pamuk & williamson, ottoman deindustrialization, 1800,1913

Ottoman de-industrialization,18001913: assessing the magnitude,

impact, and response1By SEVKET PAMUK and JEFFREY G. WILLIAMSON

India and Britain were much bigger players in the eighteenth-century world market formanufactures than were Egypt, the Levant, and the core of the Ottoman Empire, butthese eastern Mediterranean regions did export carpets, silks, and other textiles toEurope and the east. By the middle of the nineteenth century, they had lost most oftheir export market and much of their domestic market to globalization forces andrapid productivity growth in European manufacturing. How different was the Otto-man experience from the rest of the poor periphery?Was de-industrialization more orless pronounced?Was the terms of trade effect bigger or smaller? How much of Otto-man de-industrialization was due to falling world trade barriers such as oceantransport revolutions and European liberal trade policy, how much due to factory-based productivity advance in Europe, how much to declining Ottoman competitive-ness in manufacturing, how much to Ottoman railroads penetrating the interior, andhow much to Ottoman policy? This article uses a price-dual approach to seek theanswers.ehr_560 159..184

What do we mean by the term de-industrialization? To simplify, define acountry as producing only two commoditiesagricultural goods which areexported and manufacturing goods which are imported, with three factors ofproductionlabour which is mobile between the two sectors, land which is usedonly in agriculture, and capital which is used only in manufacturing. Assumefurther that our country satisfies what trade economists call the small countryassumption, and thus that it takes its terms of trade as exogenous, dictated byworld markets, a condition we think applies to the Ottoman Empire. Under theseconditions, de-industrialization can be defined most easily as the movement oflabour out of manufacturing and into agriculture.

If a country de-industrializes because its comparative advantage in agriculturehas been strengthened either by productivity advance on the land, by settling moreland, by increasing openness in the world economy, or by all three, then GDPincreases in the short run. In the case of more land or more productive land, thenthe small country faces no change in its terms of trade.2 In the case of increasing

1 We are grateful for the excellent research assistance supplied by Kyle Nasser and Miray Topay. We have alsobenefited from the useful advice and criticism offered by three anonymous referees of this journal, and by BobAllen, Greg Clark, Metin Cosgel, David Clingingsmith, Rafa Dobado, Aurora Gmez Galvarriato, PatrickOBrien, Kevin ORourke, Roger Owen, Michael Palairet, Leandro Prados de la Escosura, Ananth Seshadri,TonyVenables, and Tarik Yousef. Williamson acknowledges with pleasure financial support from the Harvard Facultyof Arts and Sciences and Pamuk from the Academy of Sciences of Turkey.

2 Alternatively, if the country is large and has an influence on world prices, then it will suffer a terms of tradedeterioration, in that it has to share part of the labour productivity gains in the agricultural export sector with itstrading partners.The large country conditions probably began to apply to Egypt and its cotton exports after the1860s, but Egypt is not included in our Ottoman Empire definition. See section I and Brown, ed., Imperial legacy,pp. xiiixvi.

Economic History Review, 64, S1 (2011), pp. 159184

Economic History Society 2010. Published by Blackwell Publishing, 9600 Garsington Road, Oxford OX4 2DQ, UK and 350 MainStreet, Malden, MA 02148, USA.

openness, the country enjoys an unambiguous terms of trade improvement asdeclining world trade barriers raise export prices and lower import prices in thehome market.3 Whether real wages in towns and living standards in villages alsoincrease depends on the direction of the terms of trade change, whether the exportcommodities include foodstuffs, and whether foodstuffs dominate the budgets ofpoor families. That is, it depends on whether the export commodities are wheat,barley, and tobacco (a big share of the workers budget) or silk and opium (a smallshare). Whether GDP growth rates rise in the long run depends on whetherindustry generates accumulation and productivity externalities that agriculturedoes not. If industrialization is a carrier of growth, as most growth theories imply,4

then de-industrialization could lead to a growth slowdown and a low-incomeequilibrium that gives the notion of de-industrialization its power in the historicalliterature.

This article documents Ottoman experience with its terms of trade over thecentury 18001913, explores the connection between de-industrialization andthose external price trends, and then compares this experience with that of the restof the eastern Mediterranean, as well as with Asia, Latin America, and the Euro-pean periphery. Section I sets the stage by reviewing Ottoman experience withtrade policy, world transport costs, Ottoman railroads, and thus with world marketintegration. Section II reviews the de-industrialization debate as it applies to theOttoman Empire and the rest of the eastern Mediterranean, while section IIIassesses the de-industrialization evidence. Section IV reports the external terms oftrade estimates for the Ottoman Turkish and Balkan core, for the Levant in theEmpires eastern wing, and for Egypt, which was more autonomous. Section Vdiscusses the economics of de-industrialization, while section VI seeks to answerthe following questions: how much of Ottoman de-industrialization was due tofalling world trade barrierstransport revolutions and European liberal tradepolicyand how much due to factory-based productivity advance in Europe, howmuch to changing Ottoman competitiveness in manufacturing, and how much toOttoman policy?

I

The Ottoman Empire covered a large part of the eastern Mediterranean for mostof the nineteenth century. Around 1820, it stretched from present-day Bosnia inthe western Balkans through central and northern Greece, Macedonia, Bulgaria,and Turkey to Syria, Lebanon, Palestine, and Iraq. Romania was formally part ofthe Empire until the middle of the century and Egypt was until the First WorldWar, but we choose to exclude them as they had different endowments, commod-ity specialization, and significant autonomy. The 1820 population of the Empire(excluding Romania and Egypt) was around 21 million or about 2 per cent ofthe world population. The Empire experienced significant territorial losses in theBalkans throughout the century, but the Ottoman 1880 population (excluding thetwo same areas) was still around 21 million. On the eve of the BalkanWars in 1912,

3 Throughout this article, the terms of trade refers to the net barter terms of trade, that is, the ratio of averageexport prices to average import prices.

4 For example, Matsuyama, Agricultural productivity; Helpman, Mystery.

160 SEVKET PAMUK AND JEFFREY G. WILLIAMSON

Economic History Society 2010 Economic History Review, 64, S1 (2011)

the total population of the Empire had increased to about 25 million whichequalled about 1.5 per cent of the world population.

In response to the influx of cheaper manufactured goods from Britain and therest of western Europe, de-industrialization was the norm for most peripherycountries during the nineteenth century, including the Ottoman Empire.The greatMiddle East scholar Issawi placed the Ottoman case in this general context:

The Revolutionary and Napoleonic wars gave the region a respite, but in the 1820s and30s it was hit by the full blast of European competition. Factories were pouring outcheap goods, and peace and increased security in the Mediterranean and improvementin shipping made it possible to land them at low costs. To this should be added theeffects of the various commercial treaties, which froze import duties at low levels andopened up the regions markets.5

Thus, the 1838 Treaty is one of the most widely discussed events pertaining tothe collapse of Ottoman industry in the nineteenth century, although liberalOttoman reforms began to appear earlier in 1826. Most European industrialcountries were still protectionist before they went liberal in the 1860s,6 but Britainwent open decades earlier. Indeed, Britains agenda was to sign free trade agree-ments with as many periphery countries as possible in order to gain foreignmarkets for their manufactures.7 The 1838 Anglo-Turkish Commercial Conven-tion was one such commercial agreement. The original treaty was signed in BaltaLiman in August of 1838 and went into effect in March of 1839. The Anglo-Turkish convention eventually became the basis of practically all foreign trade inTurkey as the Ottoman Empire signed similar treaties with several other Europeancountries in the following three years.8 This Convention was viewed as the nextstep in the Empires transition to economic liberalism after the sultan eliminatedthe Janissary corps in 1826. These urban guildsmen, on the military payroll, werethe strongest advocates of protectionism.9

The 1838 Treaty eliminated all local monopolies, allowed British merchants tobuy goods anywhere in the Empire, and exempted foreign (but not domestic)merchants from an 8 per cent internal customs duty that had been levied previ-ously on goods transported within the Empire.10 Export duties were raised from 3to 12 per cent and import duties from 3 to 5 per cent.11 One of the present authorshas argued previously that the importance of these treaties has been overstatedsince they did not represent a drastic revision from the liberal course already set in1826.12 Quataert seems to agree: Commerce between [Britain and the OttomanEmpire] already was increasing dramatically: British exports to the empire haddoubled in value during the late 1820s and doubled again before 1837.13

5 Issawi, Middle East and North Africa, p. 151.6 Bairoch, European trade policy; ORourke and Williamson, Globalization.7 In Asia, gunboats were often used as a powerful arm of that policy, by Britain and the US.8 Puryear, International economics, p. 3.9 Quataert, Age of reforms, p. 764.

10 This must have placed local industry at a significant disadvantage relative to foreign products, yet theliterature offers no explicit assessment of its impact, nor of the political economy that produced it.

11 Issawi, Middle East, 18001914, p. 38. Export duties are a tax on trade and thus offer partial relief toimport-competing industries. It appears that the literature has failed to appreciate this fact, since it offers noexplicit assessment of its impact.

