Sukkoo Kim
Department of Economics
Campus Box 1208
Washington University
One Brookings Drive
St Louis MO 63130-4899
Telephone: 314-935-4961 Fax: 314-935-4156
skim@eh.net

September 11, 1998

I am grateful to Marcus Berliant, Dora Costa, Hideo Konishi, Charles Leven, Douglass North, Kenneth Sokoloff, John Wallis, Thomas Weiss and seminar participants at the Regional Science Association Meetings in Washington DC, Cliometrics Conference at University of Toronto, Philadelphia Federal Reserve Bank, NBER Summer Institute, and Tulane for helpful comments. I thank the John M. Olin Foundation for financial support.

The history of cities in the United States has witnessed dramatic developments over the last three centuries. In the mercantile era, between the late seventeenth and the eighteenth centuries, cities were few in number, concentrated along the eastern seaboard, and were dominated by merchants who facilitated trade with Europe. In the industrial era, between the early nineteenth and the early twentieth centuries, the onset of industrialization and the expansion of the domestic market significantly increased the number and size of cities. Moreover, large industrial cities emerged and became concentrated in a few regions of the United States. In the service era, the period which began in the middle of the twentieth century, the urban economy became increasingly dominated by services, mirroring the trend in the overall economy. Major cities emerged in the southwestern region of the United States and the pace of urban population dispersion or suburbanization accelerated.

Yet, despite these dramatic historical changes in the number, size, location, density, and the economic structure of cities, one characteristic of cities has remained remarkably stable. At any given point in time, cities varied enormously in their population sizes. In particular, the size distribution of cities has followed, what urban economists refer to as, the rank-size rule. This rule states that the population of a city is equal to the population of the largest city divided by the rank of the city in question. Surprisingly, the population size distribution of cities has been relatively stable despite significant movements in the ranks of many cities. While some cities such as New York were able to maintain their rankings throughout the last two centuries, the rankings of the majority of cities shifted considerably over time.

This paper argues that the long-run trends in U.S. urban development are explained by changes in the structure of the economy which, in turn, influenced the nature of regional comparative advantage and trade. As the economy shifted from agriculture to manufacturing and then to services, the nature and scope of trade changed. Since goods must be transported for trade to occur, these changes in the level and pattern of trade, in conjunction with changes in transportation and local public goods provision technologies, led to the defining of three distinct eras of U.S. urban development. Throughout these periods, cities varied considerably in size because the larger cities performed special market-making functions. Regional comparative advantage led to trade, but gains from trade did not come freely. In addition to the physical cost of transporting goods, the geographic division of labor increased market transaction costs. The cost of coordinating the supply and demand of goods in the economy increased as goods were produced and sold in larger geographic areas. The concentration of market coordinators, as well as institutions, which inspected, certified and enforced contracts, reduced the transaction costs associated with this greater geographic division of labor.

I. The Periods of U.S. Urban Development.

This paper proposes that the history of U.S. urban development can be usefully divided into three periods: the mercantile era, 1690-1820, the industrial era, 1820-1920, and the service era, 1920-present. The three periods of U.S. urban development exhibited considerably different rates of urbanization. The growth in the number and size of cities during the mercantile period was quite modest. During this period, the number of cities with populations of greater than 2,500 increased from 4 to 61 but the percentage of urban population declined slightly from 8.3% to 7.2%. In the industrial era, the number and size of cities in the United States grew at historically unprecedented rates; the number of cities increased from 61 to 2722 and the percentage of urban population increased from 5.1% to 51.2%. The number of cities continued to increase during the service era, but the level of urbanization in central cities peaked at 65% in 1960 and declined to 62% in 1990. However, the urban population in metropolitan areas continued to increase over time, albeit in a more dispersed manner. The percentage of population in metropolitan areas rose from 51.0% to 77.5% between 1940 and 1990.

The pace and pattern of U.S. urban development were closely related to the changes in the economic structures of cities. In the mercantile era, the economic structures of cities were dominated by the merchants and the surrounding hinterland activities in agriculture and other extractive industries. In the industrial era, the economic activity of the majority of cities was dominated by manufacturing. Already at the beginning of the industrial era in 1820, a significant portion of cities’ population was engaged in manufacturing activities.1 Industrial activities continued to play a significant role in city economies between the late nineteenth and the early twentieth centuries. The manufacturing employment as a percentage of population for large cities was 14.9%, 17.1% and 18.2% in 1880, 1900, and 1920 respectively. In the service era, the importance of manufacturing in cities declined significantly. By 1990, less than 10% of the population in cities was engaged in manufacturing, or conversely, approximately 34% percent of the population was engaged in services.

