An annotated bibliography of at least five additional resources related to the research topic.No plagiarism It must be less than 30% match on authenticity report.***PLEASE DO NOT ACCEPT THIS ASS

Early globalizations: The integration of Asia in the world economy, 1800–1938 David Chilosi a,⁎, Giovanni Federico b aEconomic History Department, London School of Economics, Houghton Street, London WC2A 2AE, UK bDepartment of Economics, University of Pisa, Via Ridolfi 10, 56124 Pisa, Italy Received 14 April 2014 Available online 23 April 2015 Abstract This paper contributes to the debate on globalization and the great divergence with a comprehensive analysis of the integration of Asia in the world market from 1800 to the eve of World War II. We examine the patterns of convergence in prices for a wide range of commodities between Europe and the main Asian countries (India, Indonesia, Japan and China) and we compare them with convergence between Europe and the East Coast of the United States, hitherto the yardstick for the 19th century. Most price convergence occurred before 1870, mainly as a consequence of the abolition of the European trading monopolies with Asia, and, to a lesser extent, the repeal of duties on Atlantic trade. After 1870, price differentials continued to decline thanks to falling freights and to better communication after the lay-out of telegraph cables. There was only little disintegration in the inter-war years.

JEL classification:F14; F15; F63; O40; N74; N75 Keywords:Globalization; Market integration; International trade; Economic growth; Asia; Nineteenth century 1. Introduction Standard economic theory holds that trade and market integration foster economic growth. Indeed, the era of the so-called first globalization, before World War I, coincid- ed with a period of unprecedented economic growth in Europe and in its Western Offshoots (Maddison Project, 2013). Yet, at the same time, the Asian countries (with the partial exception of Japan) fell increasingly behind the advanced European ones (Broadberry, 2013), in spite of rapidly growing exports (Federico and Tena, 2013). Some scholars have tackled the paradox posed by this“greatdivergence”(Pomeranz, 2000) by pointing out that exports of primary products did benefit the Asian economies, but their effect was too small to foster economy-wide growth (Feuerwerker, 1980, 1983; Booth, 1988; Tomlinson, 1993; van der Eng, 1996; Roy, 2000; Brandt et al., 2013). Others blame the colonial powers for forcing the Asian economies to export primary products, thus damaging their growth potential (Dutt, 1969; Parthasarathi, 2011). ForWilliamson (2008, 2011, 2012, 2013), too, specialization in primary products damaged the long-term prospects for industrial- ization in the periphery.In his view, however, this specialization wasthe unintended consequence of market ⁎ Corresponding author.

E-mail address:[email protected](D. Chilosi).

http://dx.doi.org/10.1016/j.eeh.2015.04.001 0014-4983/© 2015 Elsevier Inc. All rights reserved. Available online at www.sciencedirect.com ScienceDirect Explorations in Economic History 57 (2015) 1–18 www.elsevier.com/locate/eeh integration, which improved the terms of trade before the 1870s. In the same vein,Allen (2011)argues that peripheral countries could have escaped this“curse of primary products”(Sachs and Warner, 2001)onlyby adopting a coherent industrialization policy, which was conspicuously lacking in all Asian countries but Japan.

Testing these competing views about exports and economic growth entails a huge and very challenging research agenda. This paper contributes to this agenda by exploring price convergence, an essential component of integration, between Europe (United Kingdom, the Netherlands, or France) and the four main Asian countries, China, British India, the Dutch East Indies (henceforth Indonesia) and Japan from the beginning of the 19th century to the eve of World War II. In so doing, we fill in two key gaps in the literature on the integration of Asia in the world economy: we analyze the period 1800–1870 and quantify the impact of the abolition of Western trading monopolies. Previous work has shown that price gaps were high before 1800 (O'Rourke and Williamson, 2002; Rönnbäck, 2009), narrowed after 1870 (O'Rourke and Williamson, 2002; Hynes et al., 2012), and widened during the Great Depression (Hynes et al., 2012). However, no empirical research, to date, has dealt with the period 1800– 1870. Yet, these years featured massive processes of integration both within Europe (Federico, 2011, 2012)and in the Atlantic economy (Jacks, 2005; Uebele, 2011; Sharp and Weisdorf, 2013), raising the question of how does Asia compare? The same years also saw the abolition of the monopolies by the Western companies trading with Asia.

Their demise must have boosted integration, but we lack measures of the actual size of this effect, relative to those of the decline in duties and advancements in transport and communication technology.

We present our dataset inSection 2and discuss the patterns of price convergence inSection 3. The key period of integration across routes was the early, rather than the late nineteenth century, while price differentials remained roughly stable in interwar years.Section 4deals with the main barriers to trade, focusing on the trading companies and on their abolition, whileSection 5 estimates the contribution of different causes (institution- al change, fall in transport costs, trade liberalization and so on) to price convergence with a panel regression. In spite of the similarities in trends across oceans, the processes of integration had roots in different institutions:

while in the Atlantic economy the repeal of duties was a key determinant, much of the price convergence between Asia and Europe was due to the demise of the British East India Company (EIC) and, to a lesser extent, the Dutch trading monopoly (Nederlandsche Handel-Maatschappij or NHM).Section 6concludes.2. The data-base The quantitative analysis of integration faces a trade-off between the quality of the data and their representativeness of changes in the overall market. In particular, asFederico (2012)argues, in order to produce reliable results, the price series should meet three conditions:

i) The price ratios should refer to pairs of markets which were actually trading. Otherwise, price differentials can be lower than costs and move (quasi-) randomly within the band of commodity points. If markets trade and are efficient à laFama (1970), in equilibrium price gaps must be equal to transaction costs, inclusive of monopoly mark-ups.

ii) Each price series should refer to a specific quality rather than to the market average and each pair of series should refer to the same quality. Otherwise, price gaps might reflect quality differentials, and any change in quality in a market might introduce spurious trends.

iii) The commodities should be representative of the actual trade flows. Extending inferences from one product only (e.g. cereals), is tantamount to assume that that movements in transport costs, barriers to trade and market efficiency are similar across all traded goods.

Unfortunately, we cannot examine integration on the import side because the data on prices of manufactures are very scattered and refer to different qualities. The data for primary products are much more abundant and thus we have been able to collect series of prices for the same commodities in 26 pairs of markets (Table 1). 1 All cities in our sample were major trading centers in their own countries, and trade statistics report bilateral trade of that specific good for about 93% of the observations. Missing data are mostly scattered, which suggests failure to record rather than absence of trade.

There is a small chance that absence of trade could be an issue in only about 2% of the cases. Quality is homogeneous across markets (Yes in the relevant column) in the overwhelming majority of pairs, 22 out of 29 pairs. In two of the other cases, one series only can be considered as qualitatively homogeneous (Yes in the Column“within market”), but the quality surely differs between series (No in the Column“across markets”), 1For a more detailed discussion of our sources seeChilosi and Federico (2013, Appendix B). We have considered a larger sample, but we have decided to drop some series (e.g. tobacco from Indonesia) which did not meet the minimum quality threshold. 2D. Chilosi, G. Federico / Explorations in Economic History 57 (2015) 1–18 while in the remaining ones the quality differs between markets and changes in time in each market (No in all three columns). In short, in the worst case, homogeneity might be a problem at most for one fourth of cases. Last but not least, these products accounted for about half or more of exports from India and Indonesia nearly the whole period, from United States before the Civil War and from China until 1900 (Table 2). The coverage is poor only for Japan.

Summing up, the data-base, though falling short of being perfect, can be considered as fit for our purposes.3. The process of integration: A statistical analysis A visual inspection of some representative series of price ratios (Fig. 1) reveals that, on the one hand, as expected, the price ratios were fairly high at the beginning of the 19th century and fell in the long run, with a spike during World War I. On the other hand, the differentials differedhugelyamongcommodities at the same time (as seen in the vertical axis) and they often suddenly collapsed rather than steadily declined.

Table 1 The data-base.

Period Market Homogeneous quality?

