Specialized industrial cities in the United States 1860–1930

Specialized industrial cities in the United States 1860–1930

Joz4rnal o/‘His/orical Geograph_v. 10, ( 1984) 37 51 Specialized industrial cities in the United States 18604930 Joseph Persky and Ronald Moses T...

1MB Sizes 0 Downloads 56 Views

Joz4rnal

o/‘His/orical

Geograph_v.

10, ( 1984) 37 51

Specialized industrial cities in the United States 18604930 Joseph Persky and Ronald Moses

There is a general view among geographers. historians and economists that a high level ot industrial specialization is likely to inhibit the growth of a city. The arguments in support of this position emphasize the importance of diversity in the generation of new work. This position is more reasonable for a large geographic region than for an individual city in a regional system. Empirical data on the growth of American cities in the late nineteenth and early twentieth centuries fail to support the proposition that specialization retards population growth. Specialized cities of a given region and age grew at about the same rate as their regional contemporaries. These data are consistent with the notion that flows 01 capital. entrepreneurship and labor among cities in a regional system render specialization of little importance to long run population growth.

How has industrial specialization affected the growth of cities? The general opinion would seem to be that specialized industrial structure is a precarious base for an urban area. The historical record of specialized cities in the United States has been described as one of failure. The attack on specialization reached an extreme level in the work of the National Resources Planning Board in the late 1930~.[‘~ The Board emphasized the difficulties faced by specialized cities in dealing with employment losses in their primary industry. This argument has continued to the present. For example, in a much cited article Benjamin Chinitz claims that specialized cities are unlikely to produce agglomeration economies attractive to other industries.“] The most striking indictment of specialization is contained in the work of Jane Jac0bs.t”’ In her view a specialized city is vulnerable because of its limited ability to generate new work. New work is the result of entrepreneurial activities which are better nourished by a diverse and hence more stimulating environment. Specialization encourages only a narrow economic efficiency that is likely to be inflexible in the face of external change. Jacobs’ analysis is interesting because of her commitment to considering technological change and product development as endogenous processes. tJ1 However, like many of the critics of specialization. Jacobs considers cities as virtually isolated social and economic systems. Although she asserts that productive activities are likely to move down the citysize hierarchy as their technologies become more standardized, she allows for no other interactions. Cities exist in a world devoid of any equilibrating tendencies. A burst of specialized industrialization is a damnable growth that can only set the stage for disaster. Many of the conclusions of Jacobs and other critics of specialization rest heavily on a belief that growth is largely determined at the level of the individual 0305 7488/84/010037+

15 $03.00/O

(’ 1984 Academic

Press Inc. (London)

Ltd

38

J. PERSKY

AND R. MOSES

city. By contrast, if one emphasizes the importance of the mobility of labor and capital a very different set of conclusions is likely to be drawn. Among cities in geographic proximity such mobility will be considerable. Rapid growth in one city is likely to affect its entire region. Much of our thinking about urban development and growth has been cast in a paradigm of natural selection and competition. But recent historical work such as that of Edward Muller suggests the importance of “complementarity rather that strictly competitive relationship(s) among the smaller cities” in the nineteenth century.t51 A related argument is based on the literature of the urban ratchet effect. This work suggests that the infrastructure of an urban area is not likely to be deserted in the face of even major industrial setbacks. thl The environment created in specialized growth may be just as valuable as that created by a more diversified economy.[‘l Although these arguments are far from conclusive, they suggest that the effect of specialization on urban growth must be determined from empirical and historical investigations. Although the historical record of specialized industrial cities in the United States has often been invoked empirical work on this history has been largely anecdotal in nature. In the late nineteenth and early twentieth centuries many American cities were highly specialized. A few of these have received attention as case studies, most notably Lowell, Massachusetts. [*I However, these specialized cities have not been studied in a comprehensive and consistent fashion. The purpose of this paper is to begin the systematic evaluation of the growth of specialized industrial cities in the United States. The empirical record of specialized cities in the late nineteenth and early twentieth centuries shows no important differences between their size distribution and that of all other cities once age and region are accounted for. If specialization was a liability in this period it must have worked not at the level of individual cities. but rather at the regional level. Because of the numerous methodological problems in applying formal hypothesis testing to historical data of the type used here, the general approach adopted is one of descriptive statistics. In the spirit of exploratory data analysis we put considerable emphasis on the tabular presentation of data. A section explaining our methodology is followed by the basic empirical results. We then consider several alternative formulations to establish the robustness of our findings. In the last section we expand on the implications of these results for the theory of urban specialization. Data and methods The basic data for this study consist of decennial population statistics for American cities. Complete population series up through 1960 were collected for all cities with populations greater than 20,000 in 1900. The figure 20,000 was in large part dictated by data limitations. For smaller places no industrial employment data are available. The next problem was to define a specialized manufacturing city. Clearly such a community had to devote a considerable share of its total labour force to a single manufacturing industry. In our view specialization is a qualitative and relative phenomenon. Nevertheless for operational purposes we must have a quantitative definition. In somewhat arbitrary fashion we chose the following criteria: first, the city must have manufacturing employment equal to at least 15”;, of its total population and secondly, at least 259, of that manufacturing employment must be concentrated in a single industry. These criteria are best judged in terms of the list of specialized

