with otherindependent determinants of city growth. The geographic pattern of technology diffusionthus allows us to identify exogenous variation in adoption. Instrumenting for adoptionwith distance from Mainz, I find very large and significant estimates of the relationshipbetween the adoption of the printing press and city growth.The printing press fostered knowledge and skills that were valuable in commerce.Print media played a key role in the development of numeracy, the emergence of businesseducation, and the adoption of innovations in book-keeping and accounting. With accessto cheap water-borne transport, port cities were positioned to profit from innovations incommercial practice. In the data, I find that printing delivered special benefits to portcities – beyond the advantages associated with printing or with port location alone.These findings add a new dimension to arguments stressing the role of cities as siteswhere information was exchanged, ideas were produced, and the business practices andsocial groups that drove the rise of European capitalism developed.2LiteratureAmong economic historians, there is some difference of opinion about the extent towhich the movable type printing press was a revolutionary innovation. Mokyr (2005a,pp. 1120-1122) notes that innovation depends on the cost of accessing existing knowledge, and that the printing press was one of the most important access-cost-reducinginventions in history. Jones (1981, pp. 60-62) also argues that “western progress owedmuch to the superior means of storing and disseminating information.” Baten and vanZanden (2008) find a significant association between simulated national-level wages andobserved differences in aggregate book production in European history.4 However, Clark(2001) finds no evidence of aggregate productivity growth associated with the diffusionof movable type printing. Mokyr (2005a) similarly argues that the aggregate effects weresmall.Social historians have hailed the movable type printing press as a revolutionary innovation. Braudel (1979c, p. 435) identifies printing as one of three great technologicalrevolutions observed 1400-1800 (alongside advances in artillery and navigation). Gilmore(1952, p. 186) states that printing drove, “the most radical transformation in the con4Baten and van Zanden (2008) draw simulated country-level real wages from Allen (2003). This papertakes the city as the unit of analysis. Within economies, there was significant variation in printing andgrowth across cities. Observed data on economic outcomes is also available at the city level. Moreover,contemporary national boundaries did not define the historic economies of Europe.3

ditions of intellectual life in the history of western civilization.” Eisenstein (1979, pp.33, 72-75) argues that printing created revolutionary new possibilities for “combinatoryintellectual activity.” Roberts (1996, p. 220) suggests the outcome was one, “dwarfingin scale anything which had occurred since the invention of writing.”Macroeconomic research identifies the central role ideas play in technological changeand growth (Jones and Romer 2010; Romer 1990; Lucas 2009). Economists observe thattechnological change is driven by the sharing and recombining of ideas (Mokyr 1995;Weitzman 1998; Romer 1990). These findings indicate that major changes in the waysideas can be stored and transmitted may have far reaching consequences.Printing was an urban technology. The market for print media was overwhelminglyurban. Motivated by these facts, this paper takes cities as the units of analysis.European cities played a central role in the emergence of modern, idea-based capitalist economic growth. Urban life generated social contacts that fostered the circulationof information and innovation (Bairoch 1988, p. 499). Cities were also seedbeds of capitalist business practices. Braudel (1979a, p. 586) observes that historically, “Capitalismand towns were the same things.”5 Historians and economists have observed that citysizes were historically important indicators of economic prosperity; that broad-based citygrowth was associated with macroeconomic growth; and that cities produced the economic ideas and social groups that transformed the European economy.6 These factssupport the use of city growth as an indicator of economic vitality.3The MechanismThis section describes how the adoption of printing technology impacted city growth inearly modern Europe. The key point is that cities that adopted print media benefittedfrom localized spillovers in human capital accumulation, technological change, and forward and backward linkages. These spillovers contributed to city growth by exerting anupward pressure on the returns to labor, making cities culturally dynamic, and attractingmigrants. They were localized by high transport costs associated with inter-city tradeand because the printing press fostered important face-to-face interactions.75Historical research has qualified this generalization (e.g Scott 2002) but confirms the importance ofcities. For discussion see Dittmar (2010).6Acemoglu, Johnson, and Robinson (2005), DeLong and Shleifer (1993), Bairoch (1988), and Braudel(1979a, 1979c).7This paper stresses the effects of print media on the development of economically useful skills andknowledge. The interplay between printing and religion is discussed below in section 5.6. The historicassociation between printing and city growth is consistent with Glaeser and Saiz’s (2003) finding that4