Large cities became concentrated in different geographic regions of the United States during the different eras of urban development.2 In the mercantile era, the majority of cities were located along the eastern seaboard in the New England, Middle Atlantic and South Atlantic regions. In the industrial era, large cities emerged in new areas of the New England and Middle Atlantic regions and in the East North Central region. During this era, 60% of large cities located in these three regions. The service era ushered in yet another significant change in the geographic distribution of cities. The share of cities in the northeast decreased dramatically as the share in the southwest increased. The New England and Middle Atlantic regions’ shares of large cities fell from an apex of 21% and 25% at the turn of the century to 4.6% and 6.6%, whereas the shares in the West South Central, Mountain and Pacific regions rose from 5%, 2%, and 5% to 14%, 9%, and 26%.

The long-run historical trends in the pace and pattern of urbanization in the United States provide a great opportunity to examine the various theories of city formation. In particular, this paper examines the two dominant theories of city formation based on comparative advantage and externalities.3 Comparative advantage leads to trade and cities lower the cost of trade. This cost is lower in cities if there are physical, port and terminal operation economies in transportation.4 In addition, cities may lower the cost of trade between firms and workers and between firms themselves if there are economies in the provision of local public goods such as local transportation, water, gas, electricity, and communications.5 Alternatively, externalities may also generate cities. Cities may arise if firms and workers locate near each other to take advantage of Marshallian externalities such as technological spillovers, labor market pooling, and non-traded industry specific inputs.6 The limit on the size of cities in both theories can come from the costs of density associated with greater exposure to disease, fire, crime, pollution, and congestion.

Data on industrial structures of large cities for 1880 and 1940 were constructed to examine these two theories of city formation. The data indicate that cities were quite specialized in their manufacturing structures at the 2-digit industries.7 In 1880, for 22% of large cities, one industry accounted for at least, and often far more than, half of a city’s manufacturing employment and, for another 24% of the cities, two industries accounted for at least half of its manufacturing employment. Despite a significant increase in their sizes, cities were equally specialized in 1940. In that year, 21.8% of the cities had at least half of their manufacturing employment in one industry and another 25.3% in two industries. Within any given year, the larger cities were likely to be more diversified than smaller cities because industries such as food, apparel, and printing were always well represented in the larger cities.

Although these patterns of city specialization cannot fully identify the sources of city formation, they do provide some clues. The high level of city specialization suggests that if externalities are at work, then they must be of the localization rather than of the urbanization type.8 However, city specialization may be caused by comparative advantage. Additional clues as to why city specialization may be caused by comparative advantage comes from the fact that cities within census regions were far more likely to be similar in terms of industrial structure than cities across census regions.9 If externalities were localized in cities, then there is little reason why cities within a region should be specialized in the same set of industries. On the other hand, if externalities were localized within a census region, then they cannot cause cities to arise. The explanation based on comparative advantage suggests that cities within regions specialized in similar sets of industries because of regional comparative advantage in these industries. In these regions, activities were organized in cities because of economies in local public goods and in transportation of inputs and final goods, both of which lowered the cost of trade.

II. The Size Distribution of Cities.

The existence of the size distribution of cities has also generated great interest among scholars. Yet, despite the apparent persistence of this phenomenon over time, there is little consensus on the exact causes of the size distribution.10 This section explores whether cities of different sizes possess different economic and social structures.11 The analysis of the economic structures of cities by size distribution is based on samples of large U.S. cities for various years between 1900 and 1990. The dependent variables in each of the cross-sectional regressions are the natural log of population, city rank and city density. The set of independent variables differ from year to year as the occupational structure has become more refined over time. Although not available for all years, they include the percentage of the total population that is foreign-born, black, and educated, and the percentage of the total population that is employed in agriculture, mining, manufacturing, wholesale trade, retail trade, transportation, FIRE, business services, personal services, amusement, professional services and government.

The results of the regressions indicate that the large cities differed from smaller cities in a number of ways. The large cities consistently had a greater proportion of their population engaged in transaction services. In 1900 and 1920, a standard deviation increase in the percentage of the population engaged in trade related activities resulted in a more than 40 percent increase in city population and an increase in the ranking of cities by 25 places. In 1940, standard deviation increases in the percentage of population engaged in wholesale trade and business services increased the population by 34% (10 places in rank) and 54% (16 places in rank) respectively. While still significantly positive, the influence of transaction services declined in 1990. The larger cities also had a greater percentage of employment in amusement and government activities and they also tended to have a significantly greater percentage of the population that was foreign born and black. As expected, the percentage of the population engaged in agriculture was negatively correlated with city size.