INDUSTRIAL

CITIES IN USA 1860-1930

39

cities they generate. Virtually all of the cities listed in the Appendix have been well known for their specialities. Rather than over-rationalize the definitions used here, we believe the appropriate check on our findings is to perform a sensitivity analysis. As discussed below, a more stringent definition of specialization yields much the same results. . It should be noted that the industries used to define specialized cities are based on the industry classifications of the censuses of 1880 and 1900. All employment data were taken from the manufacturing volumes of these censuses. We attempted to convert the industry definitions for each year to current two-digit Standard Industrial Classification (SIC) industries. The most important implication of this transformation was to include employment in various textile specialities as portions of a single textile industry, that is, employment figures for cotton mills, knitting mills, etc., were added together. This seemed appropriate because of the considerable overlap in the technologies, labour skills, and capital ownership in these industries. We did not allow cities to specialize in railroad repair, sawmilling, or foundries. Most workers in the first two of these would not now be considered employed in manufacturing industries. The foundry category was excluded because of its diverse nature and close relationship to multiple finishing industries. These industry restrictions tend to reduce growth in the specialized group of cities since the railroad centres were to be particularly rapid growers in the twentieth century. Perhaps the most important methodological question in this study was to determine the appropriate control or comparison group for the specialized cities. Every city faces obstacles and opportunities over its history. To a large extent these factors depend on the geographic location of the city and the time of its emergence. If the basic critique of specialization is to have force, it must compare a specialized city to a similarly situated non-specialized city starting from about the same absolute size. It is quite uninformative to compare the growth of a small mill town in New England to that of Chicago or Los Angeles. Presumably the critics of specialization are arguing that cities which specialize early in their history are likely to grow slower in subsequent periods. In what follows, we will comment briefly on general comparisons between all specialized cities and all nonspecialized cities. The major thrust of the paper, however, is to consider the implications of specialization, controlling for region and age. For regional definitions we will use the standard major census categories: Northeast, North Central, South and West.[” These are quite large geographic areas and there is considerable evidence to suggest that especially in the manufacturing belt the economic prospects of cities were heavily influenced by smaller regional sub-systems. [lo1 Nevertheless, for this work it seemed appropriate to err on the side of ‘under-controlling”. A finer geographic breakdown might be viewed as being so narrow as to rule out intra-regional variations. To control for a city’s age we adopt a cohort analysis which identifies each city by the period in which it first reached a population of 20,000. To simplify the data presentation we consider only two cohorts: the first consists of all cities reaching 20,000 between 1860 and 1880; the second is made up of cities achieving the same population between 1880 and 1900.[’ ‘I

Urban specialization

1860-1930

Before the Civil War there were relatively few places in the United States that

40

J. PERSKY

AND R. MOSES

would qualify by the standards of today, or even European standards of that time, as cities. In 1860 there were only sixteen cities in the United States with populations greater than 50,000 and only forty-five greater than 20,000.[‘“] Given this overall size distribution of all urban places, it is not surprising that only a few highly specialized places reached a population of any size. Most notable among these was the city of Lowell, Massachusetts with an 1860 population of 37,000. By any standard, Lowell was the epitome of a specialized city, the American proof that specialization was not to be limited to towns and villages.[’ 31 Less specialized, but somewhat larger was Providence, Rhode Island, also with a commitment to textiles.‘15] Other large specialized places in the Northeast included Manchester, New Hampshire also in textiles, and Troy, New York in apparel. At the same time there were signs of specialization elsewhere, especially if one is willing to include the large number of urban residents involved in household piece work. If piece workers are included, such cities as Cincinnati and Utica are classified as specialized in apparel. This small collection of cities of diverse ages hardly lends itself to systematic statistical treatment. These early cities, however, were important for demonstrating the ability of a specialized manufacturing structure to support urban places. In 1880 the share of city population in specialized cities as defined above was 10.4”,,. I’ ‘I This population was located in 20 cities. By 1900 specialized cities account for 12.7”,, of city population, now in 54 cities (Table 1). Thus over a period of twenty years the specialized cities slightly increased their share of population. City growth was not evenly spread over the nation during the period from 1880 to 1900. The North Central region rapidly increased its share of both the number of cities and of city population. On the other hand specialized cities were clearly concentrated in the Northeast region. In the Northeast the city population share of specialized cities was 14.8”,, in 1880 and 18.1’!,, in 1900 (Table 1). All other regions had a considerably lower share. In addition to observing regional differences, we find distinct diflerences in the importance of specialization in cities of different ages. In both 1880 and 1900 specialized cities form a substantially larger share of the population in younger cities (Table 2a). For the 1880-1900 cohort over 30”,, of its population in 1900 was in specialized manufacturing cities. As might be expected, given the generally higher prevalence of specialization in the Northeastern region, 47”” of the 1900 population of that region’s 1880.-1900 cohort was in specialized manufacturing towns. A large proportion of the TABLE 1