Historically, urban death rates exceeded urban birth rates and migration drove citygrowth. Cities drew migrants to the extent that they offered relatively high wages, cultural amenities, and economic opportunities.8 In the pre-industrial era, commerce was amore important source of urban wealth and income than tradable industrial production.9As a result, migration and city growth were typically contingent on commercial success.10Print media played a key role in the acquisition and development of skills that werevaluable to merchants.11 The ability to calculate interest rates, profit shares, and exchange rates was associated with high returns for merchants engaged in large scaleand long-distance trade. Starting in the 1480s, European presses produced a streamof “commercial arithmetics.” Commercial arithmetics were the first printed mathematics textbooks and were designed for students preparing for careers in business.12 Theytransmitted commercial know-how and quantitative skills by working students throughproblems concerned with determining payments for goods, currency conversions, interestpayments, and profit shares. The first known printed mathematics text is the TrevisoArithmetic (1478). It begins:I have often been asked by certain youths.who look forward to mercantilepursuits, to put into writing the fundamental principles of arithmetic.Herebeginneth a Practica, very helpful to all who have to do with that commercialart. (Reproduced in Swetz [1987, p. 40])Gaspar Nicolas, author of the first Portuguese arithmetic (1519), similarly explained:I am printing this arithmetic because it is a thing so necessary in Portugal fortransactions with the merchants of India, Persia, Ethiopia, and other places.(Quoted in Swetz [1987, p. 25])Hundreds of commercial arithmetics were printed 1480-1550 (see Figure I below).Print media was also associated with the development of cutting-edge business practice. Social scientists have identified double-entry book-keeping as an important technological innovation since the early 20th century, when Weber (1927) and Sombart (1953)human capital predicts population and productivity growth at the city level in our day.8On migration and historical demography see Woods (2003), de Vries (1984), and Bairoch (1988).9See inter alia Nicholas (2003, p. 7) and Braudel (1966).10Political capitals were exceptions to this rule.11A large share of print media was religious and less likely to generate positive spillovers. However,the availability of affordable religious and humanist works promoted literacy and, increasingly, normsfavoring the exchange of ideas. Literacy is discussed below. Section 5.6 discusses printing and religion.12They were employed in urban schools and by private teachers teaching commercial arithmetic. Theschools teaching commercial arithmetic operated parallel to universities, which did not provide businessoriented preparation. See Rey (2006), Speisser (2003), Swetz (1987), and Goldthwaite (1972).5

argued that it played a key role in the emergence of rational, optimizing business practice. The first published description of double-entry book-keeping appeared in 1494 (LucaPacioli’s Summa). Printed merchants’ manuals then disseminated the key ideas. Generally, merchants’ manuals combined instruction in accounting and arithmetic with nonquantitative guidance on business practice (Hoock 2008; Goldthwaite 1972). A subsetcontained tables that simplified the calculation of interest on loans, tariffs, and transportcosts. Hoock (2008, p. 149) observes that, “In some ways, [these handbooks] present thesame characteristics as the modern pocket calculator with integrated routines.” FigureI documents that hundreds of different merchants’ manuals were printed 1480-1550. Itshows that growth in the number of merchants’ manuals printed declined from high initialrates and by the late 1500s stabilized at a constant rate (approximately 1% per year).FIGURE I – INSERT HEREThe observation that print media fostered the development of business practices employed in long distance trade raises a question: Did printing deliver special benefits togeographic locations that were propitious for commerce? Historically, transport over landwas relatively expensive. Cities with access to cheap, water borne transport were positioned to realize high returns to innovations in commerce. Section 5.2 (below) documentsthat the growth advantage enjoyed by cities that adopted printing in the late 1400s waslargely driven by the growth of ports with printing presses – beyond advantages associated with the printing press or with being a port alone.The availability of inexpensive texts was a key prerequisite for the spread of literacyin Renaissance Europe (Grendler 1990). School books generated high returns for Renaissance printers (Füssel 2005; Nicholas 2003; Bolgar 1962). Schooling in languages becamepart of a progression in which pupils went from “arts to marts.” Cities began to runschools for children who were not going to learn Latin – using printed grammar schooltexts. In the 15th century, it became expected that the children of the bourgeoisie wouldattend school (Bolgar 1962). But print media also promoted opportunities for the lessprivileged to obtain education and raise their incomes. Brady (2009, p. 33) observesthat no document better captures the new opportunities than Thomas Platter’s (14991582) autobiography (Platter 1839). After wandering penniless across Europe, Platterbegan his formal schooling at age 18. Having learned Latin, Platter took a job as a ropemaker in Zurich to support his book-buying and reading habit, taught himself Hebrewand Greek, and rose to become a wealthy school master, professor, and printer.Beyond literacy, print media fostered the development of new, bourgeois competences6

