Markets in China and Europe on the Eve of the Industrial RevolutionBy C H. S W K *Why did Western Europe industrialize rst? An in uential view holds that its excep-tionally well-functioning markets supported with a certain set of institutions provided the incentives to make investments needed to industrialize. This paper examines this hypothesis by comparing the actual performance of markets in terms of market inte-gration in Western Europe and China, two regions that were relatively advanced in the preindustrial period, but would start to industrialize about 150 years apart. We nd that the performance of markets in China and Western Europe overall was comparable in the late eighteenth century. Market performance in England was higher than in the Yangzi Delta, and markets in England also performed better than those in continental Western Europe. This suggests strong market performance may be necessary, but it is not suf cient for industrialization. Rather than being a key condition for subsequent growth, improvements in market performance and growth occurred simultaneously. (JEL N13, N15, O47)Recent interpretations of the Industrial Revolu-rates much above zero became sustainable over tion suggest that it started around 1770 in Brit-the long run. How did a world of static expan-ain and spread to United States and Continental sion make its way to one of sustained increase 1Western Europe by the mid-nineteenth century. in GDP per capita, and why did the Industrial The turning point of modern economic growth Revolution start its spread from Western Europe was a remarkable event, because it was the rst and not elsewhere—in particular, China?time in human history that per capita growth One in uential view on the source of eco-nomic growth places a great deal of emphasis on how European allocative institutions are both * Shiue: Department of Economics, University of Colo-necessary and suf cient conditions for modern rado, Boulder, CO 80309 (e-mail: shiue@); growth. According to Douglass C. North and Keller: Department of Economics, University of Colorado, Boulder, CO 80309 (e-mail: kellerw@). We others, by 1700 Britain and the Netherlands had have bene ted from comments and discussions with Derek developed exceptionally well-functioning mar-Byerlee, Gregory Clark, Brad DeLong, Avinash Dixit, kets supported with a set of institutions—non-Michael Edelstein, Robert Evenson, Marcel Fafchamps, distortionary pricing systems, common law, Avner Greif, Peter Hansen, Jean-Laurent Rosenthal, and property rights—that would lead to more Vernon Ruttan, Max Stinchcombe, and David Weil, as well as many seminar participants. The comments by two refer-ef cient resource use and provide far greater ees of this journal, as well as the editors, were particularly incentives to make investments that would raise useful. We also thank Jörg Baten for the European weather income per capita (North and Barry R. Weingast data and Patrick Wild for the maps. Part of the paper was 1989; North 1981; North and Robert Paul written while the rst author was Visiting Scholar at the 2Russell Sage Foundation. We acknowledge generous nan-Thomas 1973). This facilitated the movement cial support from the National Science Foundation under grant SES 2 The year 1770 is often chosen as the starting date of the These Western European institutions have successfully Industrial Revolution because a number of key innovations, taken root in other countries as well in the last two centu-including the spinning jenny (1764), the water frame (1769), ries. The historical record of Europe attests to an expan-the steam engine (1774), and other inventions in the cotton sion of in uence and trade over long-distances (Barry R. industry, as well as iron making, were introduced around Chiswick and Timothy Hatton 2003; Kevin H. O’Rourke this time. See Angus Maddison (1989) on international and Jeffrey G. Williamson 2000). Western European set-income comparisons in the last three centuries. Note that tlers, by bringing with them their heritage of institutions, we distinguish continental Western Europe from Western in some cases created the basis of long-run growth in these Europe, which includes (Abhijit V. Banerjee and Lakshmi Iyer 2005; 1189
1190THE AMERICAN ECONOMIC REVIEWSEPTEMBER 2007of goods across locations and furthered develop-of growth, as these two regions were relatively ments of industrial as of the mid-eighteenth century, and An alternative view on growth, however, is that yet would start industrializing about 150 years these allocative institutions are not necessary and apart. A main goal of the paper is to determine suf cient for modern economic growth. Thus, whether European markets were already out-China may have been as market-oriented as the performing markets in China before the period leading areas in Western Europe, with similarly of its industrialization, as implied by the cur-good institutions of allocative ef ciency. Yet it rently in uential view, or whether differences in did not experience an industrial revolution in the performance of commodity markets are a more eighteenth century, possibly because it lacked, recent example, institutions that support technical Our analysis also sheds new light on a progress (Joel Mokyr 1990; David S. Landes number of speci c questions surrounding 1969), which are different from the allocative the Industrial Revolution. One is whether the institutions supporting market Revolution was a uniquely British Moreover, it is generally held that the state phenomenon, or whether the more advanced has a critical in uence in shaping these institu-countries of Europe, such as France, as well as tions. According to North (1981), the state has China and other non-Western areas of the world two faces. The state and associated institutions were equally plausible contenders (N. F. R. provide the legal framework that enables pri-Crafts 1995; Kenneth Pomeranz 2000; Jack A. vate contracts for economic transactions. At the Goldstone 2002). Comparisons of relative liv-same time, the state is an instrument for trans-ing standards are only starting to emerge, but ferring resources from one group to another. there is general agreement that within China, According to this view, institutions are good if the Yangzi Delta was one of the most developed 3they both support private contracts and provide areas. In a prominent recent study, Pomeranz protection against expropriation from the state maintains that whatever advantages England had and other powerful groups. Interesting recent over the Yangzi Delta, it was not in its markets, cross-country work by Acemoglu and Johnson and that, more generally, “China comes closer (2005) has sought to establish which of these the neoclassical ideal of a market economy than two functions is more critical for growth. Our Europe” (Pomeranz 2000, 17). This paper dis-detailed analysis of a single country, China, will tinguishes England from the rest of Western enable us to see how North’s paradigm can help Europe, and compares it to markets both in the us understand why China remained locked in the Yangzi Delta and in China era when Western Europe took nd that as late as 1780, markets in China Although a fair amount has been written on were comparable to most of those in Western the rise of commerce and internal trade in China Europe. The performance of English markets at during the eighteenth century (., Mark Elvin this time, however, was better than that found in 1973; Dixin Xu and Chengming Wu 2000), a the most advanced parts of Continental Western quantitative assessment of the performance of Europe, as well as in China. Furthermore, mar-markets in China in a comparative context is still ket integration in Continental Europe improved lacking. This paper compares markets in China between 1780 and 1830, and the improvement and Western Europe in terms of spatial market occurred dramatically and suddenly in compari-integration, using cointegration analysis with son to what came before. The nding of differ-data on grain prices from the seventeenth to the ences in market integration between England nineteenth century. It is the rst study, as far as and the Western European continent may prove we know, to provide a comprehensive compari-important, we think, for future research on what son of markets across Europe and most of China drives modern economic the preindustrial era. This is particularly useful A comparison of the underlying mechanisms for investigating the fundamental determinants supporting trade in each region helps give 3 For living standard comparisons, see Robert C. Allen Daron Acemoglu, Simon Johnson, and James A. Robinson et al. (2005), Allen (2002, 2001), and Paul Bairoch (1975, 2001).3–17).
VOL. 97 NO. 4SHIUE AND KELLER: CHINA AND EUROPE AROUND THE INDUSTRIAL REVOLUTION1191 qualitative support to our results. Even if the understanding of the origins of economic overall degree of market integration was com-growth, how it can be sustained, and how it may parable, what can we say, for example, about the be spread further across economies of market regulation in China compared to that in Western Europe, or about differences I. Grain Trade: Its Basis in Terms of Geography, in the transportation systems? Notwithstanding Technology, and Institutionsa number of important contributions (Pomeranz 2000; Philip C. C. Huang 2002; R. Bin Wong A comparison of trade and transport tech-2002), little is known about the relative strengths nology in China and Western Europe suggests and weaknesses of speci c institutions in each that although the regions were geographically country. The conditions and institutions that diverse, there were basic similarities in the affected trade, such as the role of guilds and the means of moving grain and goods across land provision and enforcement of property rights, and sea in the preindustrial era. Geographically, not only differed in China and Western Europe, China’s rice-growing area is located in the south but they also played a major role in these econo-and central parts of the country, and wheat was mies in a broad sense. Thus, by probing deeper grown in northern areas. The main trade routes in this analysis of grain market integration, we were along the Yangzi River and its major tribu-may improve our understanding of what advan-taries, the Grand Canal, the Yellow River, as well tages England really had over China, and also as along the coast. In Western Europe, wheat narrow the set of fundamental drivers that have was harvested throughout the area and widely been previously proposed as being crucial for traded, with waterways being an important economic of transport as well. The rivers Vistula, This paper contributes to recent literature that Oder, and Elbe connected the grain-growing emphasize the bene cial role of markets in allo-areas of Eastern Europe to the Hanseatic ports, cating resources (Jonathan Isham and Daniel while the Danube linked the Black Sea areas to Kaufman 1999), as well as the importance of Central Europe. The major rivers of Western institutions that provide the framework in which Europe, the Rhine, Rhône, Seine, Loire, and markets operate (Robert E. Hall and Charles I. Thames, together with canals, especially in the Jones 1999; Dani Rodrik, Arvind Subramanian, Netherlands, England, France, and Germany, and Francesco Trebbi 2002). Another issue on supported an expanding network of we can bring new evidence to bear is In both continents, only rough estimates whether the sources of growth in Europe might on the scale of the long-distance grain trade have originated from a long process of develop-exist. The total on all major routes in China ment stretching back for many centuries (per-amounted to perhaps million tons annually haps as far back as the year 1000), or emerged in the mid-Qing (Qing Dynasty, 1644–1911) 4rather recently in the last couple of centuries. (Fang Xing et al. 2000, 170). Assuming this Because we follow the same economies over would have been enough to feed 14 million several centuries, this reduces many impor-people (Pomeranz 2000, 34), some 8 percent tant identi cation problems because persistent of national grain consumption was supplied via factors unique to Europe—such as geography, traded grain. The fraction of grain imports var-customs, intellectual and ethical tradition, or ied both across China and over time. For exam-language—are held , in the eighteenth century, the Yangzi Delta A better understanding of changing patterns may have imported, in a typical year, about in market performance around the rst Industrial 25 percent of its rice consumption; Southern Revolution should provide a more complete Zhejiang imported in one year, 1748, more than picture of modern economic growth on many 50 percent of its rice consumption (Han-sheng fronts, as there are few explanations of growth Chuan and Richard A. Kraus 1975, 62). Grain that do not depend on the existence of, or have was exchanged for commodities such as cotton implications for, the performance of commodity markets. More generally, the question of market 4 functioning and the quality of market-supporting As conversion factors, we use a weight of 160 pounds institutions is also important in furthering for 1 shi of rice, and 2,200 pounds to one metric ton.