Percent c$city population in specialized cities 1880

Region Northeast North Central South West

1900


Fraction of cities

O0total city population’

Fraction of cities

14.8 2.4 8.7 0

15136 2126 3120 015

18.1 6.9 8.9 2.7

35191 12166 6136 l/13

’ Population of all cities greater than 20,000. See text for definition of “specialized”. Source: Computed from U.S. Census Office Compendium of’ the Tenth Census (Washington, 1883) 38&405 and Population Twelfth Census (Washington, D.C. 1901) 438-480.

D.C.

INDUSTRIAL

CITIES

II\( USA 1860 1030

TABLE 2a Percent oj‘ciry

population

in speciuked

indusrrictl cities hv cohorts

I880 o0 total city population

1900 Fraction of cities II,56

1860-1880 cohort’ 188%1900 cohort’

‘0.5

’ Cities which reached 20,fKKl between 2 Cities which reached 20.000 between

1860 and 1880. 1880 and 1900.

I’,, total city population

Fraction of cities

17.9 31.5

X/IO9

I1,!56

Source : Same as Table 1.

TABLE 2b Percent of city population

in specialized

industriLd cities by cohorts in Northeasr

1880 _.____~ __ 9, total city Fraction population of cities 1860-- 1880 cohort I 1880-1900 cohort’

36 1

I Cities which reached 20,000 between ’ Cities which reached 20,000 between

1860 and 1880. 1880 and 1900.

l/23

1900 “,0total city population

Fraction of cities

37.2 47.3

7123 21/45

Source : Same as Table 1,

population in young cities in the Northeast was in specialized cities by 1900. All things considered it would be safe to characterize the specialized cities of the latter part of the nineteenth century as largely a regional phenomenon of the Northeast. Specialization was a process that paralleled the continued growth of that region’s major older cities and the new towns further westward. Given these observations we turn to the central question of whether the populations of specialized cities lagged behind comparable non-specialized cities. Because specialized cities were not randomly distributed across region or cohorts we control for these variables. Without controlling for region and age, the general group of cities is quite disparate. Adjusting for age and region establishes that only those cities with the same starting point are compared. Clearly, we want to know how specialized cities fared in comparison with cities similarly situated in space and time. For every comparison between specialized and other. general cities, we present simple averages for the two groups over time. Because the number of cities considered is held constant, comparing changes in two groups’ average sizes is equivalent to asking whether the share of each group has changed. This seems an appropriate measure of a group’s relative growth. We also present data on standard deviations so that the interested reader can infer the extent to which the growth of a group of cities was concentrated in only a few of these cities. Beginning in 1880 with 20 specialized cities and 79 other cities, we find that the

42

.I. PERSKY

AND

R. MOSES

specialized group moved from an average size of 48,000 to an average size in 1930 of 126,000. This growth was not uniform over the period, but tended to slow in the last two decades (Table 3). At the beginning of the period cities in the general group were much larger, starting from an average of 107,000. By 1930 these cities averaged 390,000. The growth of general cities also slowed down over the period, but this slowdown is less marked than that for the specialized cities. This overview seems to offer some support for the notion that specialized cities were at a disadvantage with respect to other cities. They begin smaller in 1880 and were absolutely, if not relatively, further behind by 1930. Such a comparison is hardly appropriate in view of the diverse nature of the general group of cities. The first control we introduce is to limit the age of the cities in question. If we begin with the cohort of cities that first reached 20,000 population between 1860 and 1880 (cohort 1), the absolute differences between specialized and other cities within the cohort were much less at all dates than the corresponding differences for the totals. Between 1880 and 1930 the two groups of cities moved from almost identical averages of 30,000 to 140,000 for the general group and to 90,000 for the specialized group. Yet, if we viewed the two groups of cities as statistical samples, we would conclude that for both 1920 and 1930 the general group had achieved a significantly larger average size. This result is not surprising because the 1860-1880 cohort for the North Central region contains several rapidly growing cities in its general group, including Minneapolis, Kansas City, and Indianapolis. To the extent that one believes the environment of these cities was quite different from that of the Northeast. it is appropriate to focus more narrowly on the latter region with its relative abundance of specialized cities (seven out of eleven). Proceeding then to consider only those members of the 1860-80 cohort situated in the Northeast, we find that the two groups of cities are more homogeneous. The specialized cities moved from an average size in 1880 of 36,000 up to average sizes TABLE 3 Average populations