12 Pamuk, Ottoman Empire and European capitalism, p. 20.13 Quataert, Age of reforms, p. 825.

OTTOMAN DE-INDUSTRIALIZATION 161


However, by locking the government into fixed import duties, these treaties didprevent the Ottomans from providing any subsequent protection to domesticindustry. Indeed, between 1865 and 1905, average tariff rates were fairly stable at7.5 per cent in the Empire. These low revenue-producing tariffs were consistentwith free trade policy, and they were equivalent to those prevailing in Asia. Incontrast, average tariff rates in protectionist Latin America and the US were about30 per cent in the middle of the period, or more than four times that of theOttoman Empire.14

It may seem odd that the literature never mentions political pressure exerted byexport interest groups who, after all, stood to gain significantly from the treaties,and how they came to overwhelm import-competing interest groups, when theopposite was true of Latin America and the US. While this article does not offerany new evidence on this issue, one can imagine that the difference might havebeen attributed to two factors. First, the export-oriented agricultural producerswere not well organized in the Ottoman Empire, but they were in Latin Americaand the US. Second, the Ottoman government signed the free trade treaties inorder to obtain the political support of Britain against the threats from Russia andMohammed Ali of Egypt, not because of pressure from export interests. Suchland-based threats from major powers were absent in Latin America (except forMexico) and the US.The elimination of monopolies also had a powerful short-runimpact on de-industrialization, as we shall see in section III.

The move to free trade and the dramatic decline in transportation costs con-tributed to a boom in Ottoman trade during the nineteenth century, especiallywith western Europe.15 Imports increased from 5.2 million in 1840 to 39.4million in 1913, at about 3.3 per cent per annum, and since the prices of thetraded commodities were considerably lower on the eve of World War I than in1840, the increases in trade volumes were actually greater.16 Between 1840 and1873, trade grew even faster, with the volume doubling every 11 to 13 years.17

Placed in a comparative framework with other periphery countries:

[the] Ottoman-center trade grew faster than the periphery-center trade (from the1820s) until the early 1870s, but the rate of growth of Ottoman exports lagged behindthe rate of growth of total exports from the periphery after the 1870s . . . for the period18401913 as a whole, per capita exports from the Ottoman Empire expanded at ratesclose to but lower than those of per capita world trade and per capita center-peripherytrade.18

We note therefore that trade grew faster before 1870 than after.Steamships and railroads constitute the two major transport innovations that

contributed to the booming Ottoman trade during the period. Steamships couldbe built much larger than their sail counterparts and their rise to dominancelowered freight costs and stimulated trade for the region. Steamers were intro-duced into the eastern Mediterranean in the late 1820s, and they became a

14 The figure is for average tariff rates, but they were even higher on imported manufactures; Williamson,Globalization and the poor periphery, pp. 10916.

15 Harlaftis and Kardasis, International shipping.16 Pamuk, Ottoman Empire and European capitalism, p. 23.17 Ibid., p. 30.18 Ibid., p. 37.



common sight after the 1840s. Indeed, all sea-going countries established steam-ship companies between the 1830s and the 1850s, which competed for car-go . . . transportation in the Mediterranean.19 More to the point, the freight factoron wheat fell by nearly 80 per cent on London imports from the Black Sea and theMediterranean.20 The decline in freight rates between 1870 and 1914 was just asdramatic on routes involving Black Sea and Egyptian ports as it was in the Atlantic,perhaps even more so.21 Shipping and trade boomed in the Mediterranean. Forexample, shipping tonnage entering Beirut went from 40,000 in 1830, to 600,000in 1890, and to 1,700,000 in 1913. Other cities in the eastern Mediterranean showsimilar shipping tonnage increases.22

Over the nineteenth century as a whole, the Ottoman Empire was more activein world trade than either Asia or Africa, but less so than Latin America.Why? Didthe Ottoman Empire undergo more dramatic improvements in its terms of tradethan did Asia and Africa, but less dramatic than Latin America? Thus, did itundergo greater de-industrialization than Asia and Africa, but less than LatinAmerica?

II

De-industrialization is defined as curbing production and factor employment inmanufacturing, either in factories or in cottage industries. In the simple two-sectormodel set out in the introduction, de-industrialization can be gauged as themovement of labour out of manufacturing and into agriculture, either measured asa fall in the total employment share, or as a fall in absolute numbers. Bothmeasures could, in theory, also be constructed for value added, but such evidenceis almost impossible to find for the nineteenth-century periphery. Alternatively,absolute de-industrialization can be supported by information on the capital stockin manufacturing, such as the number of looms and spindles. In any case,de-industrialization is easy enough to define. However, the debate over the mag-nitude of de-industrialization, its causes, and its economic impact is more impor-tant and contentious.23

The literature dealing with Ottoman and Middle Eastern nineteenth-centuryde-industrialization exhibits a fair amount of solidarity, yet the de-industrializationnarrative is usually supported only by qualitative accounts, anecdotal evidence, orspotty time series, making long period quantitative analysis difficult. As if toprovoke more and better work, 25 years ago Kurmus commented that anecdotalevidence was being accepted as constituting some part of the historical truth

19 Harlaftis and Kardasis, International shipping, p. 246.20 G. Harlaftis and G. Kostelenos, Services and national economic growth: calculating the shipping income in

the Greek economy, mimeo. (2007).21 Harlaftis and Kardasis, International shipping; Shah Mohammed and Williamson, Freight rates.22 Issawi, Middle East and North Africa, p. 48.23 Quataert, Age of reforms, p. 898, cites the Indian de-industrialization literature in pointing out termino-

logical problems: Some India scholars suggest that the term [de-industrialization] be abandoned altogetherbecause its meaning is not clear. They argue that there are too many holes in the data to support the level ofgeneralization that use of the term implies. We disagree. Labels are useful for narratives, but theory andmeasurement are the keys to understanding the de-industrialization experience, regardless of descriptive labels.



which would serve to substitute antiquarianism for scientific work.24 Kurmusschallenge produced a response by one of the current authors whose book will beused extensively in this survey.25 Still, the previous debate has rarely been engagedin comparative terms. How did the Ottomans perform relative to the rest of thenineteenth-century poor periphery?Where the performance was different, why wasit different?

As is true with the de-industrialization literature on other parts of thenineteenth-century poor periphery, Ottoman industry has too often been consid-ered solely as capital-intensive, large-scale, urban, factory production since thosewere the modes of production in Britain and the rest of Europe by mid-century.This is a mistake when analysing the Ottoman Empire and the rest of the periph-ery during the nineteenth century (or even Britain before 1780), when mostmanufacturing was labour-intensive, small-scale, household-based, and rural.Indeed, Mechanized factory output was and remained relatively insignificant inthe nineteenth century when compared with domestic and handicraft produc-tion.26 Since cotton spinning and hand weaving were performed part-time byfamily members using extremely simple technology, it may seem implausible toargue that the demise of local textile production destroyed a nineteenth-centuryOttoman platform for modern industrialization.Yet economic historians assign thesame importance to home-based cotton spinning and weaving in Britain: proto-industrial cottage industries are said to have supplied the platform for the factory-based British industrial revolution that followed in the late eighteenth century.27

Furthermore, employment of women and children was central to the process thentoo.28 Hence, this article will consider both cottage and factory industry in theOttoman Empire, even though the data are often sparse for the former.

Finally, it might be argued that the Ottoman Empire was more heterogeneousthan was much of the rest of the poor periphery, primarily due to geography.Thus,given high pre-railroad, land-based transport costs, the impact of globalization wasfar greater on the Ottoman coast than in its interior, and far greater in its lowlandsthan in its mountains.While this article deals with Empire aggregates and averages,throughout the reader will be reminded of this regional heterogeneity especiallyprior to the railroads.

III

The traditional view of Ottoman manufacturing is that it steadily collapsed in thewake of the influx of European manufactured goods. The Napoleonic Wars haddisrupted international trade and shielded the eastern Mediterranean from theimpact of the industrial revolution. However, local textiles began to retreat soonafter the wars ended in 1815. The western provinces of the Empire, the Balkans,and western Anatolia were first to face the impact of imports. It is clear, forexample, that the strong manufacturing activity in Thessaly in central-northern

24 Kurmus, 1838 Treaty, pp. 41112.25 Pamuk, Ottoman Empire and European capitalism.26 Quataert, Age of reforms, p. 898.27 Mendels, Proto-industrialization; Hayami, Great transformation; de Vries, Industrial revolution; Weis-

dorf, Domestic manufacturing; and see Petmezas, Patterns.28 de Vries, Industrial revolution; idem, Industrious revolution.



Greece reached its peak in the 1810s.29 Consular and travellers reports are riddledwith anecdotal accounts relating the demise of industry.30 Certainly de-industrialization did occur between about 1815 and 1860, and in response to threeforces. First, the end of the Napoleonic Wars opened up the gates for Britishexports of manufactures. Britain had already accumulated productivity advantagesin manufacturing, and peacetime conditions made it possible for the leadingEuropean economy to exploit them in world markets. Second, additional rapidproductivity advance in British manufacturing increased its competitive edge inforeign markets still further. Third, the move to liberal policies in the OttomanEmpire between the 1830s and the 1850s deepened the de-industrialization shockstill further. After all, the terms of trade soared between 1815 and the late 1850s(figure 1), and it more than doubled over the two decades after the late 1830s. Itincreased by 2.6 times between 1800 and 1860 (table 2).