and the “social ascent of new professionals” (Scott 2002, p. 37).13 The urban middleclasses were the principle purchasers of books. Printing spread to meet, “demand forbooks among the merchants, substantial artisans, lawyers, goverment officials, doctors,and teachers who lived and worked in who needed to read, write, and calculate in order to manage their businesses and conduct civic affairs.” (Rice 1994, p. 6)The new technology underpinned an emerging culture of information exchange and thedevelopment of an urban, bourgeois public sphere (Zaret 2000; Long 1991; Smith 1984).The role of print media in the diffusion of industrial innovations was probably morelimited. Historically the diffusion of industrial technology was heavily dependent on themovement of skilled workers (Cipolla 1972). This is consistent with the emphasis thispaper places on localized spillovers from print media and the pattern of technology diffusion described below. Significantly, the knowledge required to successfully cast movabletype remained quasi-proprietary for nearly one century after Gutenberg’s innovation:Biringuccio’s Pirotechnia (1540) provides the earliest known published blueprint.Cities with printing presses derived benefits from the technology that others did not.The costs of information and human capital accumulation were significantly lower incities with printing presses. In part, these advantages were due to transport costs.Print media was costly to transport because it was a heavy and fragile commodity,sensitive to damp (Barbier 2006; Febvre and Martin 1958). The trade in hardboundbooks was relatively extensive but still significantly limited. Outside printing cities,information on the range of available print media was incomplete and many books werenot offered for sale. Flood (1998, p. 55) observes that, “Outside the towns where bookswere printed or which were main centers of the burgeoning book trade the public weredependent on what itinerant traders offered them and on word of mouth.”14 Booklets andephemera termed “city printing” (l’imprimerie de ville) accounted for a large share ofproduction and were even less widely traded.15 Transport costs in early modern Europewere sufficiently high that print media often spread through reprinting rather than intercity trade.16 Books were often shipped unbound and in very small lots – a few copies of13Mokyr (2005a) defines competence as extending beyond the ability to read, interpret, and executethe instructions of a technique to include supplemental tacit knowledge. Nicholas (2003, p. 187) andEisenstein (1979, p. 44) observe that print media transformed urban culture.14Contemporary accounts confirm that access to print media was limited outside printing centers.Platter (1839, p. 28-29) described the constraints on his education in the early 1500s: “In the schoolat St. Elizabeth, indeed, nine Bachelors of Arts read lectures at the same hour, and in the sameroom.neither had any one printed books.What was read had first to be dictated, then pointed andconstructed, and at last explained; so that the Bacchants had to carry away thick books of notes whenthey went home.”15See Nieto (2003, p. 17), Edwards (1994, p. 8), Eisenstein (1979), and section 5 for further discussion.16Edwards (1994, p. 8) observes: “If, for example, there was an interest in Strasbourg for a work firstpublished in Wittenberg, it was more common for a printer in Strasbourg to reprint the work than it7