Within markets Across Origin Destination Origin Destination Markets Atlantic Cotton 1801–1938 New York Liverpool Yes Yes Yes Wheat 1800–1937 New York London No No No Far East Silk 1834–1877 Canton London Yes Yes Yes Silk 1874–1913 China London Yes Yes Yes Silk 1874–1914 China Lyon Yes Yes Yes Silk 1894–1914 Yokohama Lyon No No No Silk 1894–1938 Yokohama New York No No No Tea 1811–1831 Canton England No No Yes a Tea 1820–1877 Canton London Yes Yes Yes India Cotton 1796–1845 Calcutta London Yes Yes Yes Cotton 1867–1938 Bombay London Yes Yes Yes Indigo 1822–1931 Calcutta London Yes Yes Yes Jute 1844–1938 Calcutta London Yes Yes Yes Linseed 1846–1938 Calcutta London Yes Yes Yes Rapeseed 1871–1921 Calcutta London Yes Yes Yes Rice 1870–1938 Rangoon London Yes Yes Yes Saltpetre 1796–1853 Calcutta London Yes Yes Yes Silk 1796–1856 Calcutta London Yes Yes Yes Silk 1857–1877 Calcutta Lyon Yes Yes Yes Sugar 1796–1856 Calcutta London Yes Yes Yes Tea 1893–1931 Calcutta London Yes Yes Yes Wheat 1861–1931 Calcutta London Yes No No Indonesia Coffee 1833–1913 Batavia Rotterdam No Yes No Pepper 1828–1938 Batavia Amsterdam No No No Rice 1848–1913 Batavia Amsterdam Yes Yes Yes Rubber 1913–1938 Batavia London Yes Yes Yes Sugar 1822–1938 Batavia London No No No Tea 1893–1938 Batavia Amsterdam Yes Yes Yes Tin 1863–1913 Batavia Amsterdam Yes No No Sources: seeChilosi and Federico (2013: Appendix B).

aSince all theseﬁgures refer to tea imported and sold at auctions by the East India Company, we expect the quality mix to be similar in the same year, but not necessarily across years.3 D. Chilosi, G. Federico / Explorations in Economic History 57 (2015) 1–18 More precise information on convergence can be obtained by estimating the equation (Razzaque et al., 2007):

ΔLn RP i¼αþβTIMEþψln RP i t 1 þφlnΔLn RP i t 1 þuð1Þ whereRP iis the relative price of theith good between two markets andTIMEis a linear trend. We compute the long-run rate of change ast= (β/ψ). The error correction model coefficientψ(ranging between 1 and 0) tests whether and estimates how rapidly price ratios return to this trend after a shock, while the lagged change in relative prices is added to address possible serial correlation. We first run Eq.(1)as a fixed effects panel for the whole sample and separately by country, considering jointly the (comparatively few) observa- tions for the two Far Eastern countries, China and Japan.

The results (Table 3) confirm the conventional wisdom: at the beginning of the series (which differed by product and route), European prices were on average double the Asian ones and by the end this difference had been cut by a half. All rates are statistically significant and very similar across routes. The half-lives of shocks are comparatively quite high. One might sum up that the market was becoming increasingly integrated, but overall it had still a long way before becoming really efficient.

However, such a conclusion might be a tad hasty: it assumes stationary efficiency and linear price conver- gence with equal rates by product. We explore difference in timing of integration by running separate panel regressions for five periods: the twilight of mercantilism (1796–1815), the early globalization (1815–1870), the heyday of globalization (1870– 1913) and the war and interwar (1914–1938). We alsorun for these periods a pooled group estimator, which allows rate of change to differ across products and then averages them (Table 4).

As posited by the conventional wisdom, the results show no integration before 1815, although they rely on a small and thus potentially non-representative number of observations. The data for the two subsequent periods are undoubtedly representative and they yield a clear conclusion: overall, convergence was twice faster in the“early globalization”than during its (alleged)“heyday”. The difference between the two periods is particularly wide for the Far East, while convergence between Indonesia and Europe was only 45% faster in 1815–1870 than in 1870–1913. Integra- tion of Indonesia during the“early globalization”is also the only case where the two estimators yield different results. The negative coefficient(s) for the war and interwar period reflects the return to normal levels of price differentials, after the sharp war-time increase.

Indeed, dropping the first 5 years, the rates become positive for the total and for all areas but the Far East.

Yet the rates of inter-war disintegration remain rela- tively small and are not statistically significant. 2In other words, the inter-war disintegration of the world trading network affected little the Asian and American exports of primary products analyzed here.

The increase in the speed of reaction between the second and third periods suggests an increase in market efficiency. However, this is modest and the reaction remains rather slow: one may surmise that in modern markets, most shocks were arbitraged away within the year and that, consequently, our yearly series capture only very large shocks, which needed more time to be absorbed. This conjecture should be tested with higher frequency data (Brunt and Cannon, 2014).

One might suspect the inter-temporal comparisons to be biased by composition effects as the coverage by product differs between periods. The absence of some bulky product from the panel before 1870 implies a negative bias in the pace of integration during the “twilight of mercantilism”, while the sample for the “heyday of globalization”includes high-value goods whose price differentials were already very small by 1870 and could not fall further. We address this concern by running fixed-effect panel regression of the natural logarithm of price ratio as a function of 5 year time dummies (Bateman, 2011). 2The overall rate of change (in percentage) is 0.50 (not signiﬁcant), while rates by area are 0.39 for the Far East (signiﬁcant at 10%), 0.63 for India, 0.53 for Dutch East Indies and 0.51 for the United States (not signiﬁcant). Table 2 Shares of covered products on total exports (in percentage).

China India Indonesia Japan United States 1810 26.4 84.0 a 34.8 1830 45.0 b 67.1 50.8 1850 40.5 76.3 46.5 1870 95.4 55.5 73.3 28.8 64.9 1890 58.7 57.1 59.4 24.0 33.5 1900 33.3 46.5 56.4 20.6 22.2 1913 29.0 53.3 43.4 26.3 26.7 1929 20.3 46.0 57.2 29.4 17.1 1938 9.5 42.2 43.1 9.0 10.0 Sources: seeChilosi and Federico (2013: Appendix B)andFederico and Tena (2013). a1823.b1828. 4D. Chilosi, G. Federico / Explorations in Economic History 57 (2015) 1–18 The results (Fig. 2) confirm that if anything the linear trends understate the difference between 1815–1870 and 1870–1913. The pattern is almost identical also if we omit non-homogenous goods and very similar (although statistically less precise) if we run separateregressions for the four trade routes. On top of this, Fig. 2shows that price convergence concentrated in fairly short periods of time, most notably between 1815 and 1825, in the 1850s, and in the early 1870s and the 1880s. Thus, we look for structural breaks in the series Price ratios: selected commodities 0.5 1.0 1.5 2.0 2.5 3.0 3.5 1796 1821 1846 1871 1896 1921 Cotton New York Liverpool 1 2 3 4 5 1796 1821 1846 1871 1896 1921 Rice Rangoon London 1.0 1.5 2.0 2.5 3.0 3.5 4.0 1796 1821 1846 1871 1896 1921 Pepper Batavia Am s terdam 0 2 4 6 8 10 1796 1821 1846 1871 1896 1921 Saltpetre Calcutta London 0.5 1.0 1.5 2.0 2.5 3.0 1796 1821 1846 1871 1896 1921 Sugar Batavia London 0.4 0.8 1.2 1.6 2.0 2.4 1796 1821 1846 1871 1896 1921 Te a C a n to n L o n d o n Notes:the price ratios are computed with the yearly means of import and export prices.

Sources: see Chilosi and Federico (2013: Appendix B). Fig. 1. Price ratios: selected commodities.5 D. Chilosi, G. Federico / Explorations in Economic History 57 (2015) 1–18 of price ratios with aBai and Perron (1998, 2003)test.

We find at least one break in 23 series out of 29 and two or more breaks in six of them, for a total of 31 breaks.

About a third of these breaks feature also a shift in level exceeding a third of the differential before the break.

Table 5sums up the results. Column 1 reminds the number of series by period and country (fromTable 1).

Columns 2 to 4 report the number of trends, distinguishing between periods of convergence (negative and significantcoefficient of the time variable), divergence (positive and significant trend) and periods without a significant trend in price ratios. The number of trends (sum of Columns 2 to 4) exceeds the number of pair of markets (Column 1) whenever there is more than one trend within a given period. Then we count the number of breaks which entail also an upward (Column 5) or downward (Column 6) jump exceeding 10% of the last year of the previous trend.