The data seem to reject two popular theories of the size distribution of cities. City sizes were not correlated with retail nor manufacturing activities. Contrary to the predictions of the central place theory based on economies in retail trade, the data indicate that smaller cities had a greater percentage of their population engaged in retail trade. For example, in 1940 and 1990, a standard deviation increase in the percentage of the population engaged in retail trade resulted in a 10% and 23% decline in city population, respectively. Data also indicate little correlation between city size and manufacturing activities. For most years, the percentage of the population engaged in manufacturing was unrelated to city sizes and in 1990, the relationship was significantly negative.

III. Concluding Remarks.

The study of cities in the United States over the last three centuries suggests that there were three distinct periods of urban development. In the mercantile era, the number and size of cities were small because the benefits to trade were relatively small. Since the American economy was dominated by agriculture and other extractive industries, trade in inputs was limited by the immobility of land. Moreover, the spatially dispersive nature of these activities made it difficult to take advantage of economies of scale in the provision of local public goods. Most of the benefits of trade came from the exchange of final goods. The regions of British America were economically integrated with Europe and each region specialized in a different set of staple crops. Cities lowered the cost of trade during this period because of economies of scale in port operations. The frequency of ship arrivals and departures, newspapers, and the concentration of merchants in cities all lowered the cost of obtaining information concerning the supply and demand conditions of a variety of products.

In the industrial era, the number and size of cities increased as the benefits to trade increased dramatically. The general increase in incomes and lower transportation costs increased the level of trade in final goods. The growth in the manufacturing sector also increased the scope of trade in raw materials and trade between workers and firms in the labor market. The growth in the volume of trade in concert with modes of transportation with significant economies in traffic density and in terminal operations, especially in the transport of heavy and bulky raw materials, contributed to the rise of large manufacturing cities. In addition, significant developments in the technologies of local public goods such as intra-urban electric railroad, gas, electric, and telephone utilities lowered the cost of trade in cities.

In the service era, the growing importance of services once again changed the nature of trade in the economy. Although the potential benefits to trade in services may have been high, there were few economies in the transportation of inputs and outputs of services. In this era, two developments led to a drastic geographic redistribution of cities. First, as city economies became predominantly service-oriented, proximity to resources became less important for city development. Second, as factors became increasingly more mobile and as technological innovations favored the development of substitutes, recycling, and less resource-intensive methods of production, differences in regional factor endowments diminished. Consequently, as cities became less industrial and as regions became more similar, cities migrated away from the manufacturing belt to the southwestern regions of the United States.

Throughout these three eras of U.S. urban development, cities exhibited considerable variation in their sizes because the larger cities lowered the costs of market transactions. In the mercantile era, when markets were still relatively small, the larger cities lowered the cost of coordinating supply and demand by maintaining a concentration of merchants, newspapers, trade journals, and insurance firms, and developing auction markets. In the industrial era, as the size of markets grew, the development of organized exchanges and the concentration of specialized merchants continued to perform similar functions. However, the geographic area serviced by the very largest cities grew as merchants in these cities coordinated the national and international supply and demand of goods. These very largest cities exhibited considerable persistence in their rankings over time. On the other hand, the fortunes of medium to large cities, which serviced their regional markets, depended a great deal on the fortunes of their regional economies and exhibited considerable movements in their rankings over time.

In recent years, the most significant trend in economic geography has been the general dispersal of economic activities. Since the middle of the twentieth century, there has been a significant decline in specialization and localization of industries at the regional level. The convergence in regional industrial structures has also been accompanied by a convergence in regional income per capita. From an urban standpoint, the trend toward population dispersion began much earlier. The introduction of more efficient intra-city transportation modes such as street railways allowed the population density to fall in the central cities as residences radiated outwards. However, since the middle of the twentieth century, the pace of spatial dispersion in cities increased significantly. The growth of low-density, so-called edge-cities, has significantly changed the urban life of Americans and will continue to have long-lasting influence as it changes the landscape of local political jurisdictions.

1 See Williamson and Swanson (1966).

2 See Leven (1993) for information on the regional distribution of urban and metropolitan population between 1790 and 1988.