The Ottoman Empire was completely self-sufficient in cotton and woollentextiles until about 1820, but the deluge of cheap European industrial goodschanged all of that. Pamuk reconstructs the decline of Ottoman cotton textilesby using an identity to estimate the domestic consumption and production of

29 Petmezas, Patterns, pp. 593601.30 Quataert, Age of reforms, p. 888.

0

50

100

150

200

250

300

1796

1802

1808

1814

1820

1826

1832

1838

1844

1850

1856

1862

1868

1874

1880

1886

1892

1898

1904

1910

Ter

ms

of t

rade

Egypt

Ottoman Empire

Middle East

Levant

Figure 1. Middle East: net barter terms of trade, 17961913Sources: Pamuk, Ottoman Empire and European capitalism; Williamson Globalization and the great divergence, fig. 7.



textiles for the areas within the 1911 borders of the EmpireMacedonia,present-day Turkey, greater Syria, and Iraq.31 He identifies the period from the1820s to the mid-1870s as the crucial one for the decline of cotton handicrafts,a result consistent with the fact that the biggest terms of trade improvementtook place in this period, as we shall see in sections IV, V, and VI. Weaverssuffered, but domestic spinning also declined dramatically in the face of importcompetition: spinning output fell from 11,550 tons per year in 18202, to 8,250tons in 18402, to 3,000 tons in 18702.32 Note that the biggest collapse inlocal spinning output was between 1840/2 and 1870/2 (74 per cent), notbetween 1820/2 and 1840/2 (29 per cent), an observation consistent with thefact that the external price shock was much bigger after the 1830s (figure 1).33 InSyria, in the eastern wing of the Empire, de-industrialization forces hit hardtoo.34 Aleppo was estimated to have had 40,000 handlooms in the eighteenthcentury, while the numbers were down to 25,000 in the 1820sa 37 or 38 percent fall from the eighteenth-century highs, and averaged 5,125 between 1838and 1850an 80 per cent fall from the 1820s. Damascus was estimated to havehad 34,000 handlooms in the eighteenth century, while the numbers were downto 12,000 in the 1820sa 65 per cent fall from the eighteenth-century highs,and averaged 2,355 between 1838 and 1850an 80 per cent fall from the1820s. One wonders how much of the decline between the eighteenth centuryand the 1820s can be attributed to competition from modern British factoriesafter 1815 and the peace, and how much of it can be attributed to the lateeighteenth century when Britain, before the factories, began to take over theAtlantic and European market from India, the Middle East, and other competi-tors.35 In any case, we note again that de-industrialization was dramatic up toabout 1860, and so too was the terms of trade boom.

One exception to this clear and strong de-industrialization pattern was theupland, mountainous areas of Ottoman Bulgaria, where woollen cloth productionfor local consumption and to some extent for urban markets in the neighbouringareas of the Empire survived, partly due to the barriers of geography and scarcityof good agricultural land and partly due to reasons of climate since the Bulgarianpopulation favoured woollen over cotton textiles to a much greater degree thanthe rest of the Empire. Thus, small-scale proto-industrial production of woollencloth production continued to rise until the 1870s, as Palairet emphasizes in hisimportant research.36 This was no modest regional exception, since the popula-tion of Bulgaria was about 2.5 million or 12 per cent of the total Empire aroundthe middle of the nineteenth century, but it is an exception that lasted until the1860s. As Palairet makes clear, commercial production of woollen cloth in

31 Pamuk, Ottoman Empire and European capitalism, pp. 11326, 1914, for the details of the reconstruction.32 Ibid., p. 118.33 Even if we follow Quataerts suggestion (Age of reforms, pp. 9057) and raise the estimates for local

spinning in 18702, the decline in overall production is clear and the half century before the 1870s remains byfar the most important period of declining Ottoman manufacturing activity.

34 Issawi, Fertile crescent, p. 374.35 Clingingsmith and Williamson, De-industrialization.36 Palairet, Balkan economies, pp. 6684. Since Bulgaria was the industrial powerhouse of the empire before the

1870s, its loss after the Russo-Turkish War of 18778 (and subsequent Ottoman tariffs on Bulgarian imports)may have given stimulus to the industries of the remaining territories, but this possibility is not explored here.



Bulgaria declined by more than 40 per cent from the 1860s to the first decade ofthe twentieth century.37

Table 1 offers another de-industrialization indicator for the nineteenth-centuryOttoman Empire, where comparisons with the rest of the Third World are offered.For four Third World regionsMexico, the Ottoman Empire, India, andIndonesiawe measure the loss of domestic textile manufactures markets toforeign imports. That is, the figures report the share of domestic consumptionsupplied by local and foreign sources. Take India first. It has been estimated (notin table 1) that Bengal exported about 21 per cent of its domestic production in1750.38 That figure had fallen, at least for India more generally, to 6 or 7 per centby 1800. Thus, even before the onset of the factory-led industrial revolution inBritain, India had lost a large chunk of its export market. By 1833, India had lostall of its (net) export market and 5 per cent of its domestic market. By 1877, thede-industrial damage was done, with domestic producers claiming only 3542 percent of their own home market. Although the Ottoman Empire did not have a largeforeign market to lose, it underwent a similar dramatic collapse in its home marketshare, domestic producers undergoing a fall in their share from 97 to 2535 percent over the half-century between the 1820s and the 1870s. It fell again after the1870s, but at a much slower rate, from 2535 to 2026 per cent.The decline in theIndonesian (or Dutch East Indian) textile industry was only a little less spectacularthan in the Ottoman Empire since the local producers share of the home marketfell from 82 to 38 per cent between 1822 and 1870. However, in the case of theDutch East Indies de-industrialization persisted much longer, with the local pro-ducer share falling still further to about 11 per cent in 1913. Finally, Mexico seemsto have resisted these de-industrialization forces more successfully, since domesticproducers could still claim 60 per cent of its home market in 1879.39 In short, the

37 Ibid., p. 296.38 Chaudhury, From prosperity.39 Dobado Gonzlez, Gmez Galvarriato, and Williamson, Mexican exceptionalism, tab. 3, p. 17.

Table 1. Comparative de-industrialization: textile import penetration, 1800s80s,around the ThirdWorld

% of home textile market supplied by:

Foreign imports Domestic industry

India, 1800 -6 to -7 1067India, 1833 5 95India, 1877 5865 3542Ottoman Empire, 1820s 3 97Ottoman Empire, 1870s 6575 2535Ottoman Empire, 1910s 7480 206Indonesia, 1822 18.1 81.9Indonesia, 1870 62 38Indonesia, 1913 88.6 11.4Mexico, 1800s 25 75Mexico, 1879 40 60

Sources: Williamson, Trade and poverty, tab. 5.3, for all but Ottoman Empire observations, which are calculated from Pamuk,Ottoman Empire and European capitalism, p. 115.



Ottoman Empire had one of the more dramatic de-industrialization episodes,although it should be stressed that these calculations apply to cotton textiles onlyand that they exclude Bulgaria.

From the 1870s or 1880s until the First World War, the rapid decline in textilemanufacturing slowed down across the Empire and it may have been even reversedin some areas despite low tariff barriers up to the First World War. Symmetrically,the rapid growth in imports from Europe also slowed down.40 For the Empire asa whole, hand spinning continued to decline but the volume of weaving probablydoubled between 1880 and the First World War. One can also discern somedifferences during this period between the western and eastern parts of the Empirewith respect to the arrival of new technology. In Bulgaria and Macedonia (includ-ing the Salonica region), factory production of yarn and cloth began to grow in the1880s, followed a little later by the Istanbul region.41 In addition to the factoryproduction of cloth, small-scale weavers in these areas began to use the yarnproduced by the new domestic mills as well as the imported yarn to produce cloth.

Factories with power-driven machinery were more limited in the eastern half ofthe Empire, in eastern Anatolia, Syria, and Iraq. In these areas, increases in textilemanufacturing occurred primarily in weaving in small-scale urban workshopswhich began to use imported yarn to expand their output of cotton and mixedcotton cloths to meet local demand.42 Significant amounts were shipped to long-distance markets in Anatolia, Syria, and Egypt. Indeed, the handloom numbers inboth Aleppo and Damascus show no trend at all between 1860 and the FirstWorldWar, after a long secular decline.43 In the rural areas, hand spinning continued todecline but the decline of hand weaving slowed down and may have been evenreversed in some areas.

In short, we observe that de-industrialization forces were powerful up to the1860s but weaker thereafter.This pattern applies to both the western and easternhalves of the Empire, even though the emergence of textile factories with power-driven machinery was stronger in the west. This slowdown if not reversal ofde-industrialization is consistent, as we shall see, with the reversal in terms of tradetrends, from a rise up to the mid-late nineteenth century to a gradual fall there-after, when the relative price of manufactures rose, thus giving local industry somerelief.

Not all industries were damaged by foreign competition, however. Due to anincreased world demand, carpet making, copper work, earthenware, inlaid wood-working, lace making, silk reeling, and embroidery were able to thrive despiteforeign competition and trade liberalism.44 Indeed, some of the above industries,and even the textile industry itself, persisted until the First World War and after.45

Some have argued that Pamuk may have overstated the extent to which handi-crafts suffered. Quataert thinks he understated hand-spun yarn production, and

40 Pamuk, Ottoman Empire and European capitalism, pp. 307.41 Palairet, Balkan economies, pp. 24397, 34656; Gounaris, Steam; Lapavitsas, Industrial development; also

Todorov, Balkan city, pp. 238306.42 Quataert, Ottoman manufacturing, pp. 21104, documents this rise in urban weaving but has less to say about

what happened in the rural areas; also Owen, Middle East, pp. 24486; idem, Study.43 Issawi, Fertile crescent, p. 374.44 Issawi, Middle East and North Africa, p. 153; Quataert, Age of reforms, p. 890.45 Unfortunately, these industries are not as well documented as are textiles, so any empirical explanation for

their different response to foreign competition cannot be assessed.



thus that his estimates of domestic production and consumption of cotton textilesmust be in error. Hence, Quataert suggests that handspinning remained animportant source for Ottoman textile producers and accounted for a considerableproportion of all cotton yarn being used in the empire, at least 25 per cent inc.1900.46 While he also claims that Pamuks reconstruction of the cotton textilesdecline does not account for the fact that imported yarn and cloth may havecreated new domestic jobs, Quataert fails to offer any explicit revision of Pamuksreconstruction, nor is Quataert explicit about the size of the demise of local textileproduction in the face of foreign competition. Thus, we will stick to Pamuksestimates and the overall decline they suggest with the exception of an upwardsrevision in the volume of local spinning.