a few texts (Febvre and Martin 1958, pp. 335-339). Contracts between printers in Lyonsand Poitiers from the late 1500s indicate that the allowance for transport costs associatedwith a journey of approximately 360 kilometers raised the sale price of transported booksby 20 percent (Febvre and Martin 1958, p. 169). Records from the archives of the Ruizmerchant family indicate that insurance and transport costs for a shipment of 21 booksfrom Lyon to Medina del Campo (280 kilometers as the crow flies) were equivalent to 30days’ wages for a skilled craftsman (Febvre and Martin 1958, p. 338). Archival holdingsprovide additional evidence on the limits on the trade in print media. The BayerischeStaatsbibliothek in Munich houses the largest and most comprehensive historic collectionof books printed 1450-1500. Figure II shows that the proportion of the editions producedin a given city and held in the Munich archives declines sharply (and non-linearly) in thedistance between the printing city and the archive.17FIGURE II – INSERT HEREPrinting cities also enjoyed benefits due to agglomeration economies. The printingpress produced new face-to-face interactions in addition to books and pamphlets. Printers’ workshops brought scholars, merchants, craftsmen, and mechanics together for thefirst time in a commercial environment, eroding a pre-existing “town and gown” divide(Eisenstein 1979, pp. 309, 521). Bookshops and the houses of printers became meeting places and temporary residences for intellectuals. Print technology also produced,in the printer-scholar, “a ‘new man’.adept in handling machines and marketing products even while editing texts, founding learned societies, promoting artists and authors,[and] advancing new forms of data collection” (Eisenstein 1979, pp. 250-251). Historical research indicates that these activities made printing cities attractive cultural andeconomic locations. Cities that were early adopters of the printing press attracted booksellers, universities, and students. Adoption of the printing press also fostered backwardslinkages: the printing press attracted paper mills, illuminators, and translators.18was for the printer in Wittenberg to ship a large number of copies [500 kilometers] to Strasbourg.”17Language barriers do not explain this phenomenon: 72% of books printed 1450-1500 were printedin Latin and the pattern holds when the sample is restricted to Latin editions. That an unusually highproportion of books printed Venice and Rome was held in foreign collections is explained by the fact thatVenice was the commercial hub and leading printing center of Europe 1450-1500 and Rome occupied aunique position as the seat of Roman Catholicism.18See Febvre and Martin (1958), Barbier (2006), Varry (2002), Fau, Saksik, Smouts, and Tisserand(2003), and Eisenstein (1979).8

4DataThis paper exploits data on the diffusion and output of printing presses over the technology’s infant industry period (1450-1500). Between 1450 and 1500, entrepreneurs established printing presses across Europe and the real price of books fell by two-thirds(van Zanden 2004; Clark 2004). Between 1500 and 1800, printing technology was largelyunchanged and declines in the price of books were relatively modest (Füssel 2005 andFebvre and Martin 1958).19 Historical research emphasizes that the period 1450-1500 wasthe “first infancy” of printing. Books produced 1450-1500 are referred to as incunabula,from the Latin for cradle or infancy (Barbier 2006; Glomski 2001; Clair 1976; Febvre andMartin 1958). Over the infant industry period supply side constraints limited technologydiffusion. As discussed below, by the early- to mid-1500s these constraints were relaxed.I construct data on the location and output of printing presses over the infant industryperiod from three principal sources. The first source is the Incunabula Short Title Catalogue (ISTC 1998) maintained by the British Library. The ISTC (1998) “records nearlyevery [incunabulum] printed from movable type before 1501.” The ISTC (1998) records27,873 printed books. Each record includes the title, publication date, and location ofpublication. The ISTC catalogues 15th century editions printed in 196 historic cities.20The second source is Febvre and Martin’s (1958) L’Apparition du Livre, which documents 181 historic cities that adopted the printing press 1450-1500. The third sourceis Clair’s (1976) A History of European Printing, which documents the establishmentof printing presses in 188 historic cities 1450-1500. As shown in Table I, the historicalsources identify 205 unique cities that adopted the printing press 1450-1500.21TABLE I – ABOUT HEREData on the locations and populations of Europe’s historic cities are from Bairoch,Batou, and Chèvre (1988). Their approach is to identify the set of cities that ever reached5,000 inhabitants between 1000 and 1800, and then to search for population data forthese cities in all periods. The data record (in thousands) the populations of urban19Clark (2004) finds that real book prices in England fell 75% between 1450 and 1530 and stabilizedat 1/3 the pre-Gutenberg level through the late 1700s. Van Zanden (2004) examines Dutch data andestimates that real prices fell by 2/3’s 1450-1500. Van Zanden estimates that between 1500 and 1800book prices declined from approximately 1/3 to 1/6 the pre-Gutenberg level.20Of the 27,873 records, 1,352 are either undated or are associated with dates outside 1450-1500 and738 indicate only a regional location or possible city locations.21This figure comprises the 196 cities from the ISTC, 4 additional printing cities identified by Febvreand Martin, 4 identified by Clair, and 1 identified by Clair and Febvre and Martin. While pressesoperated in these 9 additional cities, since we have no record of incunabula produced at these locationsthey are not recorded in ISTC (1998). Results below are not contingent on the inclusion of these cities.9