3 See Mills and Hamilton (1989).

4 See Konishi (1996) and Fujita and Mori (1996).

5 See Berliant and Wang (1993).

6 See Henderson (1988), Henderson et. al (1995), Krugman (1996), Glaeser (1998), Glaeser et. al (1992), and Glaeser et. al (1995).

7 Also see Henderson (1988).

8 Urban economists categorize externalities into two types: if externalities are captured within industries, then they are called localization economies and if externalities are captured across industries, then they are called urbanization economies.

9 Kim (1995, 1998a, 1998b) argues that the long-run trends in U.S. regional specialization and regional income per capita are consistent with explanations based on comparative advantage rather than externalities.

10 The standard textbook explanation for the existence of an urban hierarchy is the central place theory developed by Christaller (1966) and Lösch (1954). This theory explains the size distribution of cities based on economies of scale in retail markets. However, there are numerous other theories. Henderson (1988) argues that the distribution of city sizes is due to economies of scale in manufacturing. Since cities specialize in different industries with different scale economies, cities will typically differ in size. Pred (1973) argues that the size distribution of cities form a "system" based on the circulation of information, and Krugman (1996) notes that size distribution of cities can be generated by a growth process where the growth rate of any city is random and is independent of size. For more recent works on this topic, see Black and Henderson (1997) and Fujita et. al (1997).

11 See Glaeser (1998).

12 See Wallis and North (1986) for definition of transaction services.

13 See Moses and Williamson (1967).

Berliant, M. and P. Wang, "Endogenous Formation of a City Without Agglomerative Externalities or Market Imperfections: Marketplaces in a Regional Economy," Regional Science and Urban Economics (1993).

Black, D. and V. Henderson, "Urban Growth," NBER Working Paper #6008 (1997).

Christaller, W., The Central Places of Southern Germany (Prentice-Hall, 1966).

Fujita, M. and T. Mori, "The Role of Ports in Making of Major Cities: Self-Agglomeration and Hub effects," Journal of Development Economics (1996).

Fujita, M., T. Mori and P. Krugman, "On the Evolution of Hierarchical Urban Systems," mimeo (1997).

Glaeser, E., "Are Cities Dying?" Journal of Economic Perspectives (1998).

Glaeser, E., Kallal, J. Scheinkman and A. Shleifer, "Growth in Cities," Journal of Political Economy (1992).

_____, "Economic Growth in a Cross-Section of Cities," Journal of Monetary Economics (1995).

Henderson, V., Urban Development, (Oxford University Press, 1988).

Henderson, V., A. Kuncoro and M. Turner, "Industrial Development of Cities," Journal of Political Economy (1995).

Kim, S., "Expansion of Markets and the Geographic Distribution of Economic Activities: The

Trends in U.S. Regional Manufacturing Structure, 1860-1987," Quarterly Journal of

Economics (1995).

_____, "Regions, Resources, and Economic Geography: Sources of U.S. Regional Comparative Advantage, 1880-1987," Regional Science and Urban Economics, forthcoming. Also available as NBER working paper #6322 (1998a).

_____, "Economic Integration and Convergence: U.S. Regions, 1840-1987," Journal of Economic History, forthcoming. A longer version of this paper is available as NBER working paper #6335 (1998b).

_____, "Urban Development in the United States, 1960-1990," mimeo (1998c).

Konishi, H., "Hub Cities: City Formation Without Economies of Scale," mimeo, 1996.

Krugman, P., The Self-Organizing Economy (Blackwell Publishers, 1996).

Leven, C., "Regional Changes in Urban and Metropolitan Population in the United States," in

Urban Change in the United States and Western Europe, eds, A. Summers, P. Cheshire, and L. Senn (The Urban Institute Press, 1993).

Lösch, A., The Economics of Location Translated by W. Woglom (New Haven, CT: Yale University Press, 1954)

Mills, E. and B. Hamilton, Urban Economics, 4th Edition (Scott, Foresman and Co., 1989).

Moses, L. and H. Williamson, "The Location of Economic Activity in Cities," American Economic Review, Papers and Proceedings (1967).

Pred, A., Urban Growth and the Circulation of Information: The United States System of Cities, 1790-1840 (Harvard University Press, 1973).

Wallis, J. and D. North, "Measuring the Transaction Sector in the American Economy, 1870-

1970," in S. Engerman and R. Gallman, eds., Long-Term Factors in American Economic Growth (University of Chicago Press, 1986).

Williamson, J. and J. Swanson, "The Growth of Cities in the American Northeast, 1820-1870," Explorations in Entrepreneurial History (1966).