Still, although the domestic import-competing industry suffered, why was it ableto survive the foreign competition?There are a number of hypotheses suggested inthe literature. First, perhaps foreign goods were not able to penetrate into regionsdistant from major trade routes or ports, especially before the railway boom late inthe century. Ottoman geography, including the mountainous areas of Bulgaria,offered some protection to local industry, just as it did in the Latin Americaninterior.47 This pre-modern geographic environment is especially relevant becauseit speaks to the impact of the transport breakthroughs that occurred in much of theMiddle East very late in the century. Denser rail networks and more navigablewaterways and roads lowered transportation costs, and these created price con-vergence and internal trade.48 Thus, since European imports could not be broughtcheaply into the more remote areas of the Empire during most of the century,domestic manufactures could still supply local demand in those parts of theinterior. High transportation costs also provided considerable protection todomestic producers of bulky, non-textile goods from import competition even insome coastal areas.

Second, domestic tastes afforded Ottoman handicrafts some staying power.Although British companies attempted to imitate Ottoman styles, often they couldnot do so satisfactorily, and thus there was still demand for domestic cloth,including cotton, woollen, and mixed varieties of cloth.49 Their knowledge of localpreferences helped domestic manufactures survive in the short run, and theimport of foreign techniques and foreign managers increased their efficiency andcompetitiveness in the longer run. For example, local textile makers eventuallybecame more familiar with synthetic dyestuffs which allowed them to import plaincloth and take advantage of lower-priced Ottoman labour to dye it.50 Third, Issawiargues that weavers were able to cut their costs greatly by using imported yarn;thus the Industrial Revolution, which had wiped out the spinners, gave the weaversa precarious reprieve.51 This effect worked through changes in the relative pricesof local and imported yarn and cloth and was important in both the western andthe eastern halves of the Empire after the 1870s, as we have shown. A fourth

46 Quataert, Ottoman manufacturing, p. 40.47 Coatsworth and Williamson, Always protectionist?; Brtola and Williamson, Globalization; Williamson,

Explaining world tariffs.48 Yousef, Egyptian commodity markets, pp. 3546.49 Pamuk, Ottoman Empire and European capitalism, p. 124.50 Quataert, Age of reforms, p. 889.51 Issawi, Middle East and North Africa, p. 152.



potential explanation also seems promising, namely, that the late nineteenth-century resistance of Ottoman handicrafts can be explained by the fact that theterms of trade moved against the Ottomans between 1873 and 1896.This relativeprice trend reversal may have afforded import-competing sectors, like local handi-crafts, some relief. For all of these reasons, domestic handicrafts and industry,although badly hurt by foreign competition in the first half of the nineteenthcentury, resisted and adapted in the second half of the nineteenth century and inthe early twentieth century.

Large capital-intensive factories were not common in the Ottoman Empire, norwould they have made much economic sense given the factor endowments of theregion.52 There were, however, two waves of Ottoman factory building during thisperiod. The first wave occurred before the treaties, consisting of state-ownedfactories like Mohammed Alis prototype establishments in Egypt, and, to a lesserextent, like those of Mahmoud II around Istanbul. While they were protected bymonopolies from the beginning, these factories suffered from great inefficiencies,including lack of fuel and metallic raw materials and the total absence of skilledlabour.53They were also poorly managed with military elite serving as supervisors,rather than young men that had been technically trained in Europe. In addition,these factories had dilapidated machinery and inadequate power sources, both ofwhich would have been very expensive to remedy.The liberal treaties assured thesefactories a short life, but Owen believes that they would have been forced to shutdown even in the absence of free trade.54 The second wave ensued after the 1870sand consisted of factories funded and operated by private interests, and whichreceived little or no financial support from the state. It might be useful to empha-size once again that there was a difference in this respect between the western andeastern halves of the Empire, and it applied to both textile and non-textile manu-facturing. The rise of the factories was stronger in the west, from Bosnia toMacedonia and Bulgaria and later in the Istanbul region. For Anatolia and theMiddle East, Quataert states that these factories expanded rapidly in number, butIssawi and Pamuk both argue that this second wave was also modest and that theirproduction remained small compared to handicrafts.55 The latter are probablycloser to the mark but that second wave certainly looks like a supply response toa much more favourable world price environment. While Turkish (no longerOttoman) industrialization really took hold only in the 1930s,56 it might berelevant to point out that while tariffs and quotas on imports increased sharplyafter 1929, the terms of trade also took a nose dive, that is, the relative price ofmanufactures rose during that decade.

Quataert proposes three additional reasons for the lack of capital-intensiveindustrialization in the Ottoman Empirepopulation density, religion, and wars.57

Population density may have fostered industry by creating local demand: the

52 R. C. Allen, The industrial revolution in miniature: the spinning jenny in Britain, France, and India, OxfordUniversity, Department of Economics working paper, 375 (2007); idem, British industrial revolution, pp. 13555.

53 Issawi, Middle East and North Africa, p. 154; Clark, Ottoman industrial revolution, pp. 716.54 Owen, Middle East, pp. 72, 76.55 Quataert, Age of reforms, p. 901; Issawi, Middle East and North Africa, pp. 1558; Pamuk, Ottoman Empire

and European capitalism, p. 127.56 Issawi, Middle East and North Africa, p. 159.57 Quataert, Manufacturing and technology, pp. 610.



Balkans had the greatest population density in the Empire and the most pervasiveindustry, while the Arab provinces were very thinly populated and had the leastindustry. Low population density meant small markets and high landlabour ratiosmeant reduced labour supply for industry. Religion may have inhibited technologytransfer to the extent that Muslims, who constituted the majority of the popula-tion, may have been reluctant to adopt the ideas of Christians and Jews, whounderstood the technology. Finally, wars waged by the Ottomans may also havedelayed potential technology transfer at a time when such assimilation would havebeen crucial. No hard evidence is offered in support of these three propositions. Inits absence, we will explore more mundane but, we think, more promisingexplanationsfactor endowments and the external terms of trade. There wereundoubtedly many factors that contributed to Ottoman de-industrializationduring the nineteenth century and the slow industrialization afterwards, but wethink that the long-term trends in relative prices help explain a large part of theirtiming, extent, and spatial pattern.

IV

Figure 1 plots the net barter terms of trade for the Ottoman Empire (the ratio ofthe price of exports PX to the price of imports PM) covering the century 1800 to1913. The series has two parts. Pamuk has already estimated the terms of tradefollowing 1854, so what follows stresses the extension backwards to 1800 recentlyconstructed by the other author of the present article, and further revised intable 2.58 As it turns out, the first half of the century was the most crucial half. Forcomparison, figure 1 also plots Egypt, 17961913; the Levant, 18391913(present-day Iraq, Israel, Lebanon, Palestine, Jordan, and Syria); and the MiddleEast region as a whole.

Like most of the periphery, the Ottoman Empire specialized in the export ofprimary products, while importing manufactures, so between 1800 and 1854 PXrefers to an unweighted average of wheat, wool, raisins plus figs, tobacco, opium,and raw silk. As appendix 1 indicates, prices for the first four are taken from USmarkets, the price of raw silk is from British markets, and the price of opium isfrom the (infamous) Calcutta market.To derive PX, these six prices are taken as anunweighted average since the share of each commodity export in total exports isnot available until 1879, and those weights changed dramatically over the decadesbefore the late 1870s. PM refers to what was primarily manufactured goods andintermediate inputs, and it is proxied by the British export price index.

Between 1800 and the late 1860s, the price of British manufactured exports fellfar faster than did the price of British imported primary products, swinging theterms of trade against Britain and in favour of the periphery. If PM for the OttomanEmpire moved anything like PX for Britain, then the Ottoman Empire certainlyunderwent an impressive terms of trade boom over the half-century before theCrimean War.59 The magnitudes were enormous: over the four decades between

58 Pamuk, Ottoman Empire and European capitalism; Williamson, Globalization and the great divergence. PX/PMin tab. 2 and fig. 1 are exactly the same for 18601913. However, tab. 2 explores the extent to which c.i.f. andother adjustments to PM make a difference in trends. They do, documenting an even bigger PX/PM boom up to1860. See below, section VI.

59 Pamuk, Ottoman Empire and European capitalism, p. 48; based on Imlah, Economic elements, pp. 20815.



181519 and 18559, the Ottoman terms of trade rose almost 2.6 times, for anannual rate of 2.4 per cent; even over the longer 1800 to 1860 period, whichincludes two decades of Napoleonic conflict and suppressed trade, it still rose bya huge annual rate of 1.6 per cent per annum. While no actual Ottoman PM timeseries data have been tabulated prior to 1854 to confirm these otherwise plausiblePX/PM trends, qualitative evidence appears to be consistent with the implied termsof trade surge. Hence, domestic resources were pulled in to agriculture and otherprimary product sectors, while the glut of cheap, foreign manufactures flowed intothe Empire, pushing domestic resources out of import-competing industry andperhaps even out of some non-tradable activities.

