History and Industry Location: The Case of the Manufacturing Belt

Author(s): Paul Krugman

Source: The American Economic Review , May, 1991, Vol. 81, No. 2, Papers and
Proceedings of the Hundred and Third Annual Meeting of the American Economic
Association (May, 1991), pp. 80-83
Published by: American Economic Association

Stable URL: https://www.jstor.org/stable/2006830

REFERENCES
Linked references are available on JSTOR for this article:
https://www.jstor.org/stable/2006830?seq=1&cid=pdf-
reference#references_tab_contents
You may need to log in to JSTOR to access the linked references.

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide
range of content in a trusted digital archive. We use information technology and tools to increase productivity and
facilitate new forms of scholarship. For more information about JSTOR, please contact support@jstor.org.
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at
https://about.jstor.org/terms

is collaborating with JSTOR to digitize, preserve and extend access to The American Economic
Review

This content downloaded from

200.130.19.182 on Thu, 29 Oct 2020 15:22:21 UTC
All use subject to https://about.jstor.org/terms
 History and Industry Location:
The Case of the Manufacturing Belt

By PAUL KRUGMAN*

If there is one single area of economics in mineral production shifted far to the West.
which path dependence is unmistakable, it In 1870, the Northeast and East North Cen-
is in economic geography -the location of tral regions (within which the emerging
production in space. The long shadow cast manufacturing belt lay) accounted for 44
by history over location is apparent at all percent of U.S. "resource extraction" em-
scales, from the smallest to the largest- ployment (agriculture, mining, forestry,
from the cluster of costume jewelry firms in fisheries). By 1910, this share had already
Providence to the concentration of 60 mil- fallen to 27 percent; yet these regions still
lion people in the Northeast Corridor. accounted for 70 percent of manufacturing
This paper illustrates path dependence in employment. And whereas the belt's share
economic geography by describing a particu- of manufacturing employment understates
lar historical example, the persistence of its manufacturing dominance, its share of
the U.S. "manufacturing belt," and a simple resource employment overstates its resource
model that helps make sense of that base, since much of the agriculture in or
example. near the manufacturing belt took place less
because of the suitability of the land than
L. The U.S. Manufacturing Belt because of proximity to urban centers.
The manufacturing belt's persistent domi-
Early in this century, geographers noted nance evidently reflects some kind of exter-
that the great bulk of U.S. manufacturing nal economies. But what was the nature of
was concentrated in a relatively small part these external economies? I will sketch out
of the Northeast and the eastern part of the a simple model that surely does not capture
Midwest. This manufacturing belt took the full story, but is strongly suggestive of
shape in the second half of the nineteenth the kind of explanation that is needed.
century, and proved remarkably persistent.
Harvey Perloff (1960) estimated that as late IX. A Core-Periphery Model'
as 1957 the manufacturing belt still con-
tained 64 percent of U.S. manufacturing A core-periphery pattern like that of in-
employment-only slightly reduced from its dustrial America can emerge from the inter-
74 percent share at the turn of the century. action of increasing returns, transportation
Even this number understates the domi- costs, and demand. Given sufficiently strong
nance of the belt, because during its heyday economies of scale, each manufacturer
most manufacturing outside it consisted ei- wants to serve the national market from a
ther of processing of primary products or of single location. To minimize transportation
production for a very local market. That is, costs, she chooses a location with large local
the manufacturing belt contained virtually
all manufacturing that did not need either
to be close to the consumer or close to 'This paper presents only a sketch of a model. It
specific natural resources. will be apparent that this sketch is sloppy about a
The manufacturing belt persisted even as number of issues, including: what is the market struc-
ture in manufacturing; what happens to profits, if any;
the center of gravity of agricultural and
and what resources are used in both fixed costs and
transportation. It is possible to derive similar results in
a fully specified general equilibrium monopolistic com-
* Department of Economics, MIT, Cambridge, MA petition model (see my 1991 article). I adopt the more
02139. ad hoc approach here for ease of exposition.