1192THE AMERICAN ECONOMIC REVIEWSEPTEMBER 2007and cotton fabrics, silk and silk fabrics, tea, and 1800 probably averaged between to miles salt, all of which were also a signi cant share per hour for river transport and up to miles of internal trade. In Western Europe, by com-per hour when it was (mostly) canal traf c (based parison, between the years 1550 and 1800, more on W. T. Jackman 1962, 450, 725), not too differ-than 80 percent of the total long-distance grain ent from the mile per hour round-trip speed trade was the Baltic trade through the Danish for travel on the Yellow River in the vicinity of Sound (Fernand Braudel 1982, 127). At its high Luoyang, in Henan province (Laurence Evans point in the 1640s, an average of about mil-1984, 289). The most common Yangzi junk in lion tons was shipped per year, and close to Hubei had a round-trip speed of up to miles million tons was the maximum amount shipped per hour (G. R. G. Worcester 1971, 380), and the per year on average in the eighteenth century speed of junks in the Yangzi Delta was typically 5(Milja van Tielhof 2002, 49, 61). Chinese long-far higher because they took advantage of the distance grain trade was therefore larger than tidal currents in the lower Yangzi: a junk could its Western European counterpart, perhaps by a travel at miles per hour for up to 300 miles 6factor of (Worcester 1971, 184).The direct evidence on shipping speed for Available information on other transport seagoing vessels is scarce, but what informa-costs is scarce, but it suggests that freight costs tion is available indicates that the speeds were in China and Western Europe were broadly roughly comparable. The round-trip from comparable. In the Baltic trade during the early Southern Fujian to Zhapu in Zhejiang took about eighteenth century, the freight charge may have 8 days in the 1720s, including loading (Chuan been around 40 percent of the price differential and Kraus 1975, 61–62); this is about miles between Danzig and Amsterdam (van Tielhof per hour. For comparison, around the year 1580, 2002, 217). For China, estimates indicate that Dutch merchant vessels traveled from Danzig to the real costs of transport were on average 25 Amsterdam in 11 days or more (about miles percent of the grain shipped, and as much as 50 per hour), while the trip from Danzig to London percent for more involved transports, such as could take as little as 9 days (about miles per that from the Yangzi Delta to Beijing through hour) (van Tielhof 2002, 158–59).the Grand Canal (Evans 1984, 298–99).The speed of river transport in England was What does a comparison of the relative costs comparable to that in the upper reaches and trib-of different modes of transport—by sea, by utaries of the Yangzi and Yellow Rivers, while inland waterway, and overland—for a given speeds on the Lower Yangzi, and especially in weight and distance yield? Land transport in the delta, were likely higher. Average (round-late eighteenth century England was normally trip) inland waterway speeds in England around two to four times as expensive as waterway transport (Jackman 1962, appendix 8); in China, land transport was between to times as 5 expensive as waterway transport, depending on Here we use the following conversion factors: one last 7the ease of navigation of the waterway. Inland 5 hektoliter (van Tielhof 2002, 7), 100 liters of wheat 5 79 kilograms, and 1 metric ton 5 1,000 transport in China may have been 6 It is also possible to compare the size of the water-about times as expensive as transport by sea, way networks in China and Western Europe. Eight major while the corresponding factor in eighteenth European rivers—the Danube, Elbe, Rhine, Oder, Seine, century England was between 2 to (Evans Rhône, Loire, and Thames—together have a drainage area of million square kilometers, which is less than 1984, 294, Petersen 1995, 150, respectively). the Yangzi River alone ( million square kilometers). In terms of navigability, the Yangzi was far more important than the Yellow River, due to relatively high levels of silt 7 in the latter. The size of China’s inland waterway system For the well-navigable Yangzi River, overland trans-(rivers and canals) in the mid-Qing is estimated to be about port was between 3 and times as expensive as waterway 50,000 kilometers (Xing et al. 2000, 167), whereas the transport. Peking road transport was times the cost on inland navigation system of England and Wales in 1780 was the Yangzi (Evans 1984, 293); and land transport in Hubei less than 3,500 kilometers long (Christian Petersen 1995, and Shanxi is 3 to 5 times as expensive as transport on the 153). If one abstracts from the large areas in China’s west Yangzi (Perkins 1969, 120). For the less navigable Yellow where few people lived, the density of the waterway net-River and Grand Canal, overland transport may have been work was likely higher in China than that of times as costly (Evans 1984, 294).
VOL. 97 NO. 4SHIUE AND KELLER: CHINA AND EUROPE AROUND THE INDUSTRIAL REVOLUTION1193Using the mid-point estimates and normal-and conversely, private written contracts were izing the cost of sea transport to one, our esti-enforced in the ruling of the Qing courts when mates for the relative costs of sea versus inland disputes arose. The emphasis of the Qing Code, waterway versus overland transport in China however, was on maintaining public order are 1::, while the analogous gures for by providing incentives for lawful behavior England are 1::. Thus, while the scale of through the threat of punishment rather than on China’s long-distance grain trade was larger reconciling con icts among private economic than Western Europe’s, overland transport may been somewhat more expensive relative to Although formal Qing laws took a relatively water transport in China compared to Western laissez-faire approach to the daily affairs of trade, Europe. Overall, these gures suggest that the the state levied domestic customs as well as tran-ef cacy of transport technologies did not differ sit taxes. With the notable exception of England, too much between China and Western Europe in however, the amount of revenue from these taxes the eighteenth China was typically below that of countries in Both public and private institutions supported Western Europe, even as late as the eighteenth the grain trade in China. Below, we summarize century. The Chinese state also responded to some of the more notable ones for China, which perceived monopolization and collusive behav-8may be less familiar to most readers. First, ior in the marketplace by instituting a brokerage the Qing state was a direct participant in the system wherein government-licensed brokers grain trade: about 15 percent of China’s long-earned commission for supervising payments distance trade in the mid-Qing may have been between buyers and sellers, overseeing delivery, of cial government shipments—primarily in inspecting for quality and quantity, and serving the form of tribute grain to Beijing and food for as a guarantor on the exchange (Susan Mann soldiers (Xing et al. 2000, 180). The state also 1987, 63–65). While it was illegal to conduct in uenced the grain trade indirectly, by gath-any wholesale transaction without a licensed ering information about agricultural practices, broker, the requirement was not fully enforced harvest outcomes, and grain prices throughout and the number of unlicensed brokers prolifer-the empire. The Qing state supported the grain ated, weakening the brokerage by creating and maintaining transport Second, nonof cial institutions in China were routes, and by organizing local communities developed and enforced by guilds, self-govern-in the upkeep of transport routes. Some 10 per-ing organizations that were permitted a broad cent of its total revenues were devoted to public range of discretionary powers by the govern-projects, which included ood control and pas-ment. Unlike European guilds, Chinese guilds sageability of major routes. (Susan Naquin and were dominated by interregional merchant net-Evelyn S. Rawski 1987, 23; Ramon H. Myers works and typically did not keep outside arriv-and Yen-Chien Wang 2002, 597). Finally, the ing merchants from guild membership (William Qing Code, the formal legal framework of Qing T. Rowe 1984, 297). Did the Chinese guilds help China, protected private property and hence or hinder trade? The guilds’ role in providing trade through its articles against theft, sale of information as well as an institutional frame-property belonging to others, and for contract enforcement likely supported It also appears that a reasonably effective trade in China. Guilds provided lodging and court system existed and of cial arbitration services for merchants, helped them to calcu-of property disputes was available, even to the late pro ts and losses, and taught members bar-poor. There does not appear to have been much gaining techniques. Often, they also chose and of a contradiction between customary laws enforced the local weights and measurements and the of cial legal framework. Customary for transactions, and established the dates at practice was judged within the of cial laws, which markets would be open, as well as regula-tions for sales, deliveries, and market conduct (Xing et al. 2000, 180–83; Rowe 1984, 295–96). 8 Additional details and comparisons with Europe can Over time, the state may have relied more and be found in the Web Appendix ( on the guilds for market oversight. In some sept07/), as well as in Shiue and Keller (2006).cases, guilds were eventually delegated the
1194THE AMERICAN ECONOMIC REVIEWSEPTEMBER 2007unusual privilege of assessing and collecting million people lived in these ten provinces of trade taxes for the state (Mann 1987, 23–24).China at the end of the late eighteenth century Third, merchant networks, typically identi-(about 60 percent of China), while the popula- ed by kinship or common place of origin, were tion of Europe excluding the former USSR might another major feature of trade in China. These have been 120 to 150 million (John Durand networks were known for their trade in a par-1960; Colin McEvedy and Richard Jones 1978). ticular commodity (such as salt or paper, with To put this in another perspective, a typical grain and other commodities often shipped on Chinese province has roughly the same popula-the return trip), and were an important means by tion as the average European country. The two which geographically dispersed groups shared sample areas are also comparable in terms of information about conditions in distant markets. geographic size. Map 1 depicts the areas on the Contract enforcement among network members same scale; the ten Chinese sample provinces may have been facilitated by reducing commit-are shaded. Trade between the fertile agricul-ment problems. Merchant networks also estab-tural areas in the upper reaches of the Yangzi lished an interlocked chain of banks along their River and the urban regions of Shanghai at the trade routes, complementing the of cial banking Yangzi Delta involved covering distances of at developments in Qing China. Private and of -least 1,200 kilometers, approximately the dis-cial distinctions were at times heavily blurred. tance of the trade route between Antwerp and For example, merchants purchased of cial titles Vienna. The maximum distance between any or became of cials of the state, and of cials not two markets in our sample is about 1,400 kilo-only accepted but regularly solicited nancial meters for Europe and about 1,850 kilometers contributions from merchants for government for (Ping-Ti Ho 1964, 82).The data coverage for China and Europe dif-Although some of the largest fortunes in Qing fers in some respects. As Figure 1 indicates, times were built on domestic and international Europe is relatively well represented in the tem-trade, this trade clearly functioned in an insti-poral dimension, and data are available for both tutional setting that was different from those the pre– and the post–Industrial Revolution era. prevailing in Western Europe. A priori, there While most of the data on China are for a shorter does not appear to be strong reasons to expect period—the 54 years from 1742 to 1795—the a particular institutional framework to be more geographical coverage of the Chinese data is conducive to trade than broad; it consists of all 121 prefectural We now proceed to our quantitative of the ten provinces shown in Map 1, including both the most commercialized as II. Data, Econometric Methods, and Resultswell as the relatively less developed regions. In contrast, the European price data up to the A. Dataeighteenth century tend to be for relatively large and important markets. To the extent that there We have assembled a large dataset of about is selection bias coming from large market size, 250 price series, roughly equally divided between it is likely to favor Europe over and Western Europe, and ranging in time Some major characteristics of our price data period from the fteenth to the twentieth century. are as follows. Except for one of the London Within Europe, the markets are predominantly series, the sources give the market price for located in today’s Belgium, England, France, grain; the less informative data, such as prices Germany, Luxemburg, and the Netherlands, that paid by hospitals, charities, and other entities, is, in Northwest Europe. This area is generally were not used. Prices have been converted to considered to have been the most advanced part common units of currency per volume (or weight) of Europe. The Chinese markets are located in within a series. In general, we have not tried the following ten central/south-central provinces: to account for changes in the value of the cur-Anhwei, Fujian, Guangdong, Guangxi, Guizhou, rencies over time because missing information Hubei, Hunan, Jiangsu, Jiangxi, and the (typically) silver content of coins would The Chinese and European areas are com-necessarily lead to low-quality estimates. We do parable in terms of population size. About 120 not expect that in ation has a major in uence
VOL. 97 NO. 4SHIUE AND KELLER: CHINA AND EUROPE AROUND THE INDUSTRIAL REVOLUTION1195$IJOB&VSPQF
,JMPNFUFSTM 1on our results, because even though the overall commonalities in their method of recording sample period is from the seventeenth to nine-prices permits us to analyze the impact of differ-teenth century, our comparisons between China ences in data characteristics. In addition, some and Europe are typically based on periods of sources contain speci c information that allows 9only about 25 to gauge the in uence of quantity-weighing Most of the price observations are at a monthly (by using weekly prices together with quanti-frequency; however, in some series the price ties sold), nominal versus constant prices (by given is for the rst market day of the month, using the latter instead of the former), and spa-while for other series it is the average for all mar-tial aggregation (by size-weighted aggregation of ket days of the month, where this average may several nearby markets). Experimentation with or may not be quantity weighted. In addition, in these alternative price series showed that our some cases, prices pertain to spatially aggregated main results below are not driven by such dif-regions, not to a market (or several markets) in a ferences in data characteristics. The Appendix given city. Generally, the fact that prices were provides additional details, as well as the sources collected in only a few different ways and that and construction for each of the price can typically nd several markets sharing Table 1 shows summary statistics for certain key samples. For China, we rst show the full sample with 121 prefectural markets. As noted 9 earlier, these prefectures differ substantially Grain prices were broadly trending upward in both China and Western Europe, and it is not generally the case in terms of commercialization and geography. that in ation in one or the other continent was higher (see To take advantage of the breadth of this data, Wang 1992 on in ation in China). We have also experi-a number of subsamples were formed. The rst mented with adding a trend to the cointegrating relationship subsample listed is the set of prefectures that (see equation (5) below); however, tests generally reject this speci cation. belong to the expanded Yangzi Delta. This area