Ottoman terms of trade data are of far better quality for the period 18541913.Between 18559 and 18759, the terms of trade fell by 27 per cent, while it did notfall at all between 18759 and 18938.60 Indeed, the surprising fact is how stablewas the terms of trade over the four decades between 18759 and 190913 whenit drifted up at a modest annual rate of 0.4 per cent. Furthermore, all of thepre-FirstWorldWar rise took place after 1896 when the Ottomans enjoyed a 14 percent increase in their terms of trade.

60 That is, all of the fall took place in a few years before 1875 (fig. 1).

Table 2. Wage and price trends for the Ottoman Empire, 18001913 (1860 = 100)1800 1860 1880 1913

Nominal wages 54 100 103 156Consumer prices 50 100 86 106Real wages 108 100 120 147Imported textile prices, c.i.f. 380 100 76 63Imported cotton textile prices, c.i.f. 440 100 76 63Wheat export prices, f.o.b. 103 100 86 90Other export foodstuff prices, f.o.b. 86 100 92 84All export foodstuff prices, f.o.b. 93 100 89 87Raw material export prices, f.o.b. 80 100 63 54All export prices, f.o.b. 86 100 65 55Own wage in textiles 14 100 136 248Own wage in exports 63 100 158 284Price all exports/price all mports 39 100 83 89

Notes: All prices and wages are given in both silver (and gold after 1844) and in British pounds as the exchange rate between theOttoman kurus (piaster) and the British pound was consistent in all the benchmark years with the silver and gold content of thetwo currencies. The own wage is the nominal wage divided by the price of output in the given sector; for example, textiles orexports. C.i.f. = cost, insurance, freight. F.o.b. = free on board.Sources: Wages (of unskilled construction workers) and consumer prices are for Istanbul, and are taken from Pamuk, 500 years,tab. 4.1, pp. 724. Prices of Ottoman imports and exports in and after 1860 are taken from idem, Ottoman Empire and Europeancapitalism, pp. 16875.

For the period before 1860, prices of imported textiles are taken from Imlah, Economic elements, pp. 947. The import priceestimates assume that, in this early period, the costs of transportation and insurance declined at least as fast as the prices of f.o.b.textiles so that the decline in the c.i.f. prices of imported textiles in Ottoman ports was at least as rapid as the f.o.b. prices inEngland. Freight rates actually declined faster than prices of manufactures during this period, based on the North indices citedabove (n. 83).

Prices of Ottoman commodity exports for the period before 1860 are based on the indices and sources discussed in appendix I.The price indices for various categories of Ottoman exports were then calculated both with and without weights, where the tradeweights were guesstimates since these do not exist for this period. Since Ottoman exports were quite diversified throughout thenineteenth century, the two alternatives gave very similar results. British export price indices were then used as a proxy forOttoman import prices, thus yielding a terms of trade series extending from 1800 to 1860; Mitchell and Deane, Abstract, p. 331,based on Imlah, Economic elements, pp. 947. This series was then linked to Pamuks Fisher index 18541913 series, using anarithmetic average of the seven overlapping years.



To summarize, if one is looking for evidence of a big de-industrialization impacton the Ottoman Empire, we should see plenty of it in the half-century before thelate 1850s.We should see little or no evidence of de-industrialization between thelate 1850s and the mid-1890s, assuming that the pre-1860 price shock had hadenough time to shake out import-competing industries. Finally, we should see onlyvery modest evidence of de-industrialization in the two decades before the FirstWorld War.

Were these terms of trade trends common to all parts of the Empire? Issawiclaims that the terms of trade moved in favour of the Ottomans in both Syriaand Iraq between 1836 and 1913. He bases his claims concerning Syria on theprices of raw silk and cocoons prevailing in Beirut from 1836 to 1913 and onprices for some other commodities exported from Aleppo between 1891 and1913. His claims for Iraq are based on an unweighted five-commodity indexcovering the period from 1864 to 1913. Issawi extrapolates both the Syrian andIraqi terms of trade backwards, concluding that the terms of trade improvedthere from 1800 to 1913.61 These estimates are consistent with thede-industrialization experience in that part of the Empire: it destroyed a largepart of the handicrafts both directly through competition and indirectly byturning consumers taste to western-type goods, without causing technologyspillovers or import substitution.62

According to cotton export price series, Egypts terms of trade must also haverisen across the nineteenth century. Issawi estimated that Egypts terms of trade(1880 = 100) rose from 52 in 18202 to 71 in 18502, a rate of 1 per cent perannum over the three decades, much smaller than what we have found forthe rest of the Empire (2.4 per cent per annum up to 18559). It rises again to107 in 18702, reflecting world cotton scarcity induced by the US civil war.Issawis index falls to 100 in 18902 and then rises modestly to 136 in190812.63

In figure 1, we offer our own index of the terms of trade for Egypt,17961913; the Ottoman Empire, 18001913; and the Levant, 18391913. Thespectacular rise in the Egyptian terms of trade between 18204 and 185660,2.7 per cent per annum, is even bigger than that of the Ottoman Empire overthe same period, 2.3 per cent per annum. Since PM is the same in both cases,it is clear that cotton prices rose even more in world markets than did theOttoman export mix of wheat, wool, fruits, silk, tobacco, and opium. In anycase, these terms of trade figures imply an even greater de-industrializationimpact on Egypt than on the rest of the Middle East, spelling especially bad luckfor Alis early experiments with Egyptian industrialization. Between 18559 and187579, the Egyptian terms of trade fell by almost 11 per cent, less than halfthe fall experienced by the Empire, 27 per cent, and thus, presumably, lessstimulation to import-competing industry. Finally, between 18759 and190913, the Egyptian terms of trade drifted upwards at roughly the samemodest rate as elsewhere in the Middle East (0.5 versus 0.4 per cent perannum).

61 Issawi, Middle East and North Africa, pp. 14751.62 Ibid., pp. 1513.63 Issawi, Middle East and North Africa, p. 39



VWe know that a booming terms of trade contributed to Ottomande-industrialization in the nineteenth century, and especially up to the late 1850s.That is, as export prices rose and import prices fell, labour and other resourceswere pulled out of industry (and non-tradable sectors) and into the export sectorso as to augment its capacity. The size of these so-called Dutch disease effectswere, of course, intensified by the pro-global policies introduced between 1826and 1838, but we think there were other domestic supply-side forces that mighthave diminished Ottoman competitiveness with foreign manufactures up to mid-century. We also think these forces might have reversed when the terms of tradeceased to rise after the late 1850s.

During the century before the 1850s, most foodstuffs eaten by village peasantsand the urban working class (like barley, rye, lentils, chick peas, beans, olive oil,cheap raisins, and figs) were not traded internationally. True, skilled artisans andbetter-off peasants ate wheat bread made from traded wheat, but most ate cheaperbread made from coarser grains that were not traded. Also, foodstuffs were a verylarge share of family budgets, somewhere between 75 and 80 per cent in the urbanareas and even higher in the rural areas where 75 per cent of the population lived.64

Under those conditions, labour productivity in the non-traded part of food produc-tion must have influenced manufacturing competitiveness, as Alexander Gerschen-kron, W. Arthur Lewis, and even Adam Smith argued long ago.65 Their reasoningwent as follows. In a pre-industrial economy with relatively stable subsistencewages,66 any decline in Ottoman food productivity would have put upward pressureon food prices and thus on the nominal wage in manufacturing, eroding competi-tiveness with foreign producers. Any rise in Ottoman food productivity or increasein arable land in the interior would have had the opposite effect.

However, there was another force at work too. As the Ottoman Empire becamemore integrated into world commodity markets, increased specialization took theform not only of rising exports of wool, silk, and opium, but also of consumergoods such as wheat, figs, raisins, olive oil, tobacco, and even barley. Any rise in theprice of exported consumer goods would have put more upward pressure on theprices of local consumer goods and thus on the nominal wage, eroding competi-tiveness with foreign producers in import-competing sectors. This would havebeen manifested by rising food prices relative to other products, by falling profit-ability in manufacturing, and by a decline in industrial output.

Which of these domestic supply-side forces dominated the Ottoman Empire,especially before the 1860s? Alternatively, was this force offset by the increase inarable land in the interior, thus lowering the price of foodstuffs, at least in theinterior, and especially in a pre-railroad era? What was the net effect of all thesecomplex forces on the relative price of food and other key consumer goods?

Concern with the price of foodstuffs was used by Lewis to help explainde-industrialization in the tropical periphery.67 However, he did not offer anexplicit model or supply comprehensive empirical support for his thesis. Recently,

64 Pamuk, 500 years, p. 28.65 Gerschenkron, Economic backwardness; Lewis, Evolution.66 Lewis, Economic development.67 Lewis, Evolution.



an explicit and testable Lewis-like model was applied elsewhere in the poorperiphery: the thesis worked well in helping account for the spectacular demise ofIndian manufacturing in the face of British competition after 1750;68 and it alsoworked well in helping account for exceptional Mexican success in minimizing thedamage inflicted by foreign imports on its domestic textile industry.69 There, therelationship between terms of trade booms and de-industrialization was madeexplicit. Here, it should suffice to offer only the bottom line: the industrialemployment share should fall whenever the own wage in manufacturing (thenominal wage deflated by the price of manufactures) grows faster than the ownwage in wheat and other commodity exports. Moreover, de-industrialization willbe most severe when the difference in own wage growth rates is largest, and thiscan only result when the external terms of trade rises since the nominal wageshould be similar everywhere in the domestic economy. In short, the relative sizeof industry should have undergone its most severe fall when nominal wage growthwas strongest and when the terms of trade was shifting most dramatically in favourof wheat and other commodity exports.