80

This content downloaded from

200.130.19.182 on Thu, 29 Oct 2020 15:22:21 UTC
All use subject to https://about.jstor.org/terms
VOL. 81 NO. 2 PATH DEPENDENCE IN ECONOMICS 81

demand. But local demand will be largely Share of West

precisely where the majority of manufactur- in population
ers choose to locate. Thus there is a circu-
3 p
larity that tends to keep a manufacturing
core in existence once it is established. (This
is not an original story: it is more or less 2 M
explicit in the work of such geographers as M
Allan Pred, 1966, and David Myers, 1983.)
Imagine a country in which there are only
p1
two possible locations of production, East
and West, and two kinds of production.
Agricultural goods are produced using a
location-specific factor (land), and as a re-
sult the agricultural population is exoge- Share of West in manufacturing
nously divided between the locations; we
FIGURE 1
assume that the division is 50-50.
Manufactured goods (of which there are
many symmetric varieties) can be produced of MM: no western production for low
in either or both locations. If a given manu- western population, production propor-
factured good is produced in only one loca- tional to population for intermediate levels,
tion, transportation costs must be incurred no eastern production if the West is big
to service the other market. On the other enough.
hand, if the good is to be produced in both The more manufacturing that takes place
locations, an additional fixed set-up cost is in West, the larger the western share of the
incurred. The manufacturing labor force in total population; but not in proportion, be-
each location is proportional to manufactur- cause some people are immobile farmers.
ing production in that location. Thus PP is upward sloping, but flatter than
Finally, assume that the demand for each the 450 line.
manufactured good in each location is Suppose that manufacturing production
strictly proportional to that location's popu- adjusts gradually toward its equilibrium
lation. level. Then the dynamics are illustrated by
The basic workings of the model can then the arrows in Figure 1. There are three
be illustrated by Figure 1. On the horizontal stable equilibria: manufacturing may be
axis we measure the share of the manufac- concentrated in either location, at 1 or 3, or
turing labor force employed in West, on the it may be equally divided, at 2.
vertical axis the share of West in the total The picture need not look like this: a
population. The line MM represents the core-periphery geography may be unsustain-
dependence of the distribution of manufac- able because the "centripetal" pull of a
turing on the distribution of population; the manufacturing core is weaker than the
line PP the converse effect of manufactur- "centrifugal" pull of the dispersed agricul-
ing on population distribution. tural sector. We can easily derive a neces-
If West has a small share of the popula- sary condition for concentration of manu-
tion, it will not be worth incurring the fixedfacturing production in one location. Con-
costs of establishing a manufacturing facilitysider a typical manufacturing firm with total
there; it is cheaper to serve the market from sales of x; let iT be the proportion of the
facilities in East. Conversely, if West has a population that is employed in manufac-
large share of the population, it is not worth tures; let F be the fixed cost that the firm
producing manufacturers in East. A suffi- must incur to produce in both locations;
ciently equal division of population might, and let t be the transportation cost per unit
however, lead manufactures to produce lo- shipped between locations. Now we ask: is a
cally for both markets. Putting these obser- situation with all manufacturing concen-
vations together, we get the illustrated shape trated in East an equilibrium?

This content downloaded from

200.130.19.182 on Thu, 29 Oct 2020 15:22:21 UTC
All use subject to https://about.jstor.org/terms
 82 AEA PAPERS AND PROCEEDINGS MAY 1991