1196THE AMERICAN ECONOMIC REVIEWSEPTEMBER 2007 #BWBSJBO $JUJFT 'SFODI 3FHJPOT $IJOFTF 1SFGFDUVSFT $IJOFTF 1SFGFDUVSFT-POEPO 4JFOOBJ5PVMPVTF-VYFNCPVSHD6USFDIU 'SFODI .BSLFUT97JFOOB-VYFNCPVSH#SVTTFMT D .FDLMFOCVSH $JUJFT"OUXFSQ\4DIXFSJO /JKNFHFO/JKNFHFOJ1BSJT $PMPHOF M"OUXFSQ9# 'SFODI .BSLFUT4DIXFSJO D$PMPHOF\-POEPO .FDLMFOCVSH $JUJFT $IJOFTF 1SFGFDUVSFT7JFOOB 'SFODI 3FHJPOT3VSFNPOEF #BWBSJBO $JUJFTJ3VSFNPOEF6USFDIU $IJOFTF 1SFGFDUVSFT5PVMPVTF 4JFOB F 1. D A
, S L is an important one for our comparison because of countries in Northwest Europe had developed the Yangzi Delta was very likely the most a set of exceptionally advantageous institu-advanced area of China. The second Chinese tions (North and Thomas 1973). Moreover, this subsample consists of the prefectures that are sample will also be analyzed to study market the ten provincial capitals in our sample. These integration in the pre–Industrial Revolution era. markets, which are shown in Map 2 in the Web The second sample covers 15 cities in the later Appendix, were important from a political point part of the eighteenth century. This sample is of view, but were not necessarily on well-linked contemporaneous to our Chinese data, and it is trade routes. We also distinguish 34 prefectures therefore of key interest; the markets are shown that are located on the Yangzi River, directly in Map 3 in the Web Appendix. A third sample or linked through a major tributary. All avail-covers 15 European cities in the second quar-able evidence suggests that these regions traded ter of the nineteenth century (years 1825–1849). more than the average region in is the sample on which the post–Industrial 10For Western Europe, Table 1 shows a sample Revolution analysis will be based. We also of nine cities around the year 1700 (1692 to 1716). This sample is useful for at least two rea-10 We cannot analyze the same set of cities before, dur-sons. For one, the year 1700 is speci cally men-ing, and after the eighteenth century due to lack of data tioned by North as the time by which a number availability, but the overlap in the three samples is high (see m mmmmmm mmmmm mmmmm mmmmm mmmmmm mmmmm mmmmm mmmmm mmmmm mmmmmm mmmmm mmmmm mmmmm mmmmmm mmmmm mmmmm mmmmm mmmmmm mmmmm mmmmm mmmmm mmmmm mmmmmm mmmmm mmmmm mmmmm mmmmmm mmmmm mmmmm mmmmm mmmmm
VOL. 97 NO. 4SHIUE AND KELLER: CHINA AND EUROPE AROUND THE INDUSTRIAL REVOLUTION1197T 1—S
S *(I)#(II)#(III)#Frequency distribution of bilateral distance (km) (in percent)Residual rst Standard difference 300 , x 600 , x x . ChinaYearsndeviationvolatility^Skewnessx , 300, 600, 900900All prefectural markets**1742– Delta (expanded)**1742– capitals**1742– River prefectures**1742– EuropeWest European cities (1)1692– European cities (2)1770– European cities (3)1825– (4)1770–* Based on monthly (log) prices, two observations per year (March and September for Europe, second and eighth lunar month in China).^ Residual rst-difference volatility de ned as standard deviation of residual from a regression of rst-differences on month dummies.# Averages across n markets.** Details on the names and locations of the Chinese markets are available from the authors upon request.(1) Brussels, Cologne, London, Munich, Nijmegen, Ruremonde, Siena, Toulouse, Vienna.(2) Aalst, Antwerp, Boizenburg, Brussels, Cologne, London, Luxembourge, Munich, Nijmegen, Rostock, Ruremonde, Schwerin, Toulouse, Utrecht, Vienna.(3) Augsburg, Boizenburg, Brugge, Brussels, Evreux, Lindau, Munich, Nantes, Nijmegen, Nurnberg, Rostock, Rouen, Schwerin, Toulouse, Vienna.(4) All 41 English markets covered in the London a within-country sample: 41 county-closest to that of the Yangzi Delta according to level markets in England for the years 1770–these frequency . This allows us to study whether market The left side of Table 1 shows various mea-integration in England, the host country of the sures of price variability. The summary statis- rst Industrial Revolution, appears to have been tics show that price variability in China tends different from that in other regions of Western to be lower than in Europe. This is true both Europe in the late eighteenth terms of price levels (column I) and of price Because the extent of market integration and changes (column II). Previous studies have spatial price variability depends on geographic often used such price volatility measures as distance, the frequency distributions of bilateral an indication of how segmented one market is distance between all pairs in a given sample are from other markets, on the rationale that vola-shown on the right in Table 1. For example, tility is lower if markets are more connected percent of all bilateral pairs in the “all prefec-through trade (., D. N. McCloskey and John tural markets” sample in China are more than Nash 1984). Low price variability may also be 900 kilometers apart from each other, whereas caused by other factors, such as by relatively for the West European city sample for 1770–low storage costs. One indicator of that is an 1794, percent of bilateral pairs fall into asymmetric price distribution, which is driven that category. This breakdown gives additional by the fact that storage cannot be negative (., information on the relative sizes of areas com-Angus Deaton and Guy Laroque 1992). Here, pared; speci cally, the sample for England is the skewness measures range from for English markets to for the Yangzi River markets; overall, there is no clear differ-ence between markets in China and in Western the notes at the bottom of Table 1). Small changes to the sample composition do not affect our main in this respect (column III).
1198THE AMERICAN ECONOMIC REVIEWSEPTEMBER 2007B. Econometric Methodswith u 5 be. While (4) typically cannot be t11t11rejected using standard time series tests—as is This paper studies the performance of mar-the case below—the random walk hypothesis kets in China and Western Europe by compar-is not fully satisfactory. For instance, multiyear ing the spatial integration of grain markets in damage due to weather shocks is more plausi-the two regions. In this section, we present a ble for tree crops than for annuals such as rice model of an agricultural commodity, and show and wheat (Deaton 1999). We adopt it here as how testing for cointegration in prices can be the point of departure for our analysis of coin-used in this , noting that there is no consensus yet Consider the following simpli ed model. In on a model of agricultural price behavior that each period t, t 5 1, … , T, there is an inelastically is clearly preferred to it in terms of accounting supplied harvest z which follows a stochastic for both short-term and long-term dynamics (see tprocess characterized by a cumulative distribu-Deaton 1999, Deaton and Laroque 1992, 1996).tion function F(z, Z),Suppose that there are two markets, with prices p and p, satisfying equation (4). If a 1t2tlinear combination of these nonstationary vari-(1) F(z, Z) 5 Pr(z # Z| z 5 z).t11tables is stationary, the prices are said to be coin-tegrated (Clive W. J. Granger 1981; Robert F. The harvest shocks are exogenous, deter-Engle and Granger 1987). Cointegration means mined by the conditions of agricultural produc-that there exists a long-run relationship between tion, and are what ultimately determines the prices in the two markets, and prices cannot behavior of the price, P. The economy is popu-move arbitrarily far away from each other. This tlated with nal consumers who have identical is consistent with arbitrage through trade estab-inverse demand functions. Under the assumption lishing a link between markets. We thus use that the commodity cannot be stored—storage tests for cointegration to provide evidence on is discussed later, in Section IV—the harvest z the degree of market consumed in each period. With log-linear Cointegration generally supports the notion demand the price is given bythat trade and the forces of arbitrage are at work. The strength of these forces is determined by a number of factors. First, trade is reduced by high (2) p5 a 1 bz,t t transport costs, and, in general, transport costs are increasing in geographic distance. Second, where a and b , 0 are parameters, and p 5 because transporting grain over water is cheaper t1n(P). Because consumers are the only buyers than over land, transport costs also re ect an tin the market, given equation (2), the behavior area’s topography (including waterways and of prices follows directly from the behavior of mountains) as well as route maintenance and the harvests. We assume that the harvest process the sensitivity to weather-related problems (such followsas mud on streets, or the drying out of rivers and canals). Third, the degree of market integration (3) z 5 rz 1 e,t 1 1tt 1 1is also affected by the quality of institutions. It matters, for example, whether the political sys-where 21 , r # 1, and the e’s are . with tem is uni ed (as in China), or more fragmented mean . From equation (3), harvests and prices (as in Western Europe), because often govern-are . for the case of r equal to zero. The ments impose tariffs, quotas, and other trade case of positive autocorrelation may emerge if barriers at borders. Even in the absence of trade weather shocks damage crops for several periods. barriers, a uni ed system may reduce transac-In the extreme case of r 5 1, such damages (or tions costs if currencies, weights and measures, improvements) have a permanent effect. This is or languages are more standardized than in a what we assume here. The process of harvests is politically fragmented region. Further, there are nonstationary, and prices follow a random walk,differences in government support for institu-tions that have an impact on trade, in particular (4) p 5 p 1 u,property rights, contract enforcement, the rule t11tt11
VOL. 97 NO. 4SHIUE AND KELLER: CHINA AND EUROPE AROUND THE INDUSTRIAL REVOLUTION1199of law, and the provision of security for trade. C. Empirical ResultsGovernments can also be more or less prone to yielding in uence to lobbies and interest groups, This section begins by showing relatively which, in line with private incentives, often favor descriptive evidence from bilateral price cor-restrictions on trade and , which require making fewer assump-To summarize, our comparative study of spa-tions but are similar in spirit to the cointegration tial market integration in China and Western analysis to which we turn next. Figure 2 shows Europe provides information on the overall per-a scatter plot of bilateral price correlations ver-formance of grain markets in the two areas. One sus distance for a number of samples during advantage of the cointegration approach is that the years 1770 to 1794. Transport costs are, at it captures the outcome of many factors, ranging least in part, increasing in the time traveled on a from transport technology over the quality of journey, and as a rst approximation, costs are grain market-supporting institutions to factors captured here by geographic distance. The g-that affect transactions costs, such as contract ure shows the price correlations with con dence for the 15 European markets together We follow Engle and Granger (1987) and esti-with regression lines of price correlations for the mate by OLS the equationYangzi River and provincial capitals samples. At a given distance, the price correlations for (5) p 5 b 1 bp 1 tend to be higher than for the Chinese 1t012ttprovincial capitals but lower than for the Yangzi If p and p are cointegrated, there will be some River markets. Also, note that price correla-1t2tlong-run parameters b and b such that p 2 tions along the Yangzi River fall less as distance 011tb 2 bp 5 0 is satis ed. To test for this, we increases than in the other two samples. This is 012texamine the time series properties of e, because indicative of lower marginal transport costs on tpand p are cointegrated if and only if e is sta-the Yangzi than overland (or a less direct water-1t 2tttionary. An augmented Dickey-Fuller (1979) test way route). The fact that we can bound price on ˆ,e the residual of (5), is employed,correlations in eighteenth century Europe with tthose of samples from contemporaneous China (61 ) Dˆe 5 dˆ,edDˆ1 uet1t21 2t21 is consistent with the idea that the degree of t market integration was comparable in the two where the lagged dependent variable is added as areas at this regressor to reduce problems of serial correla-We now turn to examining market integra-tion. Under the null hypothesis that e is non-tion using tests for cointegration among prices tstationary, the parameter d is equal to zero, and in market pairs. After testing for nonstationar-112the stronger is the evidence that d , 0, the more ity in the individual price series p and p, 11t2tevidence there is that p and p are cointegrated. we estimate the cointegrating relationship given 1t2tBelow we will compute the t-statistics of d for in equation (5), augmented with monthly xed 1various samples in China and Western Europe effects (b):mto compare the evidence for market integration 11in the two regions.(59) p 5 b1 b 1 bp 1 e,1t0 m12ttand then ˆtest e for stationarity with an t11 Extensions of this approach have been considered as Augmented Dickey-Fuller (ADF) regression. well. In Shiue and Keller (2004), we use Johansen’s (1988) ^The lower is the t-statistic of d, the parameter 1maximum likelihood estimation (MLE) to estimate the long-run parameters together with the speed-of-adjustment parameters, and we also use MLE methods to test speci c hypotheses, in particular, the law of one price, b 5 0 and these methods would yield the same qualitative conclusions 0b 5 1. Overall this yields results consistent with what is as our cointegration described below. In addition, we have considered thresh-Typically, the null hypothesis of a unit root cannot be old estimation techniques (Alan M. Taylor 2001; Nathan rejected using standard unit root tests. For example, during S. Balke and Thomas B. Fomby 1997) as well as cross-the years 1770–1794, the average p-value for the null of a sectional spatial correlation techniques (Keller and Shiue unit root was for the 15 European cities, whereas it was forthcoming) to compare market integration; we think that for the expanded Yangzi Delta prefectures.