True, conditions were a little different in the Ottoman Empire since it was amajor exporter of foodstuffs, in contrast with India and Mexico. So, how didde-industrialization economics work in the Ottoman Empire?

VI

It is certainly comforting that the evidence in table 2 is so consistent with thepredictions of the previous section. First, the economics there invoked the Lewisassumption of stable real wages, and they were indeed, especially over the first 60years of the nineteenth century: they rose by only 11 per cent between 1800 and1880, and they fell by 8 per cent up to 1860. Second, section V predicted that arapid rise in the own wage in manufacturing (here proxied by textiles), especiallycompared with the commodity export sector, would generate powerfulde-industrialization forces. Indeed, the own wage in textiles soared across thenineteenth century. Furthermore, the biggest rise took place between 1800 and18603.3 per cent per annumprecisely the decades of most dramaticde-industrialization. The own wage in the export sector also rose up to 1860, butby not nearly as much as textiles (0.8 versus 3.3 per cent per annum). Third, therise in the own wage in manufacturing between 1800 and 1860 was being pushedpartly by a nominal wage boom (1 per cent per annum)accounting for a thirdof the own wage increase, but mostly by the collapse in manufactures prices(-2.2 per cent per annum: proxied by imported textiles)accounting for two-thirds of the own wage increase. Recall that section V predicted a nominal wageboom. Fourth, the predicted rise in food prices seems to have taken place, as theconsumer price index (CPI) (dominated by foodstuffs) doubled between 1800and 1860. The rising CPI pushed up the nominal wage by almost the sameamountconfirming the predictions of Gerschenkron, Lewis, Smith, and theeconomics of section V, thereby diminishing Ottoman wage competitiveness inmanufacturing.

68 Clingingsmith and Williamson, De-industrialization.69 Dobado Gonzlez et al., Mexican exceptionalism.



Fifth, the rise in the own wage in manufacturing slowed down considerablybetween 1860 and 1880 to 1.5 per cent per annumprecisely the decades of adiminished rate of de-industrialization and some signs of re-industrialization.Consistent with that experience, the own wage in the export sector actually grew alittle faster than it did in textiles (2.3 versus 1.5 per cent per annum). Furthermore,during these two decades the modest rise in the nominal wage (0.1 per cent perannum) accounts for only 7 per cent of the own wage increase in manufacturing,while the fall in manufacturing prices (-1.4 per cent per annum) accounts for 93 percent.This small rise in the nominal wage is consistent with the 14 per cent fall in theCPI, the latter induced, presumably, by the 11 per cent fall in the food export price.Sixth, and finally, Lewis-like thinking loses its relevance after 1880 when the realwage rose quite impressively (0.8 per cent per annum), presumably because the rateof productivity advance in both the export and import competing sectors rose.

Table 3 reports our effort to identify the domestic sources of Ottomande-industrialization which supported external terms of trade events. The table

Table 3. The impact of Ottoman policy and railroads on PX/PM,18001913:% impacton price

Coastal Interior

Imports Exports Imports Exports

Policy and transport cost conditions before free trade treaties, 183841Import tariff or export tax 5 10 5 10Tax on interior trade 0 0 8 8Transport cost 0 0 25 43

Policy and transport cost conditions from free trade treaties to 1860Import tariff or export tax 5 12 5 12Tax on interior trade 0 0 0 8Transport cost 0 0 25 43

Policy and transport cost conditions, 18619Import tariff or export tax 8 1 8 1Tax on interior trade 0 0 0 8Transport cost 0 0 25 43



Changes in 183841: assuming no change in transport cost (pre-railways)Total PM or PX 0 +2 -8 -2PX/PM +2 +6

Changes in 1860: assuming no change in transport cost (pre-railways)Total PM or PX +3 +11 +3 +11PX/PM +8 +8

Changes in 1870: including change in transport cost (railways)Transport cost 0 0 -7 +15Total PM or PX 0 0 -7 +23PX/PM 0 +30

Changes in 1907: assuming no change in transport cost (post-railways)Total PM or PX +3 0 +3 0PX/PM -3 -3

Notes: See text.



focuses on changes in Ottoman policy and the appearance of the railroads pen-etrating the interior in the late nineteenth century.70 These two changing domesticforces must be assessed against the dramatic 260 per cent increase in the externalterms of trade between 1800 and 1860 (table 2).

Consider first the policy impact of the Ottoman free trade treaties of 183841.We estimate that the initial move towards free trade raised PX/PM in the Ottomanports and coastal cities by only 2 per cent, a trivial impact compared with the260 per cent rise in PX/PM coming from world market forces. True, the policyimpact was bigger in the interior (a 6 per cent rise in PX/PM there), but still verysmall. It should be added, of course, that the treaties took away tariff autonomyand severely reduced the ability of the Ottoman government to raise tariffs later inthe century, making it impossible to protect domestic industry from the full forceof foreign competition when the free trade damage became clearer. Ottoman tariffswere changed, albeit marginally, after negotiations with European countries whilethe free trade treaties were being renewed in 1861. Duties on exports were reducedfrom 12 per cent to 1 per cent and tariffs on imports were raised from 5 per centto 8 per cent. These changes improved the terms of trade faced by the domesticproducers by 8 per cent in both the coastal areas and in the interior. The elimi-nation of internal tariffs in 1870 improved the terms of trade in the interior by anadditional 8 per cent. These changes were, however, offset by world marketconditions: Ottoman PX/PM fell by 17 per cent in world markets between 1860 and1880 (table 2) which must have eased the de-industrialization pressures on localmanufacturing. After 1880, neither the external terms of trade nor Ottoman policychanged much, so de-industrialization forces lost their destructive impact oncoastal industry. Things were different in the interior, however, as the railroadsopened up those markets to import penetration and exports to world markets.Table 3 estimates that the terms of trade in the interior rose by 30 per cent, asdeclining transport costs and the removal of the interior trade tax served to pushexportable (for example, wheat) prices up to world levels and to push importable(for example, textiles) prices down to world levels.71 So, while de-industrializationforces were quiet on the Ottoman coast after 1860, they certainly were still presentin the Ottoman interior.

70 The literature often fails to appreciate that it is changes in policy, transport costs, and world prices that matter,not levels. After all, de-industrialization implies a change, such as a decline in the relative or even absoluteimportance of manufacturing.

71 The share of transportation costs in the price of a good moved to and from the Ottoman interior variedsubstantially according to the good (low-value, high-bulk primary products versus high-value, low-bulk textiles,but also wheat versus other primary goods) and the distance from port. Hence, the estimates underlying tab. 3are rough. Quataert, Limited revolution, pp. 14354, offers an excellent discussion of this issue and someinternational comparisons with wheat transported on the Anatolian Railway after 1890.The cereal-growing areasin Anatolia could not send any wheat to Istanbul or Izmir, the leading ports in western Anatolia, before therailroad, but they did send some to the southern port of Mersin on the Mediterranean via camel caravans.Quataert indicates that per ton-kilometre costs in transporting Anatolian wheat to these ports by railroad wereinitially significantly higher than on the routes between Chicago and New York. However, he estimates thatton-kilometre charges for wheat and barley dropped across the late nineteenth century, so that they accounted for22% of the final price in Istanbul at the end of the period. Most other export commodities produced in theinterior, such as fruits, nuts, and tobacco, benefited to a smaller extent from the arrival of the railroads; Kurmus,Role of British capital. Tab. 3 summarizes this evidence by assuming that between the 1860s and 1913 therailroads by themselves lowered textile prices in the interior by about 7% and raised primary product prices in theinterior by about 15%, all relative to coastal prices.



VII

The economic impact of the core on the periphery had its source in two forces thatarose during the first global century.72 The first was the worldwide transportrevolution that served to integrate world commodity markets.73 It caused a boomin trade between core and periphery, created commodity price convergence fortradable goods between all world markets, and contributed to a rise in everycountrys external terms of trade, including the periphery.The second force camefrom the derived demand for industrial intermediates, like Ottoman raw silk andwool, which soared as manufacturing production led the way in the core.Thus, ascore economies raised their industrial output shares, manufacturing output growthraced ahead of GDP growth. Rapid productivity growth lowered the cost and priceof manufactures, and by so doing generated a soaring derived demand for rawmaterials in the core.This event was reinforced by accelerating income per capitagrowth and a high income elasticity of demand for luxury consumption goods, likeOttoman wheat, raisins, figs, and even opium. Since industrialization was driven byunbalanced productivity advance favouring manufacturing relative to agricultureand other natural-resource-based activities, the relative price of manufactures felleverywhere, especially in the periphery where they were imported. The worldtransport revolution made it possible for the distant periphery to supply thisbooming demand for primary products. Both forces produced positive, powerful,and sustained terms of trade shocks in the periphery, raising the relative price ofprimary products, and lowering the relative price of manufactures. This epochstretched over as much as 70 or 80 years.