We estimate the total change in each period/country (Column 7) and the contribution by shocks (Column 8), by averages of product specific figures. Last but not least, we compute an overall trend, comparable with the linear trends ofTable 4, as an average the product-specific yearly rates of change (Column 9).

As a whole, our analysis confirms that convergence was consistently and significantly faster during the “twilight of mercantilism”than during the“heyday of globalization”, although the difference for Indonesia is very small. The analysis adds three important results.

First, the timing of the breaks only weakly supports the traditional periodization: slightly less than half of the breaks (14 out of 31) fall within an interval of 6 years around 1815, 1870 or 1913 and almost as many (10) are Table 3 Long-run convergence: panel estimation.

N Initial ratioHalf-life (months)Rate (in percentage)Cumulated change (in percentage) All 1725 1.940 24 0.443*** 46.72 Atlantic 257 1.618 20 0.401*** 42.46 Far East 191 2.024 18 0.520*** 52.22 India 805 2.085 26 0.501*** 50.89 Indonesia 472 1.746 18 0.402*** 37.25 Significant at *10%, **5%, ***10%.

Notes: N = sample size. Fixed-effects estimation of Eq.(1)is used in all cases.

Sources: seeChilosi and Federico (2013: Appendix B).

Table 4 Trends by period: panel regression.

N Initial ratio Half-life (months) Rate (in percentage) Cumulated change (in percentage)Pooled group estimator Twilight of mercantilism (1796–1815) All 102 2.274 10 0.968 20.20 1.920 Atlantic 27 1.215 9 3.249 62.81 4.855 India 72 3.099 9 0.285 17.00 0.195 Early globalization (1815–1870) All 565 1.973 16 0.896*** 38.92 0.759** Atlantic 109 1.632 10 0.678*** 31.12 0.668*** Far East 48 2.245 12 1.501** 55.55 0.436 India 263 2.125 17 1.243*** 49.51 1.506*** Indonesia 145 1.892 12 0.504** 21.48 0.235 Heyday of globalization (1870–1913) All 765 1.291 8 0.418*** 16.44 0.379*** Atlantic 88 1.129 12 0.288*** 11.65 0.230 Far East 114 1.266 12 0.334** 13.37 0.235* India 330 1.326 7 0.517*** 19.95 0.461* Indonesia 233 1.311 8 0.347*** 13.85 0.359*** War and interwar (1914–1938) All 313 1.441 10 1.128*** 23.71 1.116*** Atlantic 37 1.084 7 0.160 3.78 0.260 Far East 27 1.157 0 0.250* 5.82 India 150 1.559 10 1.478** 28.86 1.425*** Indonesia 99 1.483 9 1.181** 24.69 1.045** Significant at * 10%; ** 5%; *** 10%.

Notes: N = sample size. N = sample size. Fixed-effects estimation of Eq.(1)is used in all cases.

Sources: seeChilosi and Federico (2013: Appendix B). 6D. Chilosi, G. Federico / Explorations in Economic History 57 (2015) 1–18 farther than 10 years from these dates. Second, breaks mattered a lot. In the“twilight of mercantilism”phase, on average, they accounted for one third of changes, as compared to a quarter during the“heyday”and a fifth during the last period. Third, the timing of the breaks often coincided with important institutional changes, detailed in the next section. Almost all the major shifts, 10 out of 12, clustered around 1815 or 1913—i.e. at the end of the EIC monopoly in Indian trade (1813) or at the outbreak of World War I. One exception is the tea exported from Canton to London, whose price ratio saw a sudden fall in 1835, 2 years after the demise of the EIC monopoly on that route; the other one is the sugar exported from India, whose London price spiked up in 1839, after the ban on the import of West Indian sugar cultivated by slaves. Moreover, the three negative shifts around 1870 are all from Indonesia, where, as will be detailed in the next section, the monopoly on foreign trade was abolished from 1868, considerably later than in India and China. 3By contrast, the Great Depression was definitely not a shock in the market for primary products.4. The causes of convergence: Transport costs and barriers to trade The conventional wisdom attributes the fall in the transaction costs of world trade before World War I mostly to the combined effects of liberalization of trade and declines in transport costs; improvements in communication and the gold standard are also routinely stressed as contributory factors. Similarly, the disinte- gration after the war is seen as the product of the protectionist backlash. We measure transport costs with the freight factor (the ratio of freights to price at the origin, assuming other costs to have been proportional to freights), and explore barriers to trade by looking at the nominal duty on each product—i.e. the sum of import duties in consuming countries and export taxes from India and Indonesia divided by the price at the origin.Fig. 3plots the transport costs and the duties for four representative commodities.

Levels are not directly comparable because the scales on the vertical axis differ, but trends are by and large similar across countries and products. As expect- ed, duties and freights did fall in the first part of the 19th century, while, consistent with the results ofTable 5, the rise in protection during the Great Depression is very limited. From 1930 to 1938, total duties were on average equivalent to 26.8% of the price in the exporting country, exceeding 50% in 22 observations out of 95 (most of them on sugar). A fortiori, the impact Long-run convergence: five year averages - 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1796 1821 1846 1871 1896 1921 Notes: The dependent variable is the natural logarithm of the ratio between import and export prices; the only independent variables are five years dummies. Fixed-effects estimation is used.

Sources: see Chilosi and Federico (2013: Appendix B). Fig. 2. Long-run convergence: 5 year averages. 3The remaining shift near 1870 is positive and signals a slowing down in the pace of integration of linseed between Calcutta and London. In fact, only in one case out of four does a structural break signal that the pace of integration increased after the 1870s (tin from Batavia).7 D. Chilosi, G. Federico / Explorations in Economic History 57 (2015) 1–18 of protection was limited over the whole period 1796– 1938: total duties exceeded 5% of prices in slightly more than a quarter of cases and 50% only in about an eighth.

On the same line, one should note that most of the fall in freights was over by 1875 and that the change was often quite sudden. The 75% collapse in the freight factor for saltpetre in 1816–1817 is an extreme case, but there are several other instances of major jumps, including a 70% fall in freights for tea from China to England in 1835. Such sudden changes cannot be accounted for by innovations in shipping organization or technology, the two competing explanations for the fall in costs of sea transport (North, 1968; Harley, 1988; Shah and Williamson, 2004), which by their nature are likely to spread gradually. Thus the evidence points to a major role for the dismantling of the institutional barriers to Asian trade.In Japan and China, these barriers were erected also by national governments to be dismantled by the force of Western powers: Japan was forced to open to trade in 1854, while the restrictions to Western merchants in China to trade in Canton only were lifted after the two opium wars in 1842 and 1858 (Dermigny, 1964; Fairbank, 1978; Wakeman, 1978). The British East India Company (EIC) enjoyed a monopoly on exports of Chinese tea to the United Kingdom and of imports of Indian opium into China until 1833, when it ceased all its mercantile activities. The monopoly of trade between its Indian possessions and the United Kingdom had been already abolished in July 1813. The Company had traditionally paid very high freights, because it used large and heavily manned and armed ships (Indiamen), very expensive to build and operate, which it rented from the owners. These Table 5 The four phases of globalization.