1200THE AMERICAN ECONOMIC REVIEWSEPTEMBER 2007NNNNNNNNNNNNNNNNNNNNNNN mNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNm NNNNNNNNNNNNNNNNNm N&VSPQFBO $JUJFT$IJOFTF 3JWFS 1SFGFDUVSFT$IJOFTF 1SPWJODJBM $BQJUBMTm ,JMPNFUFSTF 2. C E , S M , 1770–1794estimate on ˆe, the stronger is the evidence for that market integration among European cities t2113cointegration between prices p and greater than in China for short distances (less 12Figure 3 presents the average ADF t-statistics than 150 kilometers). Above that, it appears that for the sample of 15 European countries in the the level of market integration, not only among late eighteenth century as a function of distance the Yangzi River markets, but also among the class, and the implication of these results quali-Chinese prefectural capitals, was at least com- es several conclusions from the price correla-parable to levels of market integration that pre-14tions versus distance plot in Figure 2. Generally, vailed in the Western European lowest average t-statistics are found for the Figure 4 compares the cointegration mea-smallest distances—less than 150 kilometers—sures for the 15 European cities with the Yangzi which is what one would expect: the evidence in Delta region, as well as with the entire sample support of cointegration is strongest among rela-of 121 Chinese markets. Generally, there is tively nearby markets. Figure 3 also presents the more evidence of market integration for the average ADF t-statistics for the Chinese Yangzi extended Yangzi Delta region than for China River and provincial capital samples; it suggests as a whole, consistent with qualitative evidence on the relative degree of commercialization and 13 We also reduce the effect from outliers in equation 14 (5’) by adding indicator variables to the deterministic com-The city of Danzig is not located in Western Europe, ponent for periods with exceptionally strong price changes. but given its importance for the European long-distance We treat a price change that is larger than one standard grain trade (see Section II), we have experimented with deviation as an outlier; some experimentation with other including Danzig in the sample of European cities and de nitions indicates that the main results are not sensi-recalculated the cointegration statistics. Although Danzig tive to that choice. Note that although we have computed has a positive in uence on the average market integration approximate critical values using techniques laid out in levels in Europe, its inclusion does not qualitatively change James G. MacKinnon (1991), the outlier treatment and the our results; the average ADF t-statistics without (with) inclusion of seasonal effects imply that these critical values Danzig are: 0–150 km (), 150–300 km do not strictly apply anymore. For this reason, we prefer to (), 300–450 km (), 450–600 km look for general patterns in the ADF t-statistic as opposed (), 600–750 km (), 750–900 km to testing sharp hypothesis.(), and above 900 km ().1SJDF $PSSFMBUJPOmmmmm
VOL. 97 NO. 4SHIUE AND KELLER: CHINA AND EUROPE AROUND THE INDUSTRIAL REVOLUTION1201 D D DDJM MDJDMMMJ J MJJ J D&VSPQFBO $JUJFT J$IJOB :BOH[J 3JWFS NBSLFUT M$IJOB 1SPWJODJBM $BQJUBM NBSLFUTD m m m m m m %JTUBODF DMBTT LN
F 3. C C E , 1742–1795 development in the Yangzi Delta. Relative to than for Western Europe overall, and, moreover, Western Europe, there is less evidence of mar-the evidence for cointegration in England is also ket integration in the expanded Yangzi Delta for stronger than for the Yangzi Delta up to 150 kilometers, while for larger Thus, while in the late eighteenth century the distances markets appear to be more integrated Yangzi Delta region had similarly or possibly than markets in Western Europe. As the results more integrated markets than Western Europe for all 121 prefectures show, even the average as a whole, England’s markets at the time were Chinese market seems to have been as spatially more integrated still. The gap between England integrated as markets in Western Europe for all and continental Western Europe started to close but the shortest distances. This indicates that the over the next several decades. Figure 5 shows earlier analyses are not driven by selection cointegration statistics for European markets Overall, the results suggest that there were no for the years 1825–1849, in the aftermath of the large difference in terms of market integration Napoleonic Wars (1799–1815). Market integra-between China and Western Europe in the late tion was substantially higher in Western Europe eighteenth century. But what about England? in the early nineteenth century than in the late Figure 5 compares the cointegration statistics eighteenth century, and this improvement led to for eighteenth century markets in England with a level of market integration that was not so dif-those in contemporaneous Western Europe over-ferent from that in Britain during the years 1770 1516all, as well as in China’s Yangzi Delta. The evi-to for cointegration in England is stronger III. Quali cations and Sensitivity Tests15 In this section, we brie y discuss a number of The critical values of the ADF t-statistics are a func-tion of T, the time series length: it becomes easier to nd additional issues and robustness checks that are evidence in favor of cointegration the larger is T, so that the critical values need to be adjusted to hold the signi cance 16 level constant if T is not the same. We have ensured the These cointegration results are also broadly in line comparability of the t-statistics using the response surface with our ndings using bilateral price correlations, as dis-results of MacKinnon (1991).cussed in the Web Appendix."WFSBHF "%' U TUBUJTUJD
1202THE AMERICAN ECONOMIC REVIEWSEPTEMBER 2007 D MD DDJMJJJ JDJJ MM M D&VSPQFBO $JUJFT J$IJOB "MM .BSLFUT M$IJOB :BOH[J %FMUB .BSLFUTD m m m m m m %JTUBODF DMBTT LN
F 4. C C , Y D , E , 1742–1795 &VSPQFBO $JUJFT m J $IJOB :BOH[J %FMUB .BSLFUT m JM&OHMJTI $PVOUZ .BSLFUT m D &VSPQFBO $JUJFT m J JJ J JJDJ MDDMJM M D m m m m %JTUBODF DMBTT LN
F 5. C Y D , E , W E E C
, W E N C
important for our analysis. They have to do with Weather Shocks.—Differences in weather the spatial patterns of weather shocks in China shocks in China and Western Europe may affect and Europe and differences between rice and the market integration results. Speci cally, if wheat, speci cally related to shocks would change quite differently "WFSBHF "%' U TUBUJTUJD"WFSBHF "%' U TUBUJTUJD
VOL. 97 NO. 4SHIUE AND KELLER: CHINA AND EUROPE AROUND THE INDUSTRIAL REVOLUTION1203across geographic space in China compared to is less need for trade if cost-effective storage is 18Europe, this could lead to substantial differences available, and vice spatial price correlations without any implica-tions for market integration and trade. To address Seasonality.—In this paper we study coin-this issue, we compare historical weather data tegration using monthly, and not quarterly or for China and Western Europe, and nd that cli-annual, prices. While this tends to reduce prob-matic differences are not a likely explanation for lems resulting from temporal aggregation (see our results (see the Web Appendix).Taylor 2001), we are likely to pick up seasonal effects for these agricultural goods. To the extent Rice versus Wheat.—Rice markets in China that seasonal effects differ for rice cultivation in and wheat markets in Europe are selected for China and wheat cultivation in Europe, this may comparison because rice and wheat were the confound our grain in each continent, respectively. Comprehensive information on possible dif-Rice and wheat production differ in terms of ferences along these lines is not available for our cultivation and harvesting, and possibly in other markets. As a rst step to deal with differences respects such as transport cost or value-to-in seasonal effects between rice cultivation in weight and wheat cultivation in Europe, monthly The most important agronomic distinction xed effects (b) are added in the determinis-mbetween the technology of wheat production in tic component of the cointegrating equation (5). Western Europe and rice production in China The question of possible differences between is probably that wheat is grown under upland rice in China and wheat in Western Europe has (rain-fed) conditions while rice is grown under also been addressed by a number of auxiliary irrigated conditions. The irrigated environment analyses: comparisons of eighteenth century (a) lends itself to a much more labor-intensive sys-wheat prices in China and Western Europe, (b) tems of cultivation than rain-fed production. In rice and wheat prices in China, and (c) rice and addition, an irrigated system may exhibit lower wheat prices in the United States. These analy-inter-year and spatial price variability than rain-ses suggest that large biases in our market inte-fed production, and this would favor estimating gration comparison due to differences between lower levels of market integration in China than wheat in Europe and rice in China are unlikely, in Western Europe, all else equal. We now turn and to the extent that biases exist, they tend to to other important differences between rice and favor nding relatively high levels of market wheat that may affect our in Europe. A more detailed discus-sion can be found in the Web .—The model above does not include With regards to storage in Europe, there storage. The availability of storage possibilities, is generally little direct evidence of storage however, is important because storage leads in either private or public facilities (Randall to serially autocorrelated prices even if the Nielsen 1997, 12). In contrast, the Qing state harvest shocks are . (Deaton and Laroque operated large-scale granaries (Pierre-Etienne 1996, 1992; Jeffrey C. Williams and Brian D. Will and Wong 1991). The Qing public grana-Wright 1991). Moreover, if storage costs for rice ries, however, were rarely a reliable source of and wheat differ, then even if storage behavior food for the general population, even over the is ef cient and harvest shocks are . in both rather limited period of time that the granaries 19markets, storage will induce differences in were operational. Nevertheless, the institution 17the time series properties of the two prices. Storage cost differences matter also because 18 Williams and Wright (1991, chap. 9) show that allow-trade and storage are substitutes for achieving ing for storage in a model of trade leads to fewer periods consumption smoothing. All else equal, there in which trade takes place in equilibrium, to lower trade volumes, and to a lower correlation of prices than there was in the absence of storage. Evidence consistent with this for eighteenth century China is presented in Shiue (2002).17 19 Deaton and Laroque (1992, 6–7) show that the critical This is consistent with estimates that in China, the *price level p at which storage turns positive is a function amount of grain held in private storage exceeded that held of the storage public granaries (Francesca Bray 1984, 416).