Eventually these two forces abated. The rate of decline in real transport costsalong sea lanes slowed down, approaching a twentieth-century steady state.74 Therate of growth of manufacturing slowed down in the core as the transition toindustrial maturity was completed. As these two forces abated, the resultingslowdown in primary product demand growth was reinforced by resource-savinginnovations in the industrial core, induced, in large part, by those high and risingprimary product prices during the nineteenth-century terms of trade upswing.Thus, the secular boom faded, eventually turning into a secular bust. Exactly whenand where the boom turned to bust depended on the export commodities in whicha periphery region specialized, but the periphery peak ranged between the 1850sand the 1890s.

This 130-year cycle in the periphery terms of trade is illustrated in figure 2 byLatin American experience up to 1939 (1900 = 100). The regions terms of tradeunderwent a steady increase from the 1810s to the early 1890s, and, like theOttoman Empire, the improvement was especially dramatic during the first fourdecades: the annual rate of increase was 1.3 per cent per annum between startingthe half-decade 181519 and concluding the half-decade 18904, equivalent toalmost a tripling over the 75 years; and the rate between 181519 and 18559 waseven greater, about 2.1 per cent per annum. Furthermore, that increase is probablyunderstated since it fails to take account of the likely increase in the quality of

72 This section draws extensively on Williamson, Globalization and the great divergence.73 ORourke and Williamson, Globalization, ch. 3; Shah Mohammed and Williamson, Freight rates.74 Shah Mohammed and Williamson, Freight rates.



traded manufactures relative to primary products.75 Based on the estimates under-lying figure 2, the quality-adjusted terms of trade may have grown at a little morethan 2.2 per cent per annum between 181519 and 18559, and at a little morethan 1.4 per cent per annum between 181519 and 18904.There is no reason toexpect these quality adjustments to have been any different for the rest of the poorperiphery, including the Ottoman Empire.

How does the terms of trade boom for the Ottoman Empire and the rest of theeastern Mediterranean compare with Latin America, Asia, and the Europeanperiphery? Figure 3 documents terms of trade performance for the five majorregions in the poor periphery, between 1796 and 1913 (1900 = 100). Each regionis an 1870 population-weighted average of the following components: Europeanperiphery (Italy, Portugal, Spain, and Russia); Middle East (the Ottoman Empireand Egypt); Latin America (Argentina, Chile, Mexico, and Venezuela); south Asia(India and Ceylon); and south-east Asia (Indonesia, the Philippines, Siam, andMalaya). Figure 3 does not break out the country experience of which the fiveregions are aggregates, but the original source does.76

First and most important, the secular terms of trade boom was even bigger inEgypt and the Ottoman Empire than the rest. If we ignore the few years around1820 when the terms of trade spikes, it appears that south Asia (and India, whichdominates the regional series) underwent relatively modest improvements in itsterms of trade from 1800 to the mid-1820s, and in fact fell thereafter up to 1850.Over the half-century between 18004 and 18559, Indias terms of trade rose lessthan 0.5 per cent per annum. No doubt this was a significant secular price shock,

75 Reported in Williamson, Globalization, de-industrialization and underdevelopment, fig. 1.76 Williamson, Globalization and the great divergence.

0

20

40

60

80

100

120

140

160

1811

1815

1819

1823

1827

1831

1835

1839

1843

1847

1851

1855

1859

1863

1867

1871

1875

1879

1883

1887

1891

1895

1899

1903

1907

1911

1915

1919

1923

1927

1931

1935

1939

Px/

Pm

Average Latin America terms of trade unadjusted

Average Latin America terms of trade adjusted

Figure 2. Latin American terms of trade, 18111939 (1900 = 100)Source: Unadjusted: Clingingsmith andWilliamson, De-industrialization, fig. 9, based on data in Coatsworth andWilliamson, Alwaysprotectionist?. Adjusted: see app. I.



but it was far smaller than what happened in Egypt, the Ottoman Empire, andthe Middle East as a whole. The Latin American terms of trade increased by1.7 per cent per annum between 18204 and 18559. The Ottoman terms oftrade increased by 2.4 per cent per annum between 181519 and 18559, whilethe Indonesian terms of trade (which dominates the south-east Asia regionalexperience) increased by 2.5 per cent per annum between 18259 and 18659.The Egyptian terms of trade rose by 2.7 per cent per annum between 18204 and18559. In short, on the secular upswing in the terms of trade shared by all regionsin the poor periphery, Egypt and the Ottoman Empire underwent about thebiggest increase. This implies that the eastern Mediterranean region also sufferedsome of the highest levels of de-industrialization and Dutch disease up to the1860s.

What went up then came down, as the terms of trade fell everywhere in theperiphery from the 1870s or 1890s to the Second World War. However, it shouldbe noted that Egypt and the Ottoman Empire reached a relatively early peak intheir terms of trade, the series levelling off and even falling after the 1850s and1860s.This implies that Dutch disease de-industrialization forces lost their powermuch earlier in the region than in Latin America and other parts of the poorperiphery where their terms of trade continued to rise during the late nineteenthcentury.

0

50

100

150

200

250

1796

1802

1808

1814

1820

1826

1832

1838

1844

1850

1856

1862

1868

1874

1880

1886

1892

1898

1904

1910

Ter

ms

of t

rade

Middle East

Latin America

South-east Asia

European periphery

South Asia

Figure 3. The poor periphery: net barter terms of trade, 17961913 (1900 = 100)Source: Williamson, Globalization and the great divergence, fig. 4.



Bogazii University and London School of Economics; Harvard University andUniversity ofWisconsin

Date submitted 21 February 2009Revised version submitted 14 March 2010Accepted 21 April 2010

DOI: 10.1111/j.1468-0289.2010.00560.x

Footnote referencesAllen, R. C., The British industrial revolution in global perspective (Cambridge, 2009).Bairoch, P., European trade policy, 18151914, in P. Mathias and S. Pollard, eds., The Cambridge economic history

of Europe, vol.VIII: the industrial economies: the development of economic and social policies (Cambridge, 1989),pp. 1160.

Brtola, L. and Williamson, J. G., Globalization in Latin America before 1940, in V. Bulmer-Thomas, J. H.Coatsworth, and R. Corts Conde, eds., Cambridge economic history of Latin America, vol. II: the long twentiethcentury (Cambridge, 2006), pp. 1156.

Bezanson, A., Gray, R. D., and Hussey, M., Wholesale prices in Philadelphia 17841861 (Philadelphia, 1936).Brown, L. C., ed., Imperial legacy: the Ottoman imprint on the Balkans and the Middle East (New York, 1996).Chaudhury, S., From prosperity to decline: eighteenth century Bengal (New Delhi, 1999).Clark, E. C., The Ottoman industrial revolution, International Journal of Middle East Studies, 5 (1974),

pp. 6576.Clingingsmith, D. and Williamson, J. G., De-industrialization in 18th and 19th century India: Mughal decline,

climate shocks and British industrial ascent, Explorations in Economic History, 45 (2008), pp. 20934.Coatsworth, J. H. and Williamson, J. G., Always protectionist? Latin American tariffs from independence to great

depression, Journal of Latin American Studies, 36 (2004), pp. 20532.Dobado Gonzlez, R., Gmez Galvarriato, A., and Williamson, J. G., Mexican exceptionalism: globalization and

de-industrialization, 17501877, Journal of Economic History, 68 (2008), pp. 153.Gayer, A. D., Rostowm W. W., and Schwartz, A. J., Microfilmed supplement to volumes I and II of the growth and

fluctuation of the British economy 17901850 (Oxford, 1953).Gerschenkron, A., Economic backwardness in historical perspective: a book of essays (Cambridge, Mass., 1962).Gounaris, V., Steam over Macedonia, 18701912: socio-economic change and the railway factor (New York, 1993).Harlaftis, G. and Kardasis, V., International shipping in the eastern Mediterranean and the Black Sea: Istanbul

as a maritime centre, 18701910, in S. Pamuk and J. G. Williamson, eds., The Mediterranean response toglobalization before 1950 (2000), pp. 23365.

Hayami, A., A great transformation: social and economic change in sixteenth and seventeenth century Japan,Bonner Zeitschrift fr Japanologie, 8 (1986), pp. 313.

Helpman, E., The mystery of economic growth (Cambridge, Mass., 2004).Imlah, A. H., Economic elements in the Pax Britannica: studies in British foreign trade in the nineteenth century

(Cambridge, Mass., 1958).Issawi, C., The economic history of the Middle East, 18001914 (Chicago, 1966).Issawi, C., An economic history of the Middle East and North Africa (New York, 1982).Issawi, C., The fertile crescent 18001914. A documentary economic history (New York, 1988).Kurmus, O., The role of British capital in the economic development of Western Anatolia 18501913 (unpub.