Number of trends Number of breaks Total change in the periodContribution of shocksImplicit rate of change Pairs of marketsConvergence Divergence Trendless Upward shockDownward shock(in percentage) (in percentage) (in percentage) Twilight of mercantilism (1796–1815) Atlantic 2 2 1.10 0.00 0.10 India 4 4 3.50 0.00 0.31 All 6 6 2.70 0.00 0.24 Early globalization (1815–1870) Atlantic 2 1 2 2 27.60 45.50 0.45 Far East 3 1 1 2 1 28.50 25.80 0.79 India 8 4 5 1 5 46.10 41.80 1.90 Indonesia 5 3 1 3 2 10.60 16.20 0.47 All 18 9 2 12 1 10 32.50 33.39 1.20 Heyday globalization (1870–1913) Atlantic 2 2 8.40 21.40 0.21 Far East 4 3 1 1 1 2.80 5.70 0.06 India 8 6 1 2 1 2 19.30 22.70 0.54 Indonesia 6 3 3 2 14.30 41.30 0.40 All 20 12 2 8 1 5 13.40 24.76 0.37 War and interwar (1914–1938) Atlantic 2 2 1.80 0.00 0.08 Far East 1 1 2.90 0.00 0.12 India 8 4 2 2 3 1.80 17.50 0.28 Indonesia 4 1 1 2 2 1 4.90 43.50 0.16 All 15 6 3 6 5 1 1.84 20.91 0.17 Notes: The dependent variable in each regression is the natural logarithm of the ratio between import and export prices; the only independent variable is the time trend. TheBai and Perron (1998, 2003)tests detect structural breaks in the constant and the slope. The total change and the implicit rate of change for each product in each period (Columns 7 and 9) are based on the formulaDelta = (rp T rp 0)/rp 0, whereDeltais the total change and the rpsdenote the fitted values at the end and beginning of the period. As at times trends and shifts offset one another, to examine the extent to which cumulated changes are explained by them, we decompose the sum of the absolute values of the predicted changes (Chilosi, 2014), using the following formula:Delta_abs = |(rp n1 rp n0)/rp n0| + |(rp n2 rp n1)*rp n1/(rp n1*rp n0)| + |(rp n3 rp n2)*rp n2/(rp n2*rp n0)| +…whererp nichanges whenever there is a shift and again with the new trend. In practice, there are at most two trends in each period (i.e.i= 1, 2 or 3). The figures reported in Column 8 are based on the proportions ofDelta_absexplained by the shifts.

Sources:Chilosi and Federico (2013: Table A1). 8D. Chilosi, G. Federico / Explorations in Economic History 57 (2015) 1–18 Barriers to trade for selected commodities a) Transport costs (freight factors) .00 .04 .08 .12 .16 .20 1800 1820 1840 1860 1880 1900 1920 .00 .01 .02 .03 .04 .05 1800 1820 1840 1860 1880 1900 1920 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 1800 1820 1840 1860 1880 1900 1920 .00 .05 .10 .15 .20 .25 .30 .35 1800 1820 1840 1860 1880 1900 1920 Cotton New York Liver poolIndigo Calcutta London Saltpetre Calcutta LondonSugar Batavia London b) Import and export duties Notes: the freight factors are equal to the nominal freights divided by the price at the origin. Similarly, the duties are equal to the sum of import and export duties divided by the price at the origin.

Sources: see Chilosi and Federico (2013: Appendix B). .00 .02 .04 .06 .08 .10 .12 .14 .16 1800 1820 1840 1860 1880 1900 1920 .00 .02 .04 .06 .08 .10 1800 1820 1840 1860 1880 1900 1920 0.0 0.2 0.4 0.6 0.8 1.0 1800 1820 1840 1860 1880 1900 1920 0 1 2 3 4 5 1800 1820 1840 1860 1880 1900 1920 Cotton New York LondonIndigo Calcutta London Saltpetre Calcutta London Sugar Batavia London Fig. 3. Barriers to trade for selected commodities.9 D. Chilosi, G. Federico / Explorations in Economic History 57 (2015) 1–18 were a small group of Londoners, many of whom were also shareholders of the Company. Although the owners were barred from participation in the Court of Directors since 1710, they infiltrated it with the ships' husbands and captains, thus forming a cohesive oligopoly within the monopoly. According to the conventional wisdom, the “shipping interest”, as it was known at the time, allowed the inflating of costs and thus the circumvention of the Company's official charter, which prevented it from earning extra-profits from trade (Wakeman, 1978; Mui and Mui, 1984:63–64;Bowen, 2006: 252–256;Webster, 2009:32–40). 4Solar (2013)has recently suggested that the Indiamen were necessary to stave off the military threat of competing Western trading companies. In his view freights collapsed in the 1810s thanks to the combined effect of technical progress (mostly copper sheathing) and to the successes of the Royal Navy, which made it possible to use smaller and cheaper ships. Anyway, since 1813 trade with India was free and in all likelihood the market was competitive. The expectation of great profits from liberalized trade caused a glut in the market for transportation for India (Webster, 2009: 72) and indeed our series of freight collapses in 1817.

The Dutch trade with its colonies in the Indian Ocean had been a monopoly of theVereenigde Oost-Indische Compagniefrom the early 17th century to its bankruptcy in the 1790s. Trade remained free until 1825, when a monopoly on commerce with the Netherlands was granted to a new trading company, theNederlandsche Handel- Maatschappij, or NHM (Furnivall, 1976; Horlings, 1995:

142). Similarly to the EIC, the company rented ships from Dutch owners, at“exceptionally high”rates (Korthals Altes, 1994: 161;Horlings, 1995: 145). In its first years of activity, exports were reduced by a rebellion of natives, the so called Java War. At the end of the war, the Dutch government, desperate for revenues, established a system of compulsory delivery of coffee, sugar and indigo for exports, known as Cultivation System (de Klerck, 1938; Dobbin, 1983; Fasseur, 1992; Houben, 2002; van Zanden and Marks, 2012). Peasants were paid much less than the world market price—about a half of the Batavia price for coffee (Fasseur, 1992: 37). The goods were transported to Amsterdam, and there sold at auction: the profits, net of a fee for the NHM, accrued to the Dutch government. The amount was very substantial: at its peak, in the 1850s, itaccounted for over half the state revenues and for about 3.8% of Dutch GDP excluding any hidden subsidy to Dutchshippingandindustry(Smits et al., 2000; van Zanden and van Riel, 2004). However, the Cultivation System was increasingly unpopular at home and was slowly phased out. In 1850, the NHM liberalized the bidding contracts for renting ships and its monopoly was abolished altogether in 1868 (NHM, 1924:23;Furnivall, 1976:168;Korthals Altes, 1994). The Dutch trade with the colonies remained free until 1918, when the sugar producers set up a private association (VJSP) to allocate the scarce available shipping (van der Eng, 1996:215– 216;Knight, 2010). The organization continued to manage sugarexportsaftertheendofthewar,andin1932itwas substituted by a governmental organization (NIVAS) to manage sugar production quotas under the international agreement (van der Eng, 1996:224–226). The Great Depression featured also the first intervention in American agriculture: the Agricultural Adjustment Act (AAA), part of the New Deal policies, established a loan facility for cotton farmers, which in practice set a minimum price of cotton since 1934 (Federico and Sharp, 2013).

5. The causes of integration: An econometric analysis As implied by the preceding discussion, the ratio of prices for theith good (RP it) at timetcan be explained by the barriers to trade (B), the efficiency of markets (E m) and the transport costs (T c) Log RP it¼FB;E m;Tc ð2Þ In our regression the setBincludes the total duties (LOG_DUTY) and dummies for monopolies—adummy for theEIC(1796 to 1816) and separate dummies for the NHMunder the full monopoly regime (NHM1), from 1824 to 1850, and for the partially liberalized one (NHM2) from 1851 to 1868. 5We also add dummies for the AAA support to American cotton prices (since 1933) and for the two marketing boards for Javanese sugar, the private VJSP (1918–1931) and the public NIVAS (after 1932).

We expect duties and monopolies to increase price ratios, 4The EIC charter stipulated that the price of company wares in London“should not exceed the prime cost, the freight and charges of importation, the lawful interest of capital from the time of arrival of such tea in Britain, and the common premium in insurance" (Mui and Mui, 1984: ix). Indeed, the ofﬁcial proﬁts of the Company were quite low (Wakeman, 1978: 167). 5The variableLOG_DUTYis computed aslog(1 + t), wheretis the ratio of (usually speciﬁc) duties to the price in the producing country.

It is thus zero ift=0. The dummyEICis equal to 1 until 1816. We do not add a speciﬁc dummy for the end of the Napoleonic Wars because it would be collinear with theEICdummy and it might bias the estimate of its coefﬁcient. Anyway, price gaps with Asia in Britain were much less affected than in continental Europe by the Napoleonic Wars (O’Rourke, 2006). 10D. Chilosi, G. Federico / Explorations in Economic History 57 (2015) 1–18 while the interventions on the commodity market may augment or reduce price gaps according to the details of the policies.