All numbers are in thousands. of “specialized”. Total’ General,

1880

Specialized,

1880

1860-1880 cohort2 General Specialized 186e-1880 cohort, Northeast3 General Specialized

fiw specialized

and other cities:

Figures in parentheses

are standard

1880

1890

107.4 (243.7) 48.0 (48.2)

155.1 (328.2) 66.4 (68.3)

30-8 (11.7) 32.4 (10.9)

53.8 (354) 47.4 (19.1)

69.2 (44.5) 61.5 (29.2)

27.8 (7.5) 35.9 (11-8)

39.1 (13.7) 53.2 (19.3)

51.7 (20.8) 70.1 (27.9)

1860-1880 cohort ctrld oldt,r deviations. See text for definition

1910

1920

1930

204.8 265.9 (45 1.O) (608.1) 103.2 83.9 (90.4) (106%)

328.1 (726.4) 117.6 (118.2)

389.5 (893.7) 125.7 (136.9)

77

93.6 (66.7) 75.9 (36.2)

118.3 (87.5) 86.5 (39.2)

139.2 (108.2) 89.6 (44.1)

45

64.7 (30.0) 87.0 (32,2)

79.0 (38.5) 95.1 (35-6)

85.8 (42.6) 94.0 (36.8)

16

1900

’ All cities greater than 20,000 as of 1880. ’ Cities which reached 20,000 between 1860 and 1880. 3 Cities in the Northeast which reached 20.000 between 1860 and 1880. Source : Computed

from the decennial

censuses of the United States.

Number

20

11

7

INDUSTRIAL

CITIES

43

IN USA 1860- 1930

in 1930 of 94,000. The general group started somewhat smaller and ended somewhat smaller, but in no decade would a statistical test suggest a significant difference in size. In making this observation we note that there appears to have been a spurt by the specialized cities in this regional cohort, followed by a sluggish period of perhaps two decades from 1910 to 1930 (see Table 3). Turning to the cohort of cities reaching 20,000 between 1880 and 1900 we lind an even stronger case for suggesting that specialization had little effect on performance. Table 4 suggests that for the cohort as a whole the new specialized cities lagged slightly behind the larger general group. In no decade, however, would these differences appear important. Limiting the data to thosecities ofeach group in the Northeast makes the point even clearer. At this level the absolute difference between averages for the two groups never is larger than 1,500 people.

TABLES

Average populations ,fbr specialized and other r,ities: 1880-I 900 cohort and older All numbers are in thousands. Figures in parentheses are standard deviations. See text for definition of specialized. Total ’ General.

1900

Specialized,

1900

1880-1900 cohort’ General Specialized 1880-1900 cohort, General

1900

1910

1920 _~

1930

119.5 (33.2) 48.8 (6.0)

159.5 (44.6) 64.5

199.1 (53.4) 78.6 (8.3)

241.8 (66.0) 89.7 (9.61)

31.0 (14.6) 30.1 (7.8)

49.6 (47.1) 43.6 (21.8)

658 (76.5) 57.6 (41.0)

89.2 (149.6) 71.0 (56.1)

74

28.5 (7.6) 29.2 (6.7)

38.4 (16.5) 38.8 (11.6)

46.2 (22.0) 46.6 (16.8)

53.6 (27.7) 52.2 (20.8)

34

(7.2)

152 54

35

Northeast3

Specialized

’ All cities greater than 20,000 as of 1880. 2 Cities which reached 20,000 between 1880 and 1900. 3 Cities in the Northeast which reached 20,000 between Source : Computed

Number

from the decennial

21

1880 and 1900.

censuses of the United States.

Expanding the analysis

The pattern presented above can be tested in a number of ways. We conducted four exercises to recompute the basic descriptive statistics along alternative lines. These involved: (1) requiring cities to meet more stringent requirements before labelling them as specialized, (2) focusing more clearly on the emergence of very large cities, (3) extending the data forward in time, and (4) considering the growing importance of suburbanization of large urban areas. Dt$nition of specialization The definition of specialization used above is subject to debate. We explored how increasing the requirements of the definition might affect the statistics displayed. We did this for the following alternative: a