Ph.D. thesis, Univ. of London, 1974).Kurmus, O, The 1838 Treaty of Commerce re-examined, in J.-L. B. Grammont and P. Dumont, eds., Economie

et Societies dans lEmpire Ottoman (Paris, 1983), pp. 41117.Lapavitsas, C., Industrial development and social transformation in Ottoman Macedonia, Journal of European

Economic History, 35 (2006), pp. 661710.Lewis, W. A., Economic development with unlimited supplies of labour, Manchester School of Economics and

Social Studies, 22 (1954), pp. 13991.Lewis, W. A., The evolution of the international economic order (Princeton, N.J., 1978).McCusker, J. J., How much is that in real money? A historical price index for use as a deflator of money values in the

economy of the United States (American Antiquarian Soc., Worcester, Mass., 1992).Matsuyama, K., Agricultural productivity, comparative advantage, and economic growth, Journal of Economic

Theory, 58 (1992), pp. 31734.Mendels, F. F., Proto-industrialization: the first phase of the industrialization process, Journal of Economic

History, 32 (1972), pp. 24161.Mitchell, B. R. and Deane, P., Abstract of British historical statistics (Cambridge, 1962).North, D. C., Ocean freight rates and economic development 17501913, Journal of Economic History, 18

(1958), pp. 53755.



ORourke, K. H. and Williamson, J. G., Globalization and history: the evolution of a nineteenth-century Atlanticeconomy (Cambridge, Mass., 1999).

Owen, R., The study of Middle Eastern industrial history: notes on the interrelationship between factories andsmall-scale manufacturing with special references to Lebanese silk and Egyptian sugar, 19001930, Interna-tional Journal of Middle East Studies, 16 (1984), pp. 47587.

Owen, R., Middle East in the world economy 18001914 (1993 [1st edn. 1983]).Palairet, M., The Balkan economies c. 18001914: evolution without development (Cambridge, 1997).Pamuk, S., Foreign trade and foreign capital in the Ottoman Empire, 18301913 (unpub. Ph.D. thesis, Univ. of

California, 1978).Pamuk, S., The Ottoman Empire and European capitalism, 18201913: trade, investment and production (Cambridge,

1987).Pamuk, S., 500 years of prices and wages in Istanbul and other cities (Ankara, 2001).Petmezas, S. D., Patterns of protoindustrialization in the Ottoman Empire: the case of Eastern Thessaly, ca.

17501860, Journal of European Economic History, 19 (1990), pp. 575603.Puryear, V. J., International economics and diplomacy in the Near East: a study of British commercial policy in the

Levant, 18341853 (Hamden, Conn., 1969 [first pub. Stanford, 1935]).Quataert, D., Limited revolution: the impact of the Anatolian railway on Turkish transportation and the

provisioning of Istanbul, 18901908, Business History Review, 51 (1977), pp. 13960.Quataert, D., Manufacturing and technology transfer in the Ottoman Empire, 18001914 (Istanbul, 1992).Quataert, D., Ottoman manufacturing in the age of the industrial revolution (Cambridge, 1993).Quataert, D., The age of reforms, 18121914, in H. Inalcik and D. Quataert, eds., An economic and social history

of the Ottoman Empire, 13001914 (Cambridge, 1994), pp. 759943.Shah Mohammed, S. and Williamson, J. G., Freight rates and productivity gains in British tramp shipping

18691950, Explorations in Economic History, 41 (2004), pp. 172203.Simon, M., The United States balance of payments, 18611900, in Trends in the American economy in the

nineteenth century (Princeton, N.J., 1960), pp. 629716.Taylor, B., Global financial data: encyclopedia of global financial markets (Commercial database, 2003) [WWW

document]. URL http://lib.harvard.edu, under E Resources [accessed on 13 August 2009].Todorov, N., The Balkan city, 14001900 (Seattle, 1983).de Vries, J., The industrial revolution and the industrious revolution, Journal of Economic History, 54 (1994),

pp. 24970.de Vries, J., The industrious revolution: consumer behavior and the household economy 1650 to the present (Cambridge,

2008).Weisdorf, J. L., From domestic manufacturing to industrial revolution: long-run growth and agricultural

development, Oxford Economic Papers, 58 (2006), pp. 26487.Williamson, J. G., Explaining world tariffs, 18701938: Stolper-Samuelson, strategic tariffs, and state revenues,

in R. Findlay, R. Henriksson, H. Lindgren, and M. Lundahl, eds., Eli Heckscher, international trade and economichistory (Cambridge, Mass., 2006), pp. 199228.

Williamson, J. G., Globalization, de-industrialization and underdevelopment in the Third World before themodern era, Journal of Iberian and Latin American Economic History, 24 (2006), pp. 936.

Williamson, J. G., Globalization and the poor periphery before 1950 (Cambridge, Mass., 2006).Williamson, J. G., Globalization and the great divergence: terms of trade booms, volatility and the poor periphery

17821913, European Review of Economic History, 12 (2008), pp. 35591.Williamson, J. G., Trade and poverty: when the ThirdWorld fell behind (Cambridge, Mass., forthcoming).Yousef, T. M., Egyptian commodity markets in the age of economic liberalism, in S. Pamuk and J. G.

Williamson, eds., The Mediterranean response to globalization before 1950 (2000), pp. 34059.

Official publicationUS Department of Commerce, Bureau of the Census, Historical Statistics of the United States, Colonial Times to

1970, pt. 1 (Washington D.C., 1975).

APPENDIX I: OTTOMAN TERMS OF TRADE, 18001913

In order to extend the Ottoman net barter terms of trade back before 1854 and tolink it to the beginning of Pamuks 18541913 series,77 we constructed priceindices that made use of price series for six of the leading Ottoman exports takenfrom the following sources; wheat in Istanbul from Pamuk;78 opium in India from

77 Pamuk, Ottoman Empire and European capitalism, pp. 4154, 16875.78 Pamuk, 500 years, pp. 1307, 1703.



Clingingsmith and Williamson;79 tobacco and raisins in Philadelphia from Bezan-son et al.;80 and silk and wool from Gayer et al.81 All current prices were convertedto British pounds using the nominal Ottoman kurus/British pound82 and USdollar/British pound exchange rates.83 The prices of raisins, silk, and wool cited inthese sources are c.i.f., so transportation and insurance costs needed to be sub-tracted from them in order to reach f.o.b. prices in Ottoman ports. For thispurpose, we relied on the methodology and calculations undertaken by Pamuk.84

Those calculations used North for freight rates indices and Simon for insurancerates.85 We extended Simons 18611900 insurance rate trends back to 1800assuming that insurance rates moved similar to Norths freight rates but did notdecline as fast in the period before 1860.We remained on the conservative side andassumed that the difference between the c.i.f and f.o.b prices did not fall as fast asthe North freight rate indices suggest. Trends in the f.o.b. series for raisins, silk,and wool obtained through these calculations were not very different from trendsin the f.o.b. series for wheat, opium, and tobacco as can be seen in table 2. Theprice indices for Ottoman exports were then calculated both with and withoutweights, where the trade weights were guesstimates since these do not exist for thisperiod. Since Ottoman exports were quite diversified throughout the nineteenthcenturythe share of any one commodity in the total value of exports rarelyexceeding 10 per centthe two alternatives gave very similar results.86

British export price indices were used as a proxy for Ottoman import prices,thus yielding a terms of trade series extending from 1800 to 1860.87 This series wasthen linked to Pamuks Fisher index 18541913 series, using an arithmetic averageof the seven overlapping years.

APPENDIX II: EGYPTIAN TERMS OF TRADE, 17961913

Egypts terms of trade 18201913 has been constructed using only cotton pricesto estimate export price trends, a reasonable assumption since cotton was Egyptsdominant nineteenth-century export commodity, accounting for one-third ofexports in the 1840s50s, over 80 per cent in the 1880s and over 90 per cent in191014.88 For the years 182099, Alexandrian cotton prices were used.89 For theremaining 14 years, US cotton prices were taken as a proxy for Egyptian prices.90

This US cotton price proxy was chosen because Egyptian cotton followed closelyworld, and more particularly American, prices.91

The series reported by Issawi gives the cotton price in dollars per qantar, so hiswas converted to dollars per pound in order to make it consistent with the US

79 Clingingsmith and Williamson, De-industrialization, p. 24.80 Bezanson, Gray, and Hussey, Wholesale prices.81 Gayer, Rostow, and Schwartz, Microfilmed supplement.82 Pamuk, 500 years, pp. 1718.83 Taylor, Global financial data.84 Pamuk, Foreign trade, app. I, pp. 18799.85 North, Ocean freight; Simon, United States.86 Pamuk, Ottoman Empire and European capitalism, p. 150.87 Mitchell and Deane, Abstract, p. 331, based on Imlah, Economic elements, pp. 947.88 Issawi, Middle East and North Africa, p. 31.89 Issawi, Middle East, 18001914, pp. 4478.90 US Department of Commerce, Historical Statistics, p. 208.91 Issawi, Middle East and North Africa, p. 41.



series. In Egypt, the qantar was equal to 120.6135 pounds until 1835.Thus, for theperiod 18201935, the reported figures were divided by 120.6135 in order toobtain $/lb. After 1835, Egyptian qantars were equal to 99 pounds, so the reportedfigures were divided by that number to get $/lb.92

To obtain the real export price series, the resulting nominal price series wasdivided by the US consumer price index for each of the years 18201913 and thenindexed at 1880 = 100.93 In order to derive the terms of trade series, the nominalcotton price series was divided by the price of British exports (a proxy for the priceof Egyptian imports) for each year of the period.94

Although cotton was always Egypts dominant export commodity, it increased inimportance throughout the century. Egypts other important exportable, wheat,held substantial export shares earlier in the century.95 In order to see if the cottonseries accurately captures a more comprehensive export price index, the price ofwheat was correlated against the price of Egyptian cotton.96 The correlation is veryhigh, indicating that Egypts cotton terms of trade series is likely to be

pamuk & williamson, ottoman deindustrialization, 1800,1913

Documents