Arguably the most important contribution to increas- ing market efficiency was the connection by telegraph,which cut the time to transmit information from weeks to few minutes (Hoag, 2006). The cable between the United Kingdom and the United States was operational since 1866, and it was followed 4 years later by the line to India and around 1875 by the line between Europe Freight factors and adjusted freight factors: selected commodities i) Saltpetre Calcutta London 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 1796 1801 1806 1811 1816 1821 1826 1831 1836 1841 Freight factorAdjusted freight factor ii) Sugar Batavia London .00 .04 .08 .12 .16 .20 .24 1823 1833 1843 1853 1863 1873 1883 Freight factorAdjusted freight factor Notes: The freight factor is the nominal freight rate divided by the price at the origin. The adjusted freight series are constructed by firstly running the regression LOG_FREIGHT it=c +α i+β1EIC it+β2NHM1 it+β3NHM2 it+εitand secondly using the equation LOG_ADJFREIGHT it=LOG_FREIGHT it-(β1EIC it+β2NHM1 it+β3NHM2 it). Using alternative specifications, like including a time trend, yielded qualitatively identical results and only small quantitative differences.

Sources: see the text and Chilosi and Federico (2013: Appendix B). Fig. 4. Freight factors and adjusted freight factors: selected commodities. i) Saltpetre Calcutta London. ii) Sugar Batavia London.11 D. Chilosi, G. Federico / Explorations in Economic History 57 (2015) 1–18 and Indonesia (Headrick, 1988: 101). We expect the TELEGRAPHdummy to be negative. We also add dummies for a number of political events which may have disrupted the orderly working of markets, such as the Java War in 1825–1827 (JAVA WAR), the Indian Mutiny in 1857–1859 (MUTINY), the American Civil War (CIVIL WAR), the anti-slavery campaign, which boycotted Caribbean sugar, in 1840–1845 (SLAVE) and World War I (WWI).

By definition, the actual freights, the numerator of our measure of transport costs (LOG_FREIGHT), include any rent paid to privileged Dutch or British ship-owners under the NHM and EIC monopoly system. Consequently the EICandNHMdummies measure only any additional effect of the trading monopoly—e.g. from a manipulation of the market. We try to capture the total effect of monopoly, including the rents to ship-owners by using an alternative series of freight factor (LOG_ADJFREIGHT) net of the rents to shippers. We obtain this series (Fig. 4) by scaling down the actual series with coefficients from a fixed-effect panel regression, which explains freights with dummies for the NHM and the EIC. By construction, the series move in parallel toLOG_FREIGHT,exceptinthe final year of the monopoly and thus the substitution affects the coefficient of transport costs only marginally.

We add the lagged value of the dependent variable to reduce auto-correlation and to take into account the possible delay in adjustment to shocks, but its omission does not affect qualitatively the results. The coefficients are thus short term elasticities; long run elasticities can be computed asβ k/(1 γ) whereβ kis the coefficient of thekth variable andγis the coefficient of the lagged dependent variable.

The descriptive statistics for all variables and the pairwise coefficient of correlation between them do not add much new information (Chilosi and Federico, 2013).

The correlation between explicative variables is in general very low and thus there is no risk of multi-collinearity. On the other hand, most variables are clearly non-stationary (cf.Chilosi and Federico, 2013: Table A2, for a formal testing) and thus results might be spurious. We therefore test ex-post the stationarity of the residuals with aLevin et al. (2002)test for panel regressions.

Our analysis omits from the regression the Far Eastern markets, because the series are very short and their quality is comparatively poor. This leaves a total of 22 cross- sections and 1568 observation. As said before, we cannot categorically exclude that quality may be an issue for wheat (both from the United States and from India), coffee sugar, pepper and tin (all from Indonesia). Thus, in Column 3 ofTable 6we drop these products, reducing the coverage to 16 products and 1020 observations. Last butnot least, Indian trade, with 838 observations, is overrepresented in the sample. We address this issue in two different ways. First, we run the regression with a restricted panel, featuring two products per route (wheat and cotton for the United States, sugar and indigo for India, and pepper and sugar for Indonesia), and we weight each of them with their share of the value of imports on that route. Thus, in this specification (Table 6, Column 6), each route has the same weight on the results. Second, we simply run separate regressions for the United States (Table 6, Column 7), India (Column 8) and Indonesia (Column 9), which supply additional insights on the different causes of integration.

All regressions use fixed-effects specification with panel corrected standard errors to address cross-sectional heteroskedasticity (reflecting different levels of transac- tion costs) and contemporaneous correlation (which may arise from common shocks). Although we report results of an OLS estimate (Table 6, Column 1) for comparative purposes, we prefer to use instrumental variables because prices in the exporting country appear in the denominator of the dependent variable and of two explicative ones (duties and freights). Furthermore, freights might be endogenous in the short-run as well, if the supply of shipping on those routes were inelastic (Jacks and Pendakur, 2010). Specifically, we instrumentLOG_DUTY with the ratio of yearly duties to the average price throughout the whole period—so that within variations depend only on the exogenous changes in specific duties.

Likewise we use average prices as denominator of the instrument forLOG_FREIGHT,while the numerator is the trend component of a Hodrick–Prescott decomposition of the series of nominal freights.

The overall performance of the model is good. The residuals are stationary; the combined variables are highly significant (F-test) and explain about four fifths of the total variance. Almost all the signs agree with expectations. In all cases, theWooldridge's (1995)and the regression-based tests find strong evidence of that the OLS estimates sufficiently differ from the instru- mental variables' ones to recommend the use of the latter. All first stage tests find that the instruments are highly correlated with the variables, so that the expected small-sample bias is low. Turning to the results:

i) As expected, the coefficients for freights are positive, highly significant and robust to the substitution of monopoly-adjusted series (LOG_ ADJFREIGHT). A 10% increase in freights aug- mented price gaps by about 0.6–0.9% in the short run and by about 1–1.5% in the long run. Moreover, as expected, the impact of costs is much bigger for 12D. Chilosi, G. Federico / Explorations in Economic History 57 (2015) 1–18 bulky goods than for than for light products; over three times, in fact (Column 5). 6 ii) The variableLOG_DUTYis positive and significant in the baseline specification and its coefficient is similar to the coefficient of freights. Indeed, the averages of the two variables are almost identical:

18.1% for the freight factor and 18.2% (17% on imports and 1.2% on exports) for duties. Duties were high only on wheat, under the British Corn Laws before 1846, and sugar. Infact, the coefficient of LOG_DUTYis high and significant for the Atlantic trade (Column 7), low and significant only at the 13% level for Indonesia (Column 9) and very low and not significant for India (Column 8).

iii) As expected, the monopoly of the British East India Company widened the price differentials, by a quarter in the baseline specification withLOG_ FREIGHT(Column 2) and by more than a third (corresponding to a 70% increase in the long run) if we add the effect of monopoly on freights (Columns 4 and 7). In contrast, the NHM apparently fostered convergence both before (NHM1)andafter1850 (NHM2). This effect however disappears if we use the monopoly-adjusted series of freights (Columns 4 and 8). TheNHMdummies become positive and significant, although the coefficients are less than a quarter of theEICone. In short, the NHM affected negatively long-range integration only because it chargedhighfreights,whileitmayhaveeven increased the efficiency of the market, by improving the transmission of information and by reducing the risks. 7In contrast, the monopoly of the EIC harmed trade even discounting the effect on freights. The overall effect of state intervention after World War I seems modest. Neither the Dutch (private) market- ing board (VJSP) in the 1920s nor the support to American agriculture after the Great Depression (AAA) affected significantly integration. Only the Dutch public marketing board (the NIVAS) had a positive impact, reducing price gaps between Indonesia and Europe by 10–15%.v) The variableTELEGRAPHis negative as expect- ed and significant in all specifications. 8We have tried to interact it with a time trend to capture the effect of technical progress and increases in transmission (or changes in policy to set rates), but the results are poor. Our results confirm the earlier work byLew and Cater (2006)on the positive effect of telegraph on world trade.

vi) The political shocks in producing countries seem not to have affected international price differen- tials, although of course they may have had important consequences in producing areas. In contrast the campaign against slave-produced sugar in the United Kingdom (SLAVE) had a massive effect on the sugar market. The dummies forWWIare positive and significant, as expected, for India and Indonesia but not for the Atlantic.