1204THE AMERICAN ECONOMIC REVIEWSEPTEMBER 2007 J)JHI TUPSBHF (VJ[IPV (VBOHYJ M$IJOB "MM QSFGFDUVSFT D&VSPQF DJUJFT 9:BOH[J %FMUBM JJJ MDD M 9M9M 9M D Y Y %JTUBODF DMBTT LN
F 6. S C is one that countries of Europe did not support, would be biased in favor of nding relatively and it is worth employing data on public grain high levels of market integration in Europe. storage to obtain an estimate of the role that this Therefore, our result that market integration in storage might play for our comparison of market China and Europe was comparable tends toward integration. In addition, a second reason makes the conservative. Controlling for storage cost the quantitative comparison valuable. Since the differences, China’s level of market integration state was responsive to overall market trends may have been even higher than in continental (Shiue 2005a, 2004), public storage, even if it Western Europe, although judging from the alone did not exert a decisive effect on the level magnitudes in Figure 6, it seems unlikely that is or variability of grain prices, could well provide was as high as that prevailing in proxy for total (public and private) grain stor-To sum up the results of this section, it is age in that our analysis is biased toward nd-In Figure 6, we compare the evidence for ing relatively high levels of market integration cointegration in Chinese provinces that had rel-in China, and, if anything, there may be a small atively high levels of public storage with other bias in the opposite direction. This con rms our Chinese provinces, as well as with European main nding that, while in the late eighteenth 20markets. As expected, there is less evidence century England’s markets were more integrated for cointegration in markets located in prov-than those of the Yangzi Delta, China’s markets inces with high levels of storage compared to had similarly or possibly more integrated mar-the average Chinese market or the Yangzi Delta kets than Western Europe as a . If storage costs for rice in China were indeed low relative to those for wheat in Europe, IV. Summary and Concluding Discussionas suggested by Bray (1984, 385), our analysis The concept of allocative market ef ciency holds a prominent place in our understanding of 20 The two provinces with the highest level of per capita economic growth, and it has also been offered storage in our sample are Guizhou and Guangxi; see Shiue (2005a, g. 1).as a leading explanation of why Western Europe "WFSBHF "%' U TUBUJTUJD
VOL. 97 NO. 4SHIUE AND KELLER: CHINA AND EUROPE AROUND THE INDUSTRIAL REVOLUTION1205industrialized in the late eighteenth century. In high level of market integration the reason why this paper we examine the similarities and dif-Britain industrialized? In China, internal com-ferences in markets in preindustrial China and mercial taxes until the mid-eighteenth century Western Europe from the fteenth to nineteenth were probably not higher than in Britain, and in centuries. We use a range of descriptive and addition, Chinese merchant networks did much more formal methods to assess market inte-to facilitate trade, especially over long distances. gration in China and Western Europe around The British government appears to have had 1770 to see whether one continent—speci cally more public provisions for road maintenance Western Europe—was clearly ahead. According and transport, but Chinese merchants and gen-to the evidence presented in this paper, as of the try also made sometimes substantial nancial period right before the Industrial Revolution contributions for a variety of public goods. We took place in Western Europe, grain markets did take the fact that the level of market integration not perform uniformly better in Western Europe seen in continental Europe by the mid-nine-than in China. Over relatively short distances teenth century is similar to Britain’s in the late of 150 kilometers or less, there are indications eighteenth century as evidence that important that European markets were more integrated. economy-wide change had indeed occurred in This edge, however, is relatively small when Britain by around consider what occurs right after the onset What about the role of the Qing state and the of industrialization. In the early nineteenth emergence of good institutions in China? As century, soon after the dates associated with mentioned earlier in our discussion of the insti-industrialization, markets in continental Europe tutional foundations of the Chinese grain trade became, rather quickly, signi cantly more inte-in Section II (with more details given in the Web grated than in centuries prior. That the bulk of Appendix), contracting institutions (such as laws the market’s improvement took place only once and courts) were in part provided by the Chinese modern growth had started indicates that most state, and in part by guilds, merchant networks, of the improvements in the degree of market and customary laws. While there were no organi-integration in the nineteenth century in Western zations or institutions that could effectively chal-Europe may be largely a consequence of indus-lenge the power of the Qing state, we do not have trialization rather than a evidence that the state had a strongly expro-Markets today in the United States and priative or redistributive character, by European Western Europe are yet more integrated than standards. That is, the constraints imposed by the they were in the nineteenth century. They are state on economic incentives, which were in turn also more integrated than markets found in poor safeguarded by the development of “good” insti-countries today. But it would in part confound tutions in Europe, were apparently not binding in cause and effect to conclude from this that inte-China. The dichotomy proposed by North (1981) grated markets must rst be secured before mod-and examined by Acemoglu and Johnson (2005) ern growth can proceed. This, at least, does not thus appears to be too narrow to t the Chinese appear to be what happened in Western Europe case. Based on our research, we hypothesize that before the onset of the Industrial the failure of China to industrialize had at all That the overall level of market integration to do with shortcomings of the state, it was not so was not exceptional stems in part from the much that the state suppressed private economic political fragmentation of Europe. A uni ed activity, but that the state did too little to support it Europe would likely have seen higher integra-through the provision of public goods and formal tion across countries. The lack of free markets, legal institutions. Clearly more research needs however, even within most European countries, to be done on this needs to be considered. England was an Our discussion also highlights that it would exception in this regard, as internal markets be dif cult to argue that the overall comparabil-were relatively free, and consistent with that, ity of market integration in China and Western we nd market integration levels in England to Europe came about because China had excellent be higher than anywhere else in the late eigh-natural endowments in its rivers while Western teenth century. That Britain was clearly ahead Europe achieved the same through generally is an important nding. But was the relatively stronger allocative institutions and organization
1206THE AMERICAN ECONOMIC REVIEWSEPTEMBER 2007in all other respects. In both China and Western these institutions, it appears that the effects of Europe, the institutions governing grain trade institutions on growth cannot be unidimension-were often not speci c to grain, but also affected ally ranked with any ease. The ndings in this trade in other commodities and possibly other paper provide new evidence on whether trade endeavors, including the incentives for techni-causes growth. Future research is necessary to cal change. The in uence of merchant guilds, improve our understanding of what factors facil-for example, in the area of standardization, itate market integration, what factors can trigger property rights, judiciary, and security were industrialization, and how the two sets of factors wide-ranging. Because of the complexity of are : G C P S Quality: Prices in China are for mid-quality rice; for Europe we do not have much speci c infor-mation on quality. The sources indicate that the quality of grains sold varied, and the quoted price in Europe should be close to that for commonly available average quality of : In Europe, the prices come typically from mercuriales, which are of cial records of trans-actions at public markets. In China, the data come from market price reports at the prefectural level (an administrative level above the county level and below the province level).Temporal and spatial aggregation: The price series vary in terms of temporal as well as spatial aggregation. Most prices are at a monthly frequency. The calculation of the monthly price varies, but the three most important methods are (1) average price of rst market day of month, (2) average price of all transactions in a month, and (3) minimum and maximum during a month. The price averages are typically not quantity-weighted; however, we have used prices and quantities available at a weekly frequency for some markets to compute the correct average prices, and found that the difference is negligible. If the minimum and maximum prices are available, we form the average price as (min 1 max)/ is spatial aggregation to a varying extent. We explored the effects of spatial aggregation and did not nd major effects. Jesus Gonzalo (1993) discusses the extent to which time series proper-ties—order of integration, cointegration—are preserved under temporal and cross-sectional (spatial) of markets: The sample for China includes virtually all markets in the ten provinces for which rice was the major grain, which suggests that sample selection plays essentially no role for China. For Europe, our data tend to be for markets of relatively big cities that were often also centers of trade and of relatively rich cities. For instance, the 50 French markets in Sylvie Drame et al. (1991) are selected from some 900 markets where the greatest quantities were data: There are substantial gaps in the sources; approximately 23 percent (15 percent) is missing in an average Chinese (European) series. We have used the TRAMO (Time Series Regression with ARIMA Noise, Missing Observations and Outliers) program to interpolate series for which there were not too many missing data (Victor Gomez and Agustin Maravall 1997). This estimation of data does not critically affect our c Information and Sources of the Price SeriesAustria—ViennaSource: Pribram, Alfred Francis. Materialien zur Geschichte der Preise und Löhne in Österreich, Volume I. Vienna: Carl Überreuters Verlag, years: 1692 to 1914; frequency: monthly; method: quantity-weighted average of all market units: In “Wiener Metzen” (1692–1752), in “Niederösterreichen Landmetzen” (1752–1875), and in “100 Kilogram” (1875–1914). Conversion rates: 1 Wiener Metzen 5 Niederösterreiche Landmetzen; 1 Niederösterreicher Landmetzen 5 Kilogram.
VOL. 97 NO. 4SHIUE AND KELLER: CHINA AND EUROPE AROUND THE INDUSTRIAL REVOLUTION1207Monetary units: “Kreuzer” (1692–1752); “Kreuzer Konventionsmünze” (1752–1812); “Kreuzer Wiener Währung” (1812–1858); “Kreuzer Österreichischer Währung” (1858–1897); “Heller Kronenwährung” (1898–1914). Conversion rates: 1 Gulden 5 60 Kreuzer 5 60 Kreuzer Konventionsmünze 5 150 Kreuzer Wiener Währung 5 105 Kreuzer Wiener Währung 5 210 Heller Kronenwä source: “Marktprotokolle der Stadt Wien.”Belgium—BrusselsSource: For years 1568–1696, 1728–1795: Craeybeckx, J., “De Prijzen van Graan en van Brood te Brussel,” in C. Verlinden (ed.) Dokumenten voor de Geschiedenis van Prijzen en Lonen in Vlaanderen en Brabant, Vol. I. Bruges: De Tempel, : For years 1800–1889: Vandenbroeke, C., “Brusselse Merkuriale van Granen, Aardappelen, Hooi, Stro, Boter, Vlees, Koolzaad, Boskool en Steenkol,” in C. Verlinden (ed.) Dokumenten voor de Geschiedenis van Prijzen en Lonen in Vlaanderen en Brabant, Vol. III. Bruges: De Tempel, years: 1568–1889, with gaps; frequency: monthly; method: average from the month’s rst market units: In “Bruxelles setier” (1568–1696, 1728–1795); in “100 liters” (1800–1871), and in “100 Kilograms” (1872–1889). Conversion rates: 1 Bruxelles setier 5 liters, and 79 kilograms 5 100 units: In “Brabantse stuivers” (1568–1696, 1728–1795); in “French Francs” (1800–1817, 1833–1889), and in “Dutch Guilders” (1817–1832). Conversion rates: 20 Brabantse stuivers 5 1 Dutch Guilder; and in 1816, 1 Dutch Guilder 5 French Francs, and in 1832, 1 Dutch Guilder 5 French source: “Algemeen Rijksarchief te Brussel, Terminatieboeken.”Belgium—BruggeSource: Vanderpijpen, W., “Brugse Merkuriale van Granen, Brood, Aardappelen, Boter en Vlees (1796–1914),” in C. Verlinden (ed.), Dokumenten voor de Geschiedenis van Prijzen en Lonen in Vlaanderen en Brabant, Vol. IV. Bruges: De Tempel, years: 1796 to 1914, with few gaps; frequency: monthly; method: average from the month’s rst market units: In “10 Kilograms” (1796–1802), in “100 liters” (1802–1890), and in “100 kilo-grams” (1892–1914). Conversion: 100 liter 5 79 units: In “French Francs” (1796–1817, 1833–1914), and in “Dutch Guilders” (1817–1832); see notes on Belgium– sources: “Rijksarchief Brugge (.), Leiedepartement, 1076–1080 en 2831–2846” (for 1796–1812), and “Stadsarchief Brugge (.), Gazette van Brugge” (for 1813–1914).Belgium—AalstSource: Wyffels, A., “Prix du havot de froment à Alost,” in C. Verlinden (ed.), Dokumenten voor de Geschiedenis van Prijzen en Lonen in Vlaanderen en Brabant, XV–XVIII Centuries, Bruges: De Tempel, years: 1729–1802, with major gaps before 1750; frequency: monthly; method: average, with units: In “havot” (., per barrel).Monetary units: In “Gros de Flandre.”Original source: “Alost, Archives communales, n8 261.”Belgium—AntwerpSource: Craeybeckx, J., “De Prijzen van granen en van brood te Antwerpen van 1608 tot 1817,” in C. Verlinden (ed.) Dokumenten voor de Geschiedenis van Prijzen en Lonen in Vlaanderen en Brabant, Vol. I. Bruges: De Tempel, 1959.