44

J. PERSKY AND R. MOSES

specialized city must have manufacturing employment equal to 25’:; of its total population and more than SO”,;of that manufacturing employment must be a single (two-digit) industry. This definition greatly reduced the number of cities qualifying as specialized in both 1880 and 1900. For the former year only six cities met the requirement (Fall River, Lawrence, Lowell, Lynn, Manchester, and Paterson). For 1900 eight cities met the requirement. Lawrence, Lowell, and Lynn had fallen below the 500, specialization level in manufacturing, while Brockton, Haverhill, Lewiston, New Bedford, and Warwick were added to the list. Even informal inferences are difficult to draw from these small numbers. Table 5 suggests that the average performance of these highly specialized cities did not appear to differ significantly from that of the corresponding larger groups of specialized cities presented in Tables 3 and 4. Again, when we focus on cohorts 1 and 2 of the Northeast, a comparison with the general cities in the same cohort and region hardly demonstrates that the highly specialized cities lagged. For Cohort 1, the four cities involved in the Northeast: Fall River, Lawrence, Lynn, and Paterson, had an average population in 1930 of about 110,000 people in comparison with the sixteen general cities in the same region and cohort with an average population that year of 86,000. Similarly in cohort 2, the four cities in the Northeast, Brockton, Haverhill, Lewiston, and Warwick, reached an average population of 45,000 by 1930 in comparison with a general group average of 54,000.

TABLE5 Average population

for

highly

specialized

cities:

1860-1880

cohort

and 1880-1900

cohort

All numbers in thousands. Figures in parentheses are standard deviations. See text for definition of “highly specialized”. 1880

1890

1900

1910

1920

1930

44.9 (9.9)

62.5 (16.3)

82.2 (22.0)

99.4 (21.3)

106.8 (20.4)

103.0 (22.0)

6

44.4 (6.6)

63.3 (15.8)

85.3 (22.9)

105.0 (20.3)

112.4 (19.3)

110.3 (22.5)

4

44.4 (6.6)

63.3 (15.8)

105.0 (20.3) 70.1 (39.4) 38.5 (14.8)

112.4 (19.3) 79.6 (41.8) 45.2 (17.9)

110.3 (22.5) 78.1 (39.8) 45.4 (14.0)

4

Highly specialized 188&1900 cohort

85.3 (22.9) 56.5 (33.2) 30.6 (9.4)

Highly specialized 188&1900 cohort, Northeast

30.6 (9.4)

38.5 (14.8)

45.2 (17.9)

45.4 (14.0)

Highly specialized 1880 Highly specialized 186@1880 cohort Highly specialized 186&1880 cohort, Northeast Highly specialized 1900

Number

8 4

4

Source : Computed from the decennial censuses of the United States.

Large cities It can be argued that a particular type of city should be judged not so much by its average performance, but rather by the performance of its most successful members. Table 6 presents data on the upper tails of the size distributions of the two cohorts studied here as of 1930. In that year 34 of these 165 cities reached a size betwen 100,000 and 250,000 and another 15 exceeded

INDUSTRIAL

CITIES

45

IN USA IX60 I930

TABLE 6

Shtrre of cohorts

I and 2 greater

thun 100,000

and 250,000.

1930

For each entry the first number in the numerator is cities greater than 100,000: the second is cities greater than 250,000; and the denominator all cities of that type in the region and cohort.

1860-1880 cohort’ General Specialized 1880-1900 cohort’ General Specialized

All regions

Northeast

North Central

South

West

5+0/1 I 2+ 2135

4+0/1 0+0/21

1 +O;’ 2-t I,‘10

0+0/z o+ l.!z

0+0/o 0 + (I/ 1

7+0/16 1-t O/24

6+6/17 3 + Oj29

3+ I!8 4+2/13

2+2;4 I + 2:‘8

18+9/45 9 + 4174

’ Cities which reached 20,000 between ’ Cities which reached 20,000 between

1860-1880. 1880-1900.

Source: Computed from the Bureau of the Census, Population, Stutrs (Washington, D.C. 1931) 18-30.

Vol. I. F$centh

Census

4’rkr

LMc4

250,000. The data suggest that for the nation as a whole the general group of cities did somewhat better in achieving large size. The most spectacular of these was Los Angeles in the second cohort. By 1930 Los Angeles ranked as the fifth largest city in the nation with more than 1,200,OOOpeople. After Los Angeles were a group of metropolises largely in the North Central region. These included Minneapolis, Kansas City, and Indianapolis from the first cohort. It was largely this group of cities which gave the general cities their national growth advantages in the period. By contrast, if we focus on the Northeast the basic pattern of similar growth histories again emerges. Of the 45 cities in the second cohort only one, Yonkers in the general group, exceeded 100,000 by 1930. Although the first cohort understandably did better, the specialized cities, with Paterson, Trenton, Fall River, and Lynn over 100,000, did grow substantially. The largest general city of the Northeast cohort 1, Springfield, Mass., reached a rather modest 150,000 in 1930. At the regional level it would be difficult to conclude that specialized cities had a significantly lower chance of reaching large size. Rather it was difficult for al1 of these cities and especiahy the youngest cohort to move up the ranks of urban places. E.ytension to 1960 It could be argued that large differences in performance between specialized and general cities took considerable time to appear. Tables 7 and 8 extend the basic data of Tables 3 and 4 up to 1960. The results for the most important comparisons, cohorts 1 and 2 in the Northeast, are quite consistent with the earlier data. It is striking, however, how little growth in city populations occurred for these groups. Both specialized and general cities seem to have stagnated. This apparent slowdown may result from much urban growth in the recent period having taken place outside the formal city boundaries. This situation suggested the last extension of the data presented below. Metropolitan area populations Tables 9 and 10 reproduce basic statistics of this study for urban definitions more consistent with current practice. These data are based on metropolitan areas where available. For some cities this implies merging data. For those cities which neither reached metropolitan status by 1960 nor were included in a metropolitan area by that year we collected data on their county if the cities accounted for more than 25’:,,of county population. The result