We interpret them as the effect of disruption in the market on top of war-related increase in transport costs, which should already be captured by the variableLOG_FREIGHT.

How much did each variable contribute to long run convergence? To answer, inTable 7we report the share of the total change accounted for by each variable.

As the last row ofTable 7shows, the model performs very well: the divergence between the cumulated effect of the variables and the actual change is less than 5% in four cases and exceeds 20% only in the Atlantic trade. Most of the total convergence is explained by changes in transportation costs and by the abolition of the EIC, which looms large in the long run analysis, partly because the majority of observations in 1815 refer to India. 9 According to the baseline specification (Column 1), the 6We run an OLS regression withLOG_FREIGHTas dependent variable, explained by a linear trend and by route and product dummies. The product dummies yield a ranking of commodities from the lightest to the heaviest (silk, indigo, tin, cotton, tea, coffee, pepper, rubber, sugar, wheat, jute, saltpetre, linseed, rapeseed, and rice). We deﬁne‘light’theﬁrst nine products (“light”) and‘bulky’the others:

this distinction closely mirrors the conventional one between“grain and seeds”and“lighter goods”.

7An ofﬁcial history of the Company offers an alternative explanation of the negative signs in the baseline speciﬁcations: it claims that the NHM sold at a loss to help Dutch middlemen to be competitive on the European market (NHM, 1924:18–19). 8We have tested separately three additional measures of efﬁciency:

time trends, total traded quantity and a dummy forﬁxed exchange rates between countries (i.e. the gold standard). All variables are expected negative. More trade should increase theﬂow of informa- tion,ﬁxed exchange rates should reduce the risks of trading and time trends should capture all other improvements. The trends are signiﬁcant only in Indonesia, but the variable is unexpectedly positive—i.e.ceteris paribusmarkets have been disintegrating. The quantity variable is negative, but it worsens the overall results of the regression (available upon request), probably because of endogeneity issues. The dummy forﬁxed exchange rate is incorrectly signed but not signiﬁcant. Evidently, more reﬁned measures of exchange rate volatility than afforded by the available data are needed to adequately capture the effect of exchange rate risk. The time trend is positive and signiﬁcant for Indonesia—i.e. efﬁciency would have decreased, ceteris paribus. We speculate that the coefﬁcient reﬂects the increasing exposure to price shocks originating in other markets.

9But even if we give each route equal weight (i.e. using the sixth speciﬁcation fromTable 6), the EIC emerges as by far the most important factor, accounting for over 45% of the overall decline.13 D. Chilosi, G. Federico / Explorations in Economic History 57 (2015) 1–18 Table 6 The causes of integration.

(1) (2) (3) (4) (5) (6) (7) (8) (9) PCSE PCSE/IV PCSE/IV PCSE/IV PCSE/IV PCSE/IV PCSE/IV PCSE/IV PCSE/IV C0.517 0.365 0.327 0.366 0.329 0.224 0.312 0.386 0.416 (17.18)*** (8.73)*** (5.53)*** (8.73)*** (7.92)*** (4.48)*** (4.93)*** (5.51)*** (6.84)*** LOG_DUTY0.084 0.051 0.048 0.051 0.033 0.039 0.13 0.021 0.057 (3.64)*** (2.03)** (0.87) (2.04)** (1.28) (1.31) (2.75)*** (0.32) (1.54) EIC0.173 0.236 0.253 0.345 0.287 0.347 0.343 (5.05)*** (6.30)*** (5.66)*** (9.98)*** (7.38)*** (7.04)*** (9.12)*** NHM1 0.105 0.044 0.021 0.054 0.021 0.078 0.137 ( 4.48)*** ( 1.68)* ( 0.22) (2.26)** (0.74) (2.37)** (4.83)*** NHM2 0.07 0.046 0.013 0.023 0.022 0.049 0.087 ( 3.42)*** ( 2.26)** (0.34) (1.04) ( 1.05) (1.50) (3.43)*** AAA0.038 0.032 0.032 0.032 0.02 0.018 0.012 (0.79) (0.69) (0.67) (0.69) (0.46) (0.25) (0.28) VJSP 0.079 0.035 0.035 0.011 0.011 0.031 ( 2.08)** ( 0.90) ( 0.90) ( 0.28) ( 0.24) ( 0.78) NIVAS 0.185 0.125 0.126 0.105 0.105 0.153 ( 3.61)*** ( 2.35)** ( 2.35)** ( 1.93)* ( 1.60) ( 2.83)*** LOG_FREIGHT0.123 0.067 0.053 0.085 (13.27)*** (4.64)*** (2.48)** (3.65)*** LOG_ADJFREIGHT0.067 0.034 0.068 0.073 (4.64)*** (2.01)** (2.64)*** (4.44)*** LOG_FREIGHT*LIGHT0.033 (2.14)** LOG_FREIGHT*BULKY0.108 (6.73)*** TELEGRAPH 0.043 0.082 0.095 0.081 0.083 0.076 0.017 0.117 0.046 ( 2.57)** ( 4.31)*** ( 3.63)*** ( 4.31)*** ( 4.45)*** ( 3.75)*** ( 0.74) ( 3.32)*** ( 2.01)** 14D. Chilosi, G. Federico / Explorations in Economic History 57 (2015) 1–18 JAVA_WAR0.064 0.063 0.063 0.072 0.071 0.066 (0.83) (0.80) (0.80) (0.89) (0.86) (0.82) MUTINY 0.071 0.066 0.068 0.066 0.085 0.060 0.094 ( 0.82) ( 0.75) ( 0.76) ( 0.75) ( 0.99) ( 0.47) ( 1.06) SLAVE0.239 0.195 0.285 0.195 0.189 0.139 0.293 0.148 (5.62)*** (4.41)*** (4.91)*** (4.41)*** (4.20)*** (2.70)*** (4.98)*** (2.38)** CIVIL_WAR 0.009 0.055 0.08 0.055 0.085 0.082 0.006 ( 0.26) ( 1.47) ( 1.35) ( 1.46) ( 2.25)** ( 1.61) ( 0.15) WWI*ATLANTIC 0.067 0.012 0.033 0.012 0.048 0.044 0.034 ( 1.23) (0.21) (0.55) (0.21) (0.92) (0.62) ( 0.58) WWI*INDIA0.087 0.151 0.167 0.151 0.144 0.086 0.152 (2.08)** (3.40)*** (3.40)*** (3.40)*** (3.32)*** (1.14) (2.93)*** WWI*INDONESIA0.171 0.212 0.181 0.212 0.242 0.278 0.227 (5.67)*** (6.59)*** (4.34)*** (6.58)*** (7.09)*** (4.85)*** (7.27)*** LOG_RATIO10.445 0.497 0.513 0.496 0.463 0.452 0.503 0.51 0.338 (17.77)*** (18.00)*** (15.80)*** (18.00)*** (16.52)*** (12.11)*** (8.70)*** (14.90)*** (7.86)*** N1534 1534 1020 1534 1534 619 231 821 482 Adjusted R-square0.79 0.78 0.79 0.78 0.79 0.64 0.76 0.79 0.76 F153.67*** 129.36*** 120.24*** 129.37*** 129.61*** 49.38*** 77.07*** 133.66*** 76.88*** LLC t-stat 33.12*** 37.82*** 32.65*** 37.82*** 37.65*** 22.88*** 16.25*** 28.45*** 20.65*** Exogeneity chi-square test24.82*** 22.79*** 24.85*** 30.43*** 10.97*** 7.85** 20.20*** 7.70** Exogeneity F test12.90*** 12.90*** 12.83*** 11.31*** 3.81** 7.89*** 12.23*** 3.95** Shea's partial R-square 1st inst.0.89 0.86 0.89 0.88 0.75 0.96 0.94 0.74 Shea's partial R-square 2nd inst.0.46 0.42 0.46 0.46 0.32 0.41 0.45 0.44 Shea's partial R-square 3rd inst.0.6 Significant at *10%, **5%, *** 1%.