1208THE AMERICAN ECONOMIC REVIEWSEPTEMBER 2007Overall years: 1608 to 1817; frequency: monthly; method: average price on rst market day of each units: In Antwerp “Viertel.”Monetary units: In “Brabantse stuivers.”123Original sources: “Vierschaar, Antwerps Stadsarchief,” ; ; ; ; ; ; : Helin, E., Prix des Céréales à Luxembourg aux XVII et XVIII siècles, Université de Louvain, 1966; part of Joseph Ruwet et al. (1966), Marché des céréales à Ruremonde, Luxembourg, Namur et Diest aux XVIIe et XVIIIe siècles, Leuven: Presse de l’Universite de Louvain, years: 1721 to 1794, with gaps; frequency: monthly; method: average of highest and lowest price on rst market day of month, with some units: In “bichet,” where 1 bichet (or setier) 5 units: In “Sous,” where 1 Luxemburg Gulden is equal to 20 source: “Les Hallages de Luxembourg.”China—121 prefectural capitalsSource: Data collection by Carol H. Shiue. These are rice prices; the sample covers virtually all prefectures of ten provinces in China’s Central/South-Central area; see Shiue (2002) as well as Roehner and Shiue (2000) for additional years: 1742 to 1795, with gaps; frequency: monthly (collected and used: second and eighth lunar month); method: highest and lowest price from all markets in a given prefecture. We take (high-est price 1 lowest price)/2 as the average units: In “shi,” where 1 shi 5 about 103 units: In “liang,” which is a Chinese silver currency; it is also called “tael.”Original source: Gongzhong liangjiadan [Grain price lists in the palace archives]. Number One Historical Archives, —Chengdu, Chongqing, Nanchang, and XichangSource: Data provided by Madeleine Zelin, Columbia University; these are prices for rice and for years: 1736–1782, with units: In “shi,” where 1 shi 5 about 103 units: In “liang,” which is a Chinese silver currency; it is also called “tael”; 1 tael silver is about 37 source: Provincial collection of prices in Sichuan that were then reported to the imperial government and published in Gongzhong liangjiadan [Grain price lists in the palace archives], at Number One Historical Archives, —TianjinSource: Data provided by Loren Brandt, University of years: 1739–1794, with units: In “shi,” where 1 shi 5 about 103 units: In “liang.”Original source: Provincial collection of prices in Zhili that were then reported to the imperial government and published in Gongzhong liangjiadan [Grain price lists in the palace archives], at Number One Historical Archives, —LondonSource: Beveridge, William H. B. Prices and Wages in England from the Twelfth to the Nineteenth Cen-tury, Volume 1 (Price Tables: Mercantile Era; rst edition 1939). London: Frank Cass & Co. Ltd., 1965.
VOL. 97 NO. 4SHIUE AND KELLER: CHINA AND EUROPE AROUND THE INDUSTRIAL REVOLUTION1209Overall years: 1683 to 1801, with gaps; frequency: monthly; method: average from prices for spot or future (usually up to one month) are not market, but British Navy procurement, prices for grain. The Navy Victualling Board met daily to contract with dealers for the supply of provisions. There was a public announcement, after which the interested dealers were asked one by one to place their bid. The Navy tried to achieve the competitive market price outcome (for instance, indications of collusion among bidders resulted in the postponement of procurement activity). See pp. 514–35 for units: In “Quarters.”Monetary units: In “Shillings.”England—London and 40 CountiesSource: The London years: 1770 to 1794, with gaps; frequency: we employ data for the months January, March, July, and September (the source reports prices every week); method: average price of rst week of units: In “Standard Winchester Bushel of 8 gallons” (1770–93), and in “Standard Winchester Quarter of 8 bushels” (1793–94).Monetary units: In “Shillings” and “Pence.”France—ParisSource: Baulant, Micheline and Jean Meuvret. Prix des céréales extraits de la mercuriale de Paris. Paris: Ecole des Hautes Etudes, years: 1520 to 1698; frequency: monthly; method: average of maximum and minimum price on the rst market day of units: In “setier de Paris,” where 1 setier 5 156 units: In “Livres turnois.”Original source: “Mercuriale de Paris.”France—ToulouseSource: Bertrand Roehner’s preparation of gures in Frêche, Georges and Geneviève Frêche. Les prix des grains, des vins et des legumes à Toulouse. Paris: Presses Universitaires de France, years: 1486 to 1913; frequency: monthly; method: average of rst market day of the units: In “Hectoliter” (one hectoliter 5 100 liter).Monetary units: In “French Francs and Centimes.”Conversions: The Frêche-Frêche data is in different quantity and monetary units for different peri-ods; we have followed the conversion rates applied in Drame et al. (1991).Original source: “Mercuriale de Toulouse.”France—Alençon, Amiens, Bourges, Bourgogne, Bretagne, Caen, Lyon, Riom, Rouen, ToursSource: Labrousse, . Esquisse du mouvement des prix et des revenus en France au 18e siècle. Paris: Paris C. 1932. See Roehner and Shiue (2000) for additional years: 1756 to 1790, 1806 to 1900; frequency: annually; method: average. If the price is given for a region (. Bretagne), we use the location of the central city of the region. Note that administrative boundaries were partly rede ned during the French units: In “setier de Paris” (1756–1790), and in “Hectoliter” (1806–1900); conversion: one setier de Paris 5 156 units: In “100 livres” (1756–1790), and in “French Centimes”; conversion: 100 Centimes 5 1 Francs 5 1 —50 cities in French departementsSource: Drame, Sylvie, Christian Gonfalone, Judith A. Miller, and Bertrand Roehner. Un Siècle de Commerce du Blé en France, 1825–1913. Paris: Economica, years: 1825 to 1913; frequency: every 15 days (1825–1903), and every month (1903–1913); method: quantity-weighted average from all market days in a given period (15 days, or a month).
1210THE AMERICAN ECONOMIC REVIEWSEPTEMBER 2007Quantity units: In “Hectoliters.”Monetary units: In “French Centimes,” where 100 Centimes 5 1 French source: Archives Nationales (Paris), F11* 1779–2678 and F11* 2877–—CologneSource: Ebeling, Dietrich and Franz Irsigler. Getreideumsatz, Getreide- und Brotpreise in Köln 1368–1797, Erster Teil: Getreideumsatz und Getreidepreise: Wochen-, Monats- und Jahrestabelle. Part of H. Stehkämper (ed.), Mitteilungen aus dem Stadtarchiv von Köln, Vol. 65–66. Köln-Wien: Böhlau-Verlag, years: 1368 to 1797, with gaps; frequency: monthly; method: average from four to ve weekly prices. The source also lists weekly prices and quantities, which allows verifying that the difference between quantity-weighted and quantity-unweighted monthly price is typically less than 1 units: In “Kölner Malter,” which is approximately 150 liters or about 117 units: In “Albus”; conversions: 1 Albus 5 12 Heller, where the value of Albus to Gulden and Mark is 1 Gulden 5 4 Mark 5 24 sources: “Die Fruchtpreisbücher von 1531–1674,” “Die Bäckerbescheidbücher von 1658–1773,” and “Das Preise- und Umsatzverzeichnis von 1773–1797.”Germany—Rostock, Schwerin, Wismar, Boizenburg, Parchim, and GrabowSource: Die Getreidepreise im Grossherzogthum Mecklenburg-Schwerin während des Zeitraums von 1771 bis 1870, Beiträge zur Statistik Mecklenburgs. Schwerin: Mecklenburg-Schwerin Statistisches Landesamt, years: 1771 to 1870, with gaps; frequency: monthly; method: average of mid-price of all market days in a units: For Rostock and Schwerin in “Rostocker Scheffel” (RS); for Wismar in “Wismarer Scheffel” (WS); for Grabow and Parchim in “Maass” (M); and quantity units in Boizenburg are “ ¼ Sack” (¼ S). Conversions: 1 WS 5 RS, 1 M 5 RS, and ¼ S 5 RS, where 1 RS 5 units: In “Ganzen und Zehntel-Schillingen Courant” (., in Courant, with decimals).Original source: “Mecklenburgische Anzeigen” (a newspaper).Germany—MunichSource: Elsas, Moritz John. Umriss einer Geschichte der Preise und Löhne in Deutschland vom ausgehenden Mittelalter bis zum Beginn des neunzehnten Jahrhunderts. Leiden: . Sijthoff, years: 1690 to 1820, with gaps after 1779; frequency: monthly; method: average price of rst market day of the units: In “Scheffel,” where 1 Scheffel 5 about 223 units: In “Alten (schwarzen) Rechnungspfennigen.”Original source: “Schrannenzettel” of the city of —Augsburg, Bamberg, Bayreuth, Erding, Kempten, Landshut, Lindau, Memmingen, Munich, Nördlingen, Nürnberg, Regensburg, Straubing, Würzburg, ZweibrückenSource: Seuffert, Georg Karl Leopold. Statistik des Getreide- und Viktualien-Handels im Königreiche Bayern mit Berücksichtigung des Auslandes. München: . Weiss, years: 1815 to 1855 (Munich: 1790 to 1855); frequency: monthly; method: average price of all market days in a month. For Munich, the Seuffert (1857) source also contains weekly prices (for each Saturday). In principle, those are the prices listed in Elsas (1936); a comparison of the two sources essentially con rms units: In “Bavarian Scheffel.”