46 TABLE 7

Average populations,fbr specialized and other cities: 1860-1880 cohort and older,ftir 1940. 1950 and 1960 All numbers are in thousands. Figures in parentheses are standard deviations. Total ’ General,

1880

Specialized,

1880

186&1880 cohort’ General Specialized 186&1880 cohort Northeast3 General Specialized

1940

1950

1960

404.3 (94.6) 125.7 (137.5)

442.0 (10.5) 130.0 (140.0)

444.0 (98.5) 123.09 (125.5)

77

144.6 (115.1) 88.9 (43.5)

165.2 (136.7) 90.9 (45.0)

182.4 (161.3) 87.4 (44.8)

45

84.5 (42.9) 93.1 (37.9)

86.9 (45.2) 93.0 (37.8)

’ All cities greater than 20,ooO as of 1880. ’ Cities which reached 20,000 between 1860 and 1880. 3 Cities in the Northeast which reached 20,000 between Source : Computed

from the decennial

83.4 (46.8) 87.3 (37.16)

Number

20

11

16 7

1860 and 1880.

censuses of the United States.

TABLE 8

Average population,for specialized and other cities: 1880-1900 cohort and olderfor 1940,1950,1960 All numbers are in thousands. Figures in parentheses are standard deviations. See text for definition of specialized. Total’ General,

1900

Specialized,

1900

188&1900 cohort2 General Specialized 1880-1900 Cohort, General

1940 ___~_ _.

1950

1960

253.2 (699.9) 89.9 (96.2)

283.3 (750.2) 95.6 ( 100.5)

296.5 (747.2) 95.8 (93.4)

152

97.4 (180.4) 71.0 (55.9)

120.1 (240.7) 77.3 (64.8)

145.4 (311.7) 81.1 (67.7)

74

54.0 (28.5) 51.4 (20.4)

57.3 (31.1) 53.2 (21.0)

58.5 (36.8) 53.9 (22.1)

54

35

Northeast3

Specialized

’ All cities greater than 20,000 as of 1880. ’ Cities which reached 20,000 between 1880 and 1900. 3 Cities in the Northeast which reached 20,000 between Source : Computed

Number

from the decennial

1880 and 1900

censuses of the United States.

24 21

INDUSTRIAL

CITIES

47

IN USA 1860 1930

TABLE 9 .4 veruge populations

jor specialized and other metropoliturl areus: cohort and older,fbr 1900, 1930 and 1960

186&1880

All numbers in thousands. Figures in parentheses are standard deviations. Each group includes data for counties containing those cities not defined as metropolitan areas in 1960 if city accounted for more than 25”6 of county population. SMSA’s with more than one city take their specialization designation from their largest city. Total’ General,

1900 1880

Specialized,

1880

186&1880 cohort* General Specialized

1930

I960

Number

343.4 (572.4) 192.3 (254.7)

665.8 (1.181.5) 333.4 (491.9)

1.006.4 (1,631,3) 441.4 (611.3)

65

135.4 (85.9) 119.7 (54.1)

248.2 (179.5) 221.1

409.6 (3236) 317-2

34

’ All cities greater than 20,000 as of 1880. ’ Cities which reached 20,000 between 1860 and 1880. 3 Cities in the Northeast which reached 20,000 between

15

8

1860 and 1880.

Source: Growth

Computed from data presented in William Alonso and Elliott Medrich, Spontaneous Centers in Twentieth-Century American Urbanization, in N. Hansen (ed.) Growth centers in regional economic development (New York 1972) 25&265 and data from the decennial censuses of the United States.

of these geographic redefinitions is a list of 156 places. The definition of specialization remained in terms of the central city. The data suggest that for each grouping of places, specialized areas are at least as prevalent using this new geographical breakdown as they were for individual cities. For the two basic groups of interest, cohorts 1 and 2 in the Northeast, the 1960 data show that for the latter the general group was larger than the specialized group with an average population of 167,000 as compared with a specialized average of 144,000. For the earlier group, however, the specialized cities on average exceeded the general group by 371,000 to 283,000. Conclusions