Notes: PCSE = panel corrected standard errors; IV = instrumental variables estimation;N= sample size;inst.= instrument. The dependent variable in each regression is the natural logarithm of the ratio between import and export prices. Fixed-effects estimation with panel corrected standard errors is used in all cases; instrumental variable estimation for specifications (2) to (9); weighted least squares estimation for specification (6). Specification (3) includes only homogeneous goods; specification (6) includes only two goods per route;specifications (7) to (9) include only goods exported from the U.S., India and Indonesia, respectively.

Sources: seeChilosi and Federico (2013: Appendix B). 15 D. Chilosi, G. Federico / Explorations in Economic History 57 (2015) 1–18 fall in freights mattered almost as much, but this conclusion is decidedly changed if we use the coefficient fromLOG_ADJFREIGHTS(Column 4). The NHM does not appear among the variables in the aggregate analysis, because the series for Indonesia start later, but as Column 7 shows, it played a very important role in the convergence of prices between the colony and Europe, too. The telegraph did help a lot as well, accounting for between a sixth and a quarter of long-run price convergence. In contrast, the cut in duties contributed substantially to the integration only in the Atlantic economy (Column 5). The separate estimates by period (Columns 2 and 3) highlight the sharp differences in the causes of integration. The “early globalization”was mostly determined by the trade liberalization and the abolition of monopolies, while further convergence during the“heyday of globalization” was achieved thanks to the lay-out of the telegraph lines and to fall in sea-borne transport costs. On the whole, political decisions mattered more in the long run because, as shown inSection 4, most of total convergence pre-dated 1870.

6. Conclusion Our results are relevant for two literatures, that on market integration and that on trade and growth in Asia, which, although clearly related, have so far remained largely distinct. Our work contributes to filling in gaps in the literature on global market integration, strength- ening the emerging consensus view in the literature on Europe and the Atlantic trade. The process started early in the 19th century and it was determined to a large extent by institutional changes. Within Europe and betweenEurope and North America, barriers were raised essen- tially by protectionist trade policies, while commerce with Asia was hampered by the monopoly of Western trading companies. These barriers were progressively abolished and, at least for the sample of products/routes we are considering, were only partially re-instated during the Great Depression. Once trade was free from institutional constraints, further convergence was mainly achieved by cutting transportation costs. The cost of sea-borne trade, however, was fairly low already at the beginning of the 19th century and thus the scope for further convergence was limited. 10These findings are consistent with the view that market integration was at the root of the terms of trade boom experienced by Asian countries in the decades before 1870.

11 Future research should systematically assess this impact and examine the implications of market integration for trade and welfare.

Appendix A. Supplementary data Supplementary data to this article can be found online athttp://dx.doi.org/10.1016/j.eeh.2015.04.001. 10In contrast, transport costs from the producing areas to the ports were surely high and indeed there is a strong evidence of growing integration in the second half of the 19th century in the domestic market in the United States (Federico and Sharp, 2013), in India (Hurd, 1975; Studer, 2008; Andrabi and Kuehlwein, 2010) and also in the Dutch East Indies (Marks, 2010; van Zanden and Marks, 2012:

25–26). Indeed, in a recent paper,Donaldson (forthcoming)estimates that on average trade created by railways increased Indian GDP by as much as a sixth from 1870 to 1930.

11This is particularly so as new estimates of terms of trade show that India took part in this boom, too (Chilosi and Federico, 2013: 8). Table 7 The causes of integration: decomposition analysis (in percentage).

(1) (2) (3) (4) (5) (6) (7) Sample All All All All Atlantic India Indonesia Years 1816–1913 1816–1870 1871–1913 1816–1913 1816–1913 1815–1913 1849–1913 EIC39.27 49.57 57.25 69.14 NHM143.36 LOG_DUTY4.65 5.56 1.14 4.65 42.11 1.72 2.98 LOG_FREIGHT34.74 42.3 59.48 82.49 LOG_ADJFREIGHT16.79 8.16 24.99 TELEGRAPH20.34 15.8 37.3 20.31 13.08 23.47 14.6 Total 98.99 113.24 97.92 98.99 137.68 102.5 85.92 Notes: All the decompositions report the share of the percentage change in the dependent variable accounted for by each independent variable and are based on the results of the panel analysis presented inTable 6. Columns (1) to (3) are based on the second specification, Column (4) is based on the fourth specification, and Columns (5) to (8) are based on the seventh, eighth and ninth specifications, respectively. The exact starting date differs somewhat across samples in order to maximise the number of covered goods. We omit explicative variables which did not affect the dependent variables at the beginning and/or at the end of the period (such as the World War dummies). The cumulated change in the dependent variable is obtained as the average of the differences between the values fitted by the panel at time0and that of timeTdivided by the former.

Sources: see the text andChilosi and Federico (2013: Appendix B). 16D. Chilosi, G. Federico / Explorations in Economic History 57 (2015) 1–18 References Allen, R.C., 2011.Global Economic History. A Very Short Introduction.

Oxford University Press, Oxford.

Andrabi, T., Kuehlwein, M., 2010.Railways and price convergence in British India. Journal of Economic History 70, 351–377 . Bai, J., Perron, P., 1998.Estimating and testing linear models with multiple structural changes. Econometrica 66 (1), 47–78 . Bai, J., Perron, P., 2003.Computation and analysis of multiple structural change models. Journal of Applied Econometrics 18, 1–22 . Bateman, V.N., 2011.The evolution of markets in early modern Europe:

a study of wheat prices, 1350-1800. The Economic History Review 64, 447–471 . Booth, A., 1988.Agricultural Development in Indonesia. Allen and Unwin, Sidney.

Bowen, H.V., 2006.The Business of Empire. The East Company and Imperial Britain, 1756-1833. Cambridge University Press, Cambridge.

Brandt, L., Ma, D., Rawski, T., 2013.From divergence to convergence: re-evaluating the history behind China's economic boom. LSE Working Paper 175/13.

Broadberry, S., 2013.Accounting for the Great Divergence. LSE Economic History Working Paper 184/2013.

Brunt, L., Cannon, E., 2014.Integration in the English wheat market 1770-1820. Explorations in Economic History 52, 111–130 . Chilosi, D., 2014.Risky institutions: political regimes and the cost of public borrowing in early modern Italy. Journal of Economic History 74, 887–915 . Chilosi, D., Federico, G., 2013.Asian globalizations: market integration, trade and economic growth, 1800–1938. Economic History Working Papers No. 183/2013. London School of Economics.

De Klerck, E.S., 1938.History of the Netherlands East Indies vol. 2 (Rotterdam: Brusse).

Dermigny, L., 1964.La Chine et l'Occident. Le commerce a Canton au XVIIIe Siècle 1719-1833. Tome III. SEVPEN, Paris.

Dobbin, C., 1983.Islamic Revivalism in a Changing Peasant Economy. Central Sumatra, 1784-1847. Curzon Press, London and Malmö.

Donaldson, D., 2014.Railroads and the Raj: estimating the Economic Impact of Transportation Infrastructure. American Economic Review (forthcoming) . Dutt, R., 1969.first ed. The Economic History of India vols. 1-2. A.

Kelley, New York (1902-1904).

Fairbank, J., 1978.The creation of the treaty system. In: Twitchett, D., Fairbank, J. (Eds.), The Cambridge History of China vol. 10, part I. Cambridge University Press, Cambridge, pp. 213–264.

Fama, E., 1970.Efficient capital markets: a review of theory and empirical work. Journal of Finance 25, 383–417 . Fasseur, C., 1992.The Politics of Colonial Exploitation. Java, the Dutch and the Cultivation System. South-East Asia Program Cornell University, Ithaca.

Federico, G., 2011.When did the European market integrate? European Review of Economic History 15, 93–126 . Federico, G., 2012.How much do we know about market integration in Europe? The Economic History Review 65 (2), 470–497 . Federico, G., Sharp, P., 2013.The Cost of railroad regulation: the disintegration of American agricultural markets in the interwar period. The Economic History Review 66, 1017–1038 . Federico, G., Tena, A., 2013.World trade 1800–1938. paper presented to the Seventh World Cliometric Conference (Hawaii).