VOL. 97 NO. 4SHIUE AND KELLER: CHINA AND EUROPE AROUND THE INDUSTRIAL REVOLUTION1211Monetary units: In “Gulden and Kreuzer”; 1 Gulden 5 60 Kreuzer; conversion to prices for Munich from Elsas (1936): 1 Kreuzer 5 “Alte Rechnungspfennige.”Italy—SienaSource: Parente, Giuseppe. Prezzi e Mercato del Grano Siena 1546–1765. Florence: Carlo Cya, years: 1546 to 1765, with gaps; frequency: monthly; method: average from all prices of the units: In “Staio Senese”; 1 Staio Senese 5 units: In “Soldi”; this source gives conversion factors to compute constant prices. They are for the years 1542–1557: ; for the years 1558–1676: ; for the years 1677–1738: ; and for 1739–1766, the conversion factor is . We have experimented with both current and constant prices, with similar source: “Archivio degli Esecutori della Gabella,” Archives of the State of Siena, docu-ments 1283 to Netherlands—UtrechtSource: Sillem, Jérome Alexandre. Tabellen van Marktprijzen van Granen te Utrecht in de Jaren 1393 tot 1644,Verhandelingen der Koninklijke Akademie van Wetenschappen te Amsterdam, Johannes Müller: Amsterdam, 1901; a supplement with the eighteenth and nineteenth century gures is published in Posthumus, ., Inquiry into the history of prices in Holland, . Brill: Leiden, years: 1534–1644, 1760–1814; frequency: monthly; average of all prices for a given units: In “Modius”; 1 Modius 5 units: In Dutch source: “Rekeningen en weeklijsten der Domprossdij.”The Netherlands—NijmegenSource: Tijms, W. Historia Agriculturae. Groningen: Nederlands Agronomisch-Historisch Instituut, years: 1558–1916, with gaps; frequency: monthly; method: average price of the rst market day of each units: In “Malder” (1558–1822), and in “Hectoliter” (1824–1916); conversion: 1 Malder 5 liter 5 Units: In “Guldens” (Dutch Guilders).The Netherlands—Røermond (Ruremonde)Source: Ruwet, J., F. Ladrier, E. Helin, and L. van Buyten (1966), Marché des céréales à Ruremonde, Luxembourg, Namur et Diest aux XVIIe et XVIIIe siècles, Leuven: Presse de l’Universite de Louvain, years: 1599 to 1796; frequency: monthly; method: average of minimum and maximum of the rst market day of each units: In “Malder,” where 1 Malder is about 170 liters (a range from to is given, ).Monetary units: In “Stuivers,” which in French are “Patards.”Poland—Danzig (Gdansk)Source: Furtak, Tadeusz (1935), Ceny w Gda´nsku w latach 1701–1815, Lwow, years: 1703–1806; frequency: monthly, with gaps; the source gives minimum and maxi-mum price in a given month; we use the average of units: In “łaszt” (or last), where 1 last is about 3,010 units: In “złoty.”
1212THE AMERICAN ECONOMIC REVIEWSEPTEMBER 2007Original source: Reports no. 559 and 560 of the Danzig city library, and the newspapers “Danziger Erfahrungen” (1739–40) and “Danziger Nachrichten” (1740–1815).The United States—Boston, Charleston, New York, PhiladelphiaSource: Cole, Arthur H. (1938), Wholesale Commodity Prices in the United States, 1700–1861: Statistical Supplement: Actual Wholesale Prices of Various Commodities. Cambridge, MA: Harvard University Press. This is the price for rice in Philadelphia and Charleston, and the price for wheat in Boston, New York, and years: 1700–1861, varies by series; with units: units: in US , Micheline, and Jean Meuvret. 1960. Prix des céréales extraits de la mercuriale de Acemoglu, Daron, and Simon Johnson. 2005. Paris. Paris: Ecole des Hautes Etudes.“Unbundling Institutions.” Journal of Politi-Bernhardt, Kathryn, and Philip . Huang. cal Economy, 113(5): 949–. Civil Law in Qing and Republican Acemoglu, Daron, Simon Johnson, and James China. Stanford: Stanford University . Robinson. 2001. “The Colonial Origins Beveridge, William H. B. 1965. Prices and of Comparative Development: An Empirical Wages in England from the Twelfth to the Investigation.” American Economic Review, Nineteenth Century (Volume 1). London: 91(5): 1369– Cass. Allen, Robert C. 2001. “Real Wages in Europe Bogart, Dan. 2005. “Did Turnpike Trusts Increase and Asia: A First Look at the Long-Term Pat-Transportation Investment in Eighteenth-terns.” England?” Journal of Economic His-Allen, Robert C. 2002. “Involution, Revolution, tory, 65(2): 439– What? Agricultural Productivity, Income, Braudel, Fernand. 1992. The Structures of Every-and Chinese Economic Development.” Un- day Life: The Limits of the Possible. Berkeley: of California Press. (Orig. pub. in Allen, Robert C., Jean-Pascal Bassino, Debin Ma, French 1979.)Christine Moll-Murata, and Jan Luiten van Bray, Francesca. 1984. “Agriculture.” In Science Zanden. 2005. “Wages, Prices, and Living and Civilisation in China (Volume 6), ed. Standards in China, Japan, and Europe 1738–Joseph Needham. Cambridge: Cambridge 1925.” Paper presented at the Global Economic University Press. History Network conference, , Timothy. 2002. Geographical Sources of ndBairoch, Paul. 1975. “The Main Trends in Ming–Qing History. 2 edition. Ann Arbor: National Economic Disparities since the Indus-University of Michigan Revolution.” In Disparities in Economic Buck, John Lossing. 1937. Land Utilization in Development since the Industrial Revolution, China. Shanghai: Commercial . Paul Bairoch and Maurice Levy-Leboyer, Buxbaum, David. 1971. “Some Aspects of Civil 3–17. New York: St. Martin’s and Practice at the Trial Level in Balke, Nathan S., and Thomas B. Fomby. 1997. Tanshui and Hsinchu from 1789 to 1895.” “Threshold Cointegration.” International Eco- Journal of Asian Studies, 30(2): 255– Review, 38(3): 627–, Barry R., and Timothy Hatton. 2003. Banerjee, Abhijit V., and Lakshmi Iyer. 2005. “International Migration and the Integration “History, Institutions, and Economic Perfor-of Labor Markets.” In Globalization in His-mance: The Legacy of Colonial Land Ten-torical Perspective, ed. Michael D. Bordo, ure Systems in India.” American Economic Alan M. Taylor, and Jeffrey G. Williamson, Review, 95(4): 1190––119. Chicago: University of Chicago , Jörg. 2000. “Climate, Grain Production Cheng, Tianquan, Yongling Jiang, and William and Nutritional Status in Southern Germany C. Jones, trans. 1994. The Great Qing Code. thDuring the 18 Century.” : Oxford University Press.
VOL. 97 NO. 4SHIUE AND KELLER: CHINA AND EUROPE AROUND THE INDUSTRIAL REVOLUTION1213Chuan, Han-sheng, and Richard A. Kraus. 1975. Local Elites and Patterns of Dominance, Mid–Ch’ing Rice Markets and Trade: An ed. Joseph W. Esherick and Mary Backus Essay in Price History. Cambridge, MA: Rankin, 305–45. Berkeley: University of Harvard University , Arthur H. 1938. Wholesale Commodity Evans, Laurence. 1984. “Junks, Rice, and Empire: Prices in the United States, 1700–1861: Civil Logistics and the Mandate from Heaven.” Statistical Supplement: Actual Wholesale Historical Re ections, 11(3): 271– of Various Commodities. Cambridge, Feuerwerker, Albert. 1995. Studies in the Eco-MA: Harvard University History of Late Imperial China. Ann Crafts, N. F. R. 1995. “Macroinventions, Eco-Arbor: University of Michigan Growth, and ‘Industrial Revolution’ Frêche, Georges, and Geneviève Frêche. 1967. in Britain and France.” Economic History Les prix des grains des vins et des legumes Review, 48(3): 591–98.à Toulouse. Paris: Presses Universitaires de Deaton, Angus. 1999. “Commodity Prices and in Africa.” Journal of Economic Per-Furtak, Tadeusz. 1935. Ceny w Gda´nsku w spectives, 13(3): 23– 1701–1815. Lwow, Slkl. gl.: Kasa im. Deaton, Angus, and Guy Laroque. 1992. “On the J. Mianowskiego. of Commodity Prices.” Review of Glaser, Rüdiger. 2001. Klimageschichte Mittel-Economic Studies, 59(1): 1–: 1000 Jahre Wetter, Klima, Katas-Deaton, Angus, and Guy Laroque. 1996. “Com-trophen Rüdiger Glaser. Darmstadt: Storage and Commodity Price Goldstone, Jack A. 2002. “Ef orences and Eco-Dynamics.” Journal of Political Economy, nomic Growth in World History: Rethinking 104(5): 896– ‘Rise of the West’ and the Industrial Dickey, David, and Wayne A. Fuller. 1979. Revolution,” Journal of World History, 13(2): “Distribution of the Estimates for Auto-323– Time Series with a Unit Root.” Gomez, Victor, and Agustin Maravall. 1997. Journal of the American Statistical Asso-Program TRAMO (Time Series Regression ciation, 74(366): 427–31. with ARIMA Noise, Missing Observations, Drame, Sylvie, Christian Gonfalone, Judith A. and Outliers) and SEATS (Signal Extraction Miller, and Bertrand Roehner. Un Siècle de in ARIMA Time Series) Instructions for Commerce du Blé en France, 1825–1913. the User. Madrid: Secretaria de Estado de Paris: . Durand, John. 1960. “The Population Statistics Gongzhong liangjiadan.“Grain Price Lists in of China, . 2 – 1953.” Population Studies, the Palace Archives.” Beijing: Number One 13(3): 209– Archives. Ebeling, Dietrich, and Franz Irsigler. 1976. Gonzalo, Jesus. 1993 “Cointegration and Ag-Getreideumsatz, Getreide-und Brotpreise in gregation.” Richerche Economiche, 47(3): Köln, 1368–1797 (Volume 65). Cologne: Mit-281–91. teilungen aus dem Stadtarchiv von Köln. Granger, Clive W. J. 1981. “Some Properties of Elsas, Moritz John. 1936. Umriss einer Geschichte Time Series Data and Their Use in Econo-der Preise und Löhne in Deutschland vom metric Model Speci cation.” Journal of ausgehenden Mittelalter bis zum Beginn Econometrics, 16(1): 121– neunzehnten Jahrhunderts. Leiden: . Hall, Robert E., and Charles I. Jones. 1999. Sijthoff.“Why Do Some Countries Produce So Much Elvin, Mark. 1973. The Pattern of the Chinese More Output Per Worker Than Others?” Past. Stanford: Stanford University Journal of Economics, 114(1): Engle, Robert F., and Clive W. J. Granger. 83–. “Co-Integration and Error Correction: Henderson, William O. 1939. The Zollverein. Representation, Estimation, and Testing.” Cambridge: Cambridge University , 55(2): 251–, Ping-ti. 1954. “The Salt Merchants of Yang–Esherick, Joseph W., and Mary Backus Rankin. chou.” Harvard Journal of Asiatic Studies, 1993. “Concluding Remarks.” In Chinese 17: 130–68.