In the light of these results, serious reconsideration of the theory of specialized cities would seem to be in order. The basic argument of Jane Jacobs and other proponents of diversified industrial structures is founded on a presumption that a city is the fundamental unit of economic activity. There is little or no room in these theories for the equilibrating potential of the mobility ofcapital and labour. By taking the specialized city out of its regional and national context they emphasize the likely risks of specialization. Implicitly they assume that the human and physical capital of such cities will be of little value once their specialized base has begun to erode. For an isolated island community or a

48

J. PERSKY

Average

AND R. MOSES

TABLE 10 populations,f& specialized and other metropolitan areas: cohort and older ,for 1900. 1930 and 1960

1880-1900

All numbers in thousands. Figures in parentheses are standard deviations. Each group includes data for counties containing those cities not defined as metropolitan areas in 1960 if city accounted for more than 259; of county population. SMSA’s with more than one city take their specialization designation from their largest city. Total ’ General,

1900

Specialized,

1900

1880-1900 cohort’ General Specialized 1880-1900 cohort, Northeast’ General Specialized

1900

1930

224.0 (444.9) 111.7 (169.0)

449.6 (931.1) 212.5 (328.2)

121.2 (1.391.6) 288.5 (712.7)

118

71.6 (37.5) 73.2 (45.8)

180.0 (324.4) 146.9 (117.2)

386.7 (946.9) 200.5 ( 165.9)

51

84.4 (19.8) 756 (57.8)

119.1 (31.7) 123.5 (101.8)

167.3 (49.9) 143.9 (120.6)

’ All cities greater than 20,000 as of 1900. ’ Cities which reached 20,000 between 1880 and 1900. 3 Cities in the Northeast which reached 20,000 between Source: See Table 9.

1960

Number

38

25

7 13

1880 and 1900.

Nevada mining town this makes sense. For a city closely interconnected in a system of cities, the result is not so obvious. The traditional critique of specialization has put too great an emphasis on the ability of a city to generate new work internally. This criteria of innovation may be appropriate for considering the fate of a nation or even an international metropolis like New York. At the level of most actual specialized cities, components of a mixed system of urban places, self-generated growth is likely to be of less significance. Although there is considerable variation in urban growth rates, this variation is not systematically related to the variation in the degree ofearly specialization in cities’ histories. To summarize, our analysis of late nineteenth and twentieth century population data is not supportive of urban growth theories that emphasize the dangers of specialization. When appropriate controls for region and city age are made, highly specialized cities performed almost as well as the general population of all cities. This result is surprising in light of the considerable literature that attempts to explain the poor performance of specialized cities. Although this paper does not attempt a theoretical reconstruction, we suspect that more attention should be paid to the pressures that arbitrage labour and capital markets within regional systems of cities.

Department of Economics University of Illinois at Chicago

INDUSTRIAL

CITIES

IN USA 1860 1930

49

Notes

[II

For a discussion of the National Resources Planning Board position and earlier work on specialization see A. L. Rodgers, Some aspects of industrial diversification in the United States, Economic Grogruph.v 33 (1957) 16-30 Rcvic~ PI B. Chinitz, Contrasts in agglomeration: New York and Pittsburgh, .4/ncricrr/r Eco~~o~~~ic~ 51 (1961) 279-89 [31 J. Jacobs, 7’17~ecot7om~~ o/‘citirs (New York 1969) [4; In these matters she is an important forerunner of the product-cycle theories of location. These theories also emphasize the flow of activity from more diversified places to more specialized ones and the selective growth of towns in North American [5] E. Muller. Regional urbanization regions, Jourtxd qf’Historica/ Gq4w7pl7~~ 3 ( 1977) 21 39. 36. Muller’s quote is with particular reference to capital mobility among the cities of Wisconsin C61For a discussion of the urban-ratchet effect see W. Thompson, Prefucc~ to urhm7 eco77or77ic~.r (Baltimore 1965) 21-24. Despite thisdiscussion Thompson feels that “industrial diversitication is probably the most efficacious way to achieve growth stability”, p. 164 New York and [71For an empirical test of this point see G. Carlino, Contrasts in agglomeration: Pittsburgh reconsidered, Urban Studies 17 (1980) 343-51. Carlino concludes that “the forces of specialization do not offset the forces of economies of scale due to agglomeration”, p. 349 PI For a recent study of Lowell that emphasizes the impact of the mills on the nature of the city’s community see T. Dublin, B’onrm at M.ork (New York 1979). For a broader introduction to the history of industrial cities in the United States see J. Vance. Tlli,ssc,c~r7rof’n7~77 (New York 1977) 32449 [91See U.S. Census for regional definitions [lOIThis argument is persuasively made by D. Meyer. Emergence of the American manufacturing belt: An Interpretation, Journal of' Historicul Geographic 9 (1983) 145-74 1111 See the Appendix for the speciahzed cities in each cohort, Note that for the Appendix the classification of cohort 1 and earlier cities is determined by their industrial structure in 1900. Data presented for these cities in the tables, however, are determined by their industrial structure in 1880. See Appendix footnotes for a reconciliation of these two alternatives Data from Populutior7 qf tke United States it7 1860; the eighth cet7sus (Washington 1865) 1121 10,000 “millhands” out of about 13,000 “hands employed” [I31In 1860 Lowell had approximately in manufacturing. The total population of Lowell in that year was 37.000. See Dubin O/J. ci/. 230 and Census 1860 1141By 1880 Providence no longer meets the specialization requirements set above and is thus not included in the Appendix which presents cities specialized as of 1900. This is also the case for Cincinnati and Utica mentioned below [ 151 “City population” refers to the total population of all cities with more than 20,000 people