Feuerwerker, A., 1980.Economic trends in the late Ch'ing Empire, 1870- 1911. In: Fairbank, J., Liu, K. (Eds.), The Cambridge History of China vol. 11, part II. Cambridge University Press, Cambridge, pp. 1–69.Feuerwerker, A., 1983.Economic trends, 1912–1949. In: Fairbank, J.

(Ed.), The Cambridge History of China vol. 12, part I. Cambridge University Press, Cambridge, pp. 28–127.

Furnivall, J.S., 1976.Netherlands India. A Study of a Plural Economy.

B. M. Israël., Amsterdam.

Harley, K., 1988.Ocean freight rates and productivity, 1740–1913:

the primacy of mechanical invention reaffirmed. Journal of Economic History 48, 851–876 . Headrick, D., 1988.The Tentacles of Progress. Technology Transfer in the Age of Imperialism, 1850-1940. Oxford University Press, Oxford.

Hoag, Christopher, 2006.The Atlantic cable and capital market information flows. Journal of Economic History 66, 342–353 . Horlings, E., 1995.The Economic Development of the Dutch Service Sector, 1800-1850. NEHA, Amsterdam.

Houben, V.J.H., 2002.Java in the 19th century: consolidation of a territorial state. In: Dick, H., Houben, V.J.H., Lindblad, J.T., Wie, T.K. (Eds.), The Emergence of a National Economy. An Economic History of Indonesia, 1800-2000. Allen & Unwin and University of Hawai'i Press, Crows Nest and Honolulu, pp. 56–81.

Hurd, J., 1975.Railways and the expansion of markets in India, 1861- 1921. Explorations in Economic History 12, 263–288 . Hynes, W., Jacks, D.S., O'Rourke, K.H., 2012.Commodity market disintegration in the interwar period. European Review of Economic History 16, 119–143 . Jacks, D.S., 2005.Intra–and international commodity market integration in the Atlantic economy. Explorations in Economic History 42, 381–413 . Jacks, D.S., Pendakur, K., 2010.Global trade and the maritime transport revolution. The Review of Economics and Statistics 92, 745–755 . Knight, R., 2010.Exogenous colonialism: Java sugar between Nippon and Taikoo before and during the interwar depression, c.

1920–1940. Modern Asian Studies 44, 477–515 . Korthals Altes, W.L., 1994.Changing Economy in Indonesia vol. 15 Prices (non Rice) 1814-1940. Royal Tropical Institute, The Hague.

Levin, A., Lin, C.F., Chu, C., 2002.Unit root tests in panel data:

asymptotic and finite-sample properties. Journal of Econometrics 108, 1–24 . Lew, B., Cater, B., 2006.The telegraph, co-ordination of tramp shipping and growth in world trade 1870-1910. European Review of Economic History 10, 147–173 . Maddison Project, 2013. . At,www.ggdc.net/maddison/maddison- project/data.htm(Accessed on January 25th 2013).

Marks, D., 2010.Unity or diversity? On the integration and efficiency of rice markets in Indonesia, c.1920-2006. Explorations in Economic History 47, 310–324 . Mui, H., Mui, L.H., 1984.The Management of Monopoly. A Study of the English East India Company's Conduct of its Tea Trade, 1784- 1833. University of British Columbia Press, Vancouver.

Nederlandsche Handel-Maatschappij, 1924.A Brief History of the Netherlands Trading Society, 1824-1924. Mouton & Company, The Hague.

North, D.C., 1968.Sources of productivity change in ocean shipping, 1600-1850. Journal of Political Economy 76, 953–970 . O'Rourke, K.H., 2006.The worldwide economic impact of the French Revolutionary and Napoleonic Wars, 1793-1815. Journal of Global History 1, 123–149 . O'Rourke, K.H., Williamson, J.G., 2002.When did globalization begin? European Review of Economic History 6, 23–50 . Parthasarathi, P., 2011.Why Europe Grew Rich and Asia Did Not.

Global Economic Divergence, 1600-1850. Cambridge University Press, Cambridge.17 D. Chilosi, G. Federico / Explorations in Economic History 57 (2015) 1–18 Pomeranz, K., 2000.The Great Divergence. Europe, China and the Making of the Modern World Economy. Princeton University Press, Princeton.

Razzaque, M., Osafa-Kwaako, P., Grynberg, R., 2007.Long-run trend in the relative price: empirical estimation for individual commodities.

In: Grynberg, R., Newton, S. (Eds.), Commodity Prices and Development. Oxford University Press, Oxford, pp. 35–67 (2007).

Rönnbäck, K., 2009.Integration of global commodity markets in the early modern era. European Review of Economic History 13, 95–120 . Roy, T., 2000.The Economic History of India, 1857–1947. Oxford University Press, Oxford.

Sachs, J., Warner, A.M., 2001.Natural resources and economic development: the curse of natural resources. European Economic Review 45, 827–838 . Shah, M., Williamson, J.G., 2004.Freight rates and productivity gains in British tramp shipping 1869-1950. Explorations in Economic History 41, 174–203 . Sharp, P., Weisdorf, J., 2013.Globalization revisited: market integration and the wheat trade between North America and Britain from the eighteenth century. Explorations in Economic History 50, 88–98 . Smits, J.P., Horlings, E., van Zanden, J.L., 2000.The Measurement of Gross National Product and its Components, 1800-1913(Groningen Growth and Development Centre Monograph series no 5 2000. data available at,http://nationalaccounts.niwi.knaw.nl/(accessed in January 2009).

Solar, P.M., 2013.Opening to the East. Shipping between Europe and Asia, 1780-1830. Journal of Economic History 73, 625–661 . Studer, R., 2008.India and the great divergence: assessing the efficiency of grain markets in 18th and 19th century India. Journal of Economic History 68, 393–437 . Tomlinson, B.R., 1993.The Economy of Modern India, 1860–1970.

Cambridge University Press, Cambridge.Uebele, M., 2011.National and international market integration in the 19th century: evidence from comovement. Explorations in Economic History 48, 226–242 . Van der Eng, P., 1996.Agricultural Growth in Indonesia. Macmillan, London and Basingstoke.

Van Zanden, J.L., Marks, D., 2012.An Economic History of Indonesia, 1800–2010. Routledge, London and New York.

Van Zanden, J.L., van Riel, A., 2004.The Strictures of Inheritance.

The Dutch Economy in the Nineteenth Century. Princeton University Press, Princeton.

Wakeman, F., 1978.The Canton trade and the Opium war. In: Twitchett, D., Fairbank, J. (Eds.), The Cambridge History of China vol. 10, part I.

Cambridge University Press, Cambridge, pp. 163–212.

Webster, A., 2009.The Twilight of the East India Company. Boydell Press, Woodbridge.

Williamson, J.G., 2008.Globalization and the great divergence: terms of trade boom, volatility and the poor periphery, 1782-1913.

European Review of Economic History 12 (3), 355–391 . Williamson, J.G., 2011.Trade and Poverty. When the Third World Fell Behind. MIT Press, Cambridge, Mass.

Williamson, J.G., 2012.Commodity prices over two centuries: trends, volatility, and impact. Annual Review of Resource Economics 4, 185–206 . Williamson, J.G., 2013.The commodity export, growth, and distribution connection in Southeast Asia 1500–1940. paper presented to the Southeast Asian, Economic Growth Conference (Bangkok).

Wooldridge, J.M., 1995.Score diagnostics for linear models estimated by two stage least squares. In: Maddala, G.S., Phillips, P.C.B., Srinivasan, T.N. (Eds.), Advances in Econometrics and Quantita- tive Economics: Essays in Honor of Professor C. R. Rao.

Blackwell, Oxford, pp. 66–87. 18D. Chilosi, G. Federico / Explorations in Economic History 57 (2015) 1–18

An annotated bibliography of at least five additional resources related to the research topic.***No plagiarism*** It must be less than 30% match on authenticity report.***PLEASE DO NOT ACCEPT THIS ASS

An annotated bibliography of at least five additional resources related to the research topic.No plagiarism It must be less than 30% match on authenticity report.***PLEASE DO NOT ACCEPT THIS ASS