1214THE AMERICAN ECONOMIC REVIEWSEPTEMBER 2007Ho, Ping-ti. 1964. The Ladder of Success in Robert Engle and Clive W. J. Granger. New Imperial China: Aspects of Social Mobility. York: Oxford University , MA: Harvard University , Angus. 1989. The World Economy in Huang, Philip C. C. 1996. Civil Justice in China. the 20th Century. Paris: Organisation for Eco-Representation and Practice in the Qing. nomic Co-operation and Development Publi-Stanford: Stanford University and Development , Philip C. C. 2002. “Development or Mann, Susan. 1987. Local Merchants and the Involution in Eighteenth-Century Britain and Chinese Bureaucracy, 1750–1950. Stanford: China?” The Journal of Asian Studies, 61(2): Stanford University Press. 501–38. Mann Jones, Susan. 1972. “Finance in Ningpo: Isham, Jonathan, and Daniel Kaufmann. 1999. The ‘Ch’ien Chuang,’ 1750–1850.” In Eco-“The Forgotten Rationale for Policy Reform: nomic Organization in Chinese Society, ed. The Productivity of Investment Projects.” W. E. Willmott, 47–77. Stanford: Stanford Quarterly Journal of Economics, 114(1): University –, Robert B. 1998. Tigers, Rice, Silk, and Jackman, W. T. 1962. The Development of Silt. Cambridge: Cambridge University in Modern England. 2 edi-McCloskey, D. N., and John Nash. 1984. “Corn at tion. London: Frank Cass & Company : The Extent and Cost of Grain Storage Junjian, Jing. 1994. “Legislation Related to the in Medieval England.” American Economic Civil Economy in the Qing.” In Civil Law in Review, 74(1): 174– and Republican China, ed. Kathryn McEvedy, Colin, and Richard Jones. 1978. Atlas of Bernhardt and Philip Huang, 42–84. Stanford: World Population History. New York: Penguin. Stanford University -Schwerin Statistisches Landesamt. Johansen, Soren. 1988. “Statistical Analysis of 1873. Beiträge zur Statistik Mecklenburgs. Cointegration Vectors.” Journal of Economic Schwerin: Grossherzogliches Statistisches Dynamics and Control, 12(2/3): 231–, Steven L. 1984. Provisioning Paris. Mingay, Gordon E. ed. 1989. The Agrarian His-Ithaca: Cornell University of England and Wales Volume VI, 1750–Keller, Wolfgang, and Carol. H. Shiue. Forth-1850. Cambridge: Cambridge University coming. “The Origins of Spatial Interaction.” of , B. R. 1988. British Historical Statistics. Labrousse, C. Ernest. 1932. Esquisse du mou-Cambridge: Cambridge University des prix et des revenus en France Mokyr, Joel. 1990. The Lever of Riches. Tech-au 18e siècle. Paris: Editions des archives nological Creativity and Economic Progress. contemporaines. New York: Oxford University Press. Landes, David S. 1969. The Unbound Pro-Myers, Ramon H., and Yeh-Chien Wang. 2002. metheus, Technological Change and Indus-“Economic Developments, 1644–1800.” In trial Development in Western Europe from The Cambridge History of China Volume 9, 1750 to the Present. Cambridge: Cambridge ed. Willard J. Peterson, 563–645. Cambridge: University Press. Cambridge University , Lillian M. 1992. “Grain Prices in Zhili Naquin, Susan, and Evelyn S. Rawski. 1987. Province, 1736–1911: A Preliminary Study.” Chinese Society in the Eighteenth Century. In Chinese History in Economic Perspective, New Haven: Yale University . Thomas G. Rawski and Lillian Li, 69–99. Nielsen, Randall. 1997. “Storage and English Los Angeles: University of California Intervention in Early Modern The London Gazette. Serial Publication News-Grain Markets.” Journal of Economic His-paper, on micro lm 1666–1800, Issues tory, 57(1): 1––13736. San Antonio College Libraries, North, Douglas C. 1981. Structure and Change San Antonio, Economic History. New York: , James G. 1991. “Critical Values for North, Douglass C., and Barry R. Weingast. Cointegration Tests.” In Long-Run Economic 1989. “Constitutions and Commitment: The Relationships, Readings in Cointegration, ed. Evolution of Institutions Governing Public
VOL. 97 NO. 4SHIUE AND KELLER: CHINA AND EUROPE AROUND THE INDUSTRIAL REVOLUTION1215Choice in Seventeenth-Century England.” Rowe, William T. 1984. Hankow. Commerce and Journal of Economic History, 49(4): 803– in a Chinese City, 1796–1889. Stan-O’Rourke, Kevin H., and Jeffrey G. Williamson. ford: Stanford University . Globalization and History: The Ruwet, Joseph. 1966. Marché des céréales à Evolution of a Nineteent-Century Atlantic Ruremonde, Luxembourg, Namur et Diest Economy. Cambridge, MA: MIT XVIIe et XVIIIe siècles. Louvain: Publi-Parente, Giuseppe. 1942. Prezzi e Mercato del cations Universitaires de Siena (1546–1765). Florence: Carlo Seuffert, Georg Karl Leopold. 1857. Statistik Getreide-und Viktualien-Handels im Perdue, Peter. 1987. Exhausting the Earth: State Königreiche Bayern mit Berücksichtigung and Peasant in Hunan, 1500–1850. Chicago: des Auslandes. München: . Publishing. Shiue, Carol H. 2002. “Transport Costs and Perdue, Peter. 1992. “The Qing State and the the Geography of Arbitrage in Eighteenth-Gansu Grain Market, 1739–1864.” In Chinese Century China.” American Economic Review, History in Economic Perspective, ed. Thomas 92(5): 1406–. Rawski and Lillian M. Li, 100–25. Los Shiue, Carol H. 2004. “Local Granaries and Angeles: University of California Government Disaster Relief: Moral Perkins, Dwight D. 1969. Agricultural Develop-Hazard and Intergovernmental Finance in ment in China 1368–1968. Chicago: Aldine Eighteenth- and Nineteenth-Century China.” of Economic History, 64(1): 100–, Karl-Gunnar. 1999. Grain Markets in Shiue, Carol. H. 2005a. “The Political Economy Europe, 1500–1900, Integration and Deregu-of Famine Relief.” Journal of Interdisciplin-lation. Cambridge: Cambridge University ary History, 36: 33–, Carol H. 2005b. “From Political Petersen, Christian. 1995. Bread and the Fragmentation Towards a Customs Union: British Economy, c1770–1870. Scolar Press: Border Effects of the German Zollverein, to 1855.” European Review of Economic P ster, Christian. 2002. “The Potential of Docu-History, 9(2): 129– Data for the Reconstruction of Shiue, Carol H., and Wolfgang Keller. 2004. Past Climates in Europe.” . / Markets in China and Europe on the Eve of esf_holivar/p Industrial Revolution. National Bureau of Pomeranz, Kenneth. 2000. The Great Diver-Economic Research Working Paper : China, Europe, and the Making of Shiue, Carol H., and Wolfgang Keller. 2006. the Modern World Economy. Princeton, NJ: “Institutional Foundations of Grain Trade Princeton University China: A Comparative Perspective.” Posthumus, N. W. 1964. Inquiry into the History Prices in Holland. Leiden: . , Jérome Alexandre. 1901. Tabellen van Pribram, Alfred Francis. 1938. Materialien zur Marktprijzen van Granen te Utrecht in de Geschichte der Preise und Löhne in Öster- Jaren 1393 tot 1644, uit de rekeningen en reich (Volume I). Vienna: Carl Überreuters weeklijsten der Domprossdij. Amsterdam: J. Verlag. Mü, Dani, Arvind Subramanian, and Skinner, G. William. 1964. “Marketing and Francesco Trebbi. 2002. “Institutions Rule: Social Structure in Rural China: Part I.” Jour-The Primacy of Institutions over Geography nal of Asian Studies, 24(1): 3–43. and Integration in Economic Development.” State Meteorological Society. 1981. Zhongguo National Bureau of Economic Research Work-jin wubai nien hanloa fenbu tuji (Collected ing Paper of Droughts and Floods in China in the Roehner, Bertrand, and Carol H. Shiue. 2000. Past Five Hundred Years.) Beijing: Ditu chu-“Comparing the Correlation Length of Grain ban in China and France.” Interna-Taylor, Alan M. 2001. “Potential Pitfalls for the tional Journal of Modern Physics C, 11(7): Purchasing-Power-Parity Puzzle? Sampling 1383– Speci cation Biases in Mean-Reversion
1216THE AMERICAN ECONOMIC REVIEWSEPTEMBER 2007Tests of the Law of One Price.” Econometrica, Wong, R. Bin. 2002. “The Search for European 69(2): 473– and Domination in the Early Tong, James W. 1991. Disorder under Heaven. Modern World: A View from Asia.” Ameri-Stanford: Stanford University Historical Review, 107(2): 447–69. Tijms, W. 1977. Historia Agriculturae. Groningen: Agronomisch-Historisch Tielhof, Milja. 2002. The “Mother of All Wong, R. Bin, and Peter Perdue. 1992. “Grain thTrades.” Leiden: and Food Supplies in 18 Century Verlinden, C. ed. 1959. Dokumenten voor de Hunan.” In Chinese History in Economic Geschiedenis van Prijzen en Lonen in Vlaan-Perspective, ed. Thomas G. Rawski and Lillian deren en Brabant (Vol. I.) Bruges: De Tempel. M. Li, 126–44. Los Angeles: University of Verlinden, C. ed. 1972. Dokumenten voor de Ge- California van Prijzen en Lonen in Vlaanderen Worcester, G. R. G. 1971. The Junks and Sam-en Brabant (Vol. III). Bruges: De Tempel. pans of the Yangtze. Annapolis: . Naval Verlinden, C. ed. 1973. Dokumenten voor de van Prijzen en Lonen in Xing, Fang, Shi Qi, Jian Rui, and Wang Shixin. Vlaanderen en Brabant (Vol. IV). Bruges: De 2000. “The Growth of Commodity Circulation the Rise of Merchant Organisations.” In Vogel, Hans Ulrich. 1987. “Chinese Central Chinese Capitalism, 1522–1840, ed. Xu Dixin Monetary Policy, 1644–1800.” Late Imperial and Wu Chengming, 165–83. New York: St. China (Vol. 8), 1–’s Press. (Orig. pub. in Chinese 1985.)Wang, Yeh–Chien. 1992. “Secular Trends of Xu, Dixin, and Wu Chengming. 2000. Chinese Rice Prices in the Yangzi Delta, 1638–1935.” Capitalism, 1522–1840. London: MacMillan In Chinese History in Economic Perspective, Press. (Orig. pub. in Chinese 1985.)ed. Thomas G. Rawski and Lillian M. Li. Los Yumin, Zhou. 2000. Wan Qing caizhen yu she-Angeles: University of California bianqian (Finance and Social Change in Wilkinson, Endymion. 1969. “The Nature of the Late Qing.) Shanghai: Shanghai renmin Chinese Grain Price Quotations, 1600–.” Transactions of the International Con-Zelin, Madeleine. 1993. “The Rise and Fall of ference of Orientalists in Japan/Kokusai Toho the Fu-Rong Salt-Yard Elite: Merchant Domi-Gakusha Kaigi Kiyo (Vol. 14): 54–65. Tokyo: nance in Late Qing China.” In Chinese Local Toho and Patterns of Dominance, ed. Joseph Will, Pierre–Etienne, and R. Bin Wong. 1991. E. Esherick and Mary Backus Rankin, 82–Nourish the People. Ann Arbor: University 109. Berkeley: University of California Michigan Press. Zurndorfer, Harriet T. 2004. “Imperialism, Glo-Williams, Jeffrey C., and Brian D. Wright. 1991. balization, and Public Finance: The Case of Storage and Commodity Markets. Cambridge: Late Qing China.” Global Economic History Cambridge University Working Paper 06/04.