50

J. PERSKY AND R. MOSES APPENDIX Cities Specialized in 1900’

Gil

1’

I .- Akron. Ohio 2. Allentown. Pa. 3. Altoona, Pa. 4. Amsterdam, N.Y. 5. Augusta, Ga. ’ 6. Bayonne, N.J.’ 7. Birmingham. Ala. 8. Brockton, Mass. 9. Chattanooga, Tenn. 10. Cohoes, N.Y.5 11. Columbia, S.C. 12. Erie, Pa.? 13. Fall River. Mass. 14. Fitchburg, Mass.’ 15. Grand Rapids, Mich. 16. Hamilton. Ohio 17. Haverhill, Mass.’ 18. Holyoke. Mass. 19. Jamestown, N.Y. 20. Johnstown, Pa. 21. Joliet, 111.’ 22. Kansas City, Kansas” 23. Lawrence, Mass. 24. Lewiston, Me. 25. Lowell, Mass.” 26. Lynn, Ma~s.~ 27. McKeesport. Pa.‘l 28. Manchester, N.H.” 29. Meriden, Conn. 30. Muncie. Ind. 31. New Bedford, Mass.” 32. New Britain, Conn. 33. Newcastle. Pa. 34. North Adams, Mass.’ 35. Oshkosh, Wis. 36. Paterson, N.J. 37. Pawtucket, R.1.” 3X. Petersburg, Va.” 39. Pittsburgh, Pa.’ 40. Pittsfield. Mass. 41. Pueblo, Colo. 42. Richmond, Va.” 43. Salem. M~ss.“.~ 44. Sheboygan. Wis. 45. South Bend. Ind. 46. Springfield, Ohio 47. Taunton. Mass.‘+r 48. Trenton. N.J. 49. Troy. N.Y.‘.J 50. Warwick, R.1.’

IX60 1880 CCd?O,./

*

*

*

*

* *

* * *

INXO- I VOO C0l10r.l Itdustriul md SIC Code * Rubber * Textiles * Rubber * Textiles Textiles Petroleum Primary Metals Leather Fabricated Metals Textiles Textiles Fabricated Metals Textiles * Textiles Furniture * Fabricated Metals * Leather Paper Textiles Textiles Textiles Food Textiles * Textiles Textiles Leather * Primary Metals Textiles * Stone, Clay & Glass * Stone, Clay & Glass Textiles * Fabricated Metals * Primary Metals * Textiles * Lumber Textiles * Textiles Tobacco Primary Metals * Textiles * Primary Metals Tobacco Leather * Furniture * Transport Equipment Machinery Textiles Stone. Clay & Glass Apparel * Textiles

(30) (22) (30) (22) (22) (29) (33) (31) (34) (22) (22) (34) (22) (32) (75) (34) (31) (26) (22) (22) (22) (20) (22) (23) (32) (31) (33) (22) (32) (32) (22) (34) (33) (22) (24) (22) (22) (21) (33) (22) (33) (21) (31) (15) (37) (35) (22) (32) (33) (21)

INDUSTRIAL

51. 52. 53. 54.

Watertown, N.Y. Woonsocket. R.I.’ Youngstown, Ohio Zanesville. Ohio

CITIES

IN l’SA

* * * *

IX60

1930

Fabricated Metals Textiles Primary Metals

51

(34) (22) (23) (32)

Five cities specialized in 1880 wet-e no longer specialized in 1900. Portland, Me. (30). Syracuse, N.Y. (23) and Utica, N.Y. (23) from the pre-1860 group and Auburn. N.Y. (35) and Wheeling, West Virginia, (33) from Cohort I. Augusta, Ga., Erie, Pa.. and Taunton. Mass. are both in Cohort 1, but are there classed as general cities since in 1880 they did not meet the criteria of specialization. These cities were at least 20,000 by 1860. From available evidence neither Pittsburgh nor Salem was specialized when it first reached this population. New Bedford is not specialized in textiles before 1900, but of course it had been heavily committed to fishing. These sixteen cities are not primary cities of an SMSA in 1960 or (if less than 50,000) primary cities of well identified counties. Ten of these are secondary cities in another SMSA, while the other six account for only ;I small share of their county (less than X’,,).