agglomerations, not simply populations within administratively defined boundaries.22These data – henceforth the “Bairoch data” – are recorded every 100 years up to 1700,and then every 50 years to 1850. The data set contains a total of 2,204 historic Europeancities. Populations are observed every 100 years 1300-1800 for a balanced panel of 202cities.23 The leading alternate source of data on historic city populations is the panel inde Vries (1984). This paper anaylzes the Bairoch data because the de Vries (1984) dataare restricted to 1500-1800 and cover only cities that reached a population of 10,000.24The econometric work below also exploits a new database on the historical characteristics of European cities, including: which cities were located on navigable rivers, seaports, and the sites of Roman settlement; which were political or religious centers; andmeasures of economic institutions. These data are described in the appendix.55.1EmpiricsOverviewPer capita income data is not available at the city level and the existing data on urbanwages is confined to a small number of cities.25 However, the consensus in the literatureon urbanization in Europe is that population size was an indicator of the overall vitalityand well-being of cities in early modern Europe.26 Moreover, city growth may reflecttechnological progress. In modern economies with mobile labor, high productivity citiesare likely to draw migrants (Glaeser, Scheinkman, and Shleifer 1995). In a Malthusianeconomic regime, or one with Lewis-style unlimited supplies of surplus labor in agriculture, technological change in the urban sector will also show up in city growth. For thesereasons, this paper focuses on the relationship between the adoption of print technologies22Bairoch, Batou, and Chèvre (1988, p. 289) include populations of, “the ‘fauborgs’, the ‘suburbs’,‘communes’, ‘hamlets’, ‘quarters’, etc. that are directly adjacent” to historic city centers. Bairoch,Batou, and Chèvre draw data from urban censuses, tax records, archaelogical work, as well as otherprimary and secondary sources. These data are examined in greater detail in Dittmar (2010).23ISTC (1998), Clair (1976), and Febvre and Martin (1958) identify printing presses at some locationsthat do not appear in the Bairoch city data. These were overwhelmingly non-urban religious establishments (principally monasteries). Other “missing” print centers were adjacent to cities that did havepresses and represent a sort of duplication. Westminster with its proximity to the city of London is anexample. In keeping with the economic understanding of urban agglomeration, and the construction ofthe Bairoch data, this paper treats production of print media at Westminster as London output.24Dittmar (2010) analyses and compares these data in greater detail.25Allen (2007) compiles data on real wages in 20 cities. These extend to the early 1400s for only 8cities, all of which adopted printing 1450-1500. Similar coverage is available in the data collected by theGlobal Price and Income History Group (UC Davis) and the International Institute for Social History.26See Acemoglu, Johnson, and Robinson (2005), Bairoch (1988), and de Vries (1984).10

and city growth.Because data on the number of presses in operation are only available for a smallsubset cities, and because the available measures of output are coarse, I focus on adoption. Data on the production of incunabula editions provide valuable but imperfectmeasures of production. Pamphlets, booklets, and other ephemera constituted a large,unmeasured share of output. The production of ephemera was less concentrated thanthe production of expensive books and the inter-city trade in ephemeral forms of printmedia was relatively limited: historians designate these ephemeral media as “city printing” (l’imprimerie de ville).27 These media played an important role in the developmentof literacy and print culture that measures of book production may not capture. Thesefacts support an emphasis on the printing press itself.5.2Comparison of Average OutcomesThis section compares the population growth of cities that were early adopters of printtechnology to the growth of cities that were not. It documents that cites in which printingpresses were established in the late 1400s grew relatively quickly 1500-1600.28Table II compares, by country, the growth 1500-1600 of cities that were early adoptersto the growth of cities that were not. It includes all cities for which population data isavailable. It shows that, on average, cities that adopted the press in the late 1400sgrew 20 percentage points more than non-printing cities 1500-1600. Table II also showsthat the cities that adopted were unusually large: 30 percent of cities with populationdata adopted, but adopting cities accounted for 58 percent of total urban population in1500. Moreover, the Netherlands stand out as an economy in which printing

The movable type printing press was the great innovation in early modern information technology. The rst printing press was established in Mainz, Germany between 1446 and 1450. Over the next fty years the techology di used across Europe. Between 1450 and 1500, the price of books fell by