With all manufacturing in East, West has ing belt-about why Detroit emerged as the
a share of total population equal to only automotive center, New York as the gar-
(1 - 17)/2. The transportation cost of serv- ment center, Grand Rapids as the furniture
ing this market from East for a typical man- center, etc. Yet it surely captures an impor-
ufacturer is therefore tx(1 - 17)/2. The cost tant aspect of what happened. And it also
of setting up a plant in West is F. So a contains elements (increasing returns at the
concentration of production in East, once level of individual firms, and external
established, will persist as long as F> economies resulting from the interaction of
tx(1 - 17)/2. If this criterion is not met, there
these firms' decisions) that will reappear as
is no path dependence: the long-run geogra- one further elaborates the story.
phy of manufactures will follow that of agri-
culture.2 III. Further Thoughts
We can immediately see that whether ge-
ography is path dependent is determined by The case of the U.S. manufacturing belt
three parameters: large F, that is, suffi- is of substantial interest in its own right.
ciently strong economies of scale; small t, The rise and persistence of that belt is an
that is, sufficiently low costs of transporta- important yet much neglected aspect of U.S.
tion; and large iT, that is, a sufficiently large economic history. More important than its
share of "footloose" production not tied immediate significance, however, is what the
down by natural resources. history of manufacturing location says about
We can now tell a stylized story of the the nature of our economy in general. And
emergence of the manufacturing belt. In the what it says is Nicholas Kaldor (1972), Paul
early United States, with its primarily agri- David (1985), and Brian Arthur (1989) were
cultural population, where manufacturing right-that increasing returns and cumula-
was marked by few scale economies, and tive processes are pervasive and give an
where transportation was costly, no strong often decisive role to historical accident.
geographical concentration could occur. As It is also interesting that the story of the
the country began its industrial transition, manufacturing belt reaches back to the
manufacturing arose in areas that contained mid-nineteenth century. It is common to
most of the agricultural population outside argue that external economies and cumula-
the South. During the second half of the tive processes have become more important
nineteenth century, however, manufacturing in recent decades because of the growing
economies of scale increased, transporta- importance of high technology. The geo-
tion costs fell, and the share of the popula- graphical concentration of manufacturing in
tion in nonagricultural occupations rose. the United States took shape, however, in
The result was that the initial advantage of the age of steam, not that of microproces-
the manufacturing belt was locked in. Even sors. So it is not simply true that our econ-
though new land and new resources were omy is not now well described by the con-
exploited to the West and even though slav- ventional constant returns model. It never
ery ended, for three-quarters of a century was.
the pull of the established manufactured
areas was strong enough to keep the manu-
facturing core virtually intact. REFERENCES
Of course this story oversimplifies. In par-
ticular, it says nothing about the sources of Arthur, Brian, " Positive Feedbacks in the
local specialization within the manufactur- Economy," Scientific American, February
1989, 262, 92-99.
David, Paul, "Clio and the Economics of
2This is a necessary condition for concentration. A
QWERTY," American Economic Review
sufficient condition for concentration of manufacturers
production in one location is F> tx /2. In this case,
Proceedings, May 1985, 75, 332-37.
the 450 section of MM disappears, and it becomes a Kaldor, Nicholas, "The Irrelevance of Equi-
horizontal line at an equal population division. librium Economics," Economic Journal,

This content downloaded from

200.130.19.182 on Thu, 29 Oct 2020 15:22:21 UTC
All use subject to https://about.jstor.org/terms
VOL. 81 NO. 2 PATH DEPENDENCE IN ECONOMICS 83

December 1972, 82, 1237-55. 1983, 9, 145-74.

Krugman, Paul, "Increasing Returns and Perloff, Harvey, Regions, Resources, and Eco-
Economic Geography," Journal of Politi- nomic Growth, Baltimore: Johns Hopkins
cal Economy, forthcoming 1991. University Press, 1960.
Myers, David, "Emergence of the American Pred, Allan, The Spatial Dynamics of U.S.
Manufacturing Belt: An Interpretation," Urban-Industrial Growth, 1800-1914,
Journal of Historical Geography, April Cambridge: MIT Press, 1966.

This content downloaded from

200.130.19.182 on Thu, 29 f on Thu, 01 Jan 1976 12:34:56 UTC
All use subject to https://about.jstor.org/terms