(

Urban Economics
FIFTH EDITION

Arthur O'Sullivan
Departme11t of Economics
Lewis & Clark College

LIBROS
COMPAÑIA
Calle 19 No. 3- 16 • Local 104
1ibrosyco 111 p@ hot rna i l .corn
Teleíax: 341 71 00 • Bo¡;otá. D.C.

B McGraw-Hill
•irwin
Boston Burr Ridge, IL Dubuque, IA Madison, WI New York San Francisco St. Louis
Bangkok Bogotá Caracas Kuala Lumpur Lisbon London Ma::lrid Mexico City
Milan Montreal New Delhi Santiago Seoul Singapore Sydney Taipei Toronto
 CHAPTER 3

Big and Small Cities

I wil/ tell the story as I go a long ofsmall cities no less than of

g rec//. Most of titase which were great once are small today;
and those which in my own lifetime ha ve grown to greatness.
were small enough in the old days.
-HERODONTUS

As we saw in Chapter 2, cities exist because sorne activities are subject to

economies of scale. Trading cities develop as a result of scale economies in trans-
portation, while factory cities develop as a result of scale economies in production.
In both cases, the scale economies are internal to the firm; an increase in the firm's
output decreased its average production cost. Based on this simple analysis, we can
predict lhe development of two sorts of small cities-small trading centers with a
few firms at a marketplace, and company towns with ali workers employed in a
single factory.
In this chapter, we'll see why most cities are not small trading posts or one-
factory towns but instead are large and diverse collections of econornic activity. When
the production cost of one firm decreases as the output of another firm increases, we
say the firm experiences externa) economies of scale. The word "externa!" indicates
that one firm benefits from the decisions made by olher firms, decisions lhat are
externa! to the firm. These positive spillovers cause firms to cluster in cities, leading
to the large agglomerations of employment. The extemal economies increase labor
productivity, allowing firms to pay the higher wages necessary to attract workers to
large cities, where comrnuting costs are relativcly high. Many of the ideas in this
chapter are discussed in Jacobs (1969).
One of the purposes of this chapter is to provide sorne insights into why cities
vary in size and scope, addressing the question Why aren't ali cities the same size?
As shown in Figure 3-1, there is substantial variation in city sizes. Panel A shows
the size distribution of the 50 largest urbanized areas. At the upper end, there is
one area (New York) with a population above 16 million, two areas (Los Angeles
and Chicago) with populations between 6 million and 12 million, five areas with

39
40 Part l Market Forces in the Development of Cities

population between 3 million and 6 million, and 22 areas with populations between
1.5 million and 3 million. In other words, the smaller the urbanized area, the more
numerous the areas of similar size. Panel B of Figure 3-1 shows the size distribution
for the 5 lst through the 396th urbanized areas.

FIGURE 3-1 Size Distribution of U.S. Urbanized Areas

16

12
,...._
"'c
~
g
c 8
.~
.§
~
p.
o

o-1.1..L1..1.J...L1...1.Ll.LJ.J.J..Ll..U.L..L.U.U.UJ..LLU..U..U....UL.LLJ..LJ..LJL.L.J..L.Ll..1.J...Ll...l..LJ..LJL.L.J..L.Ll..1.J...L1...l.Ll.LJ.LJ..LJU..U..l..U..l.J..L.l.LI.J.J..LJ.J...L1..1.J.J..U...
5 11 17 23 29 35 41 47
Rank of urbanized area
(A) 50 largest urbanized areas

600
~

"'
-o
§
¿""' 400
e
.Q
.§
~
& 200

o 100 200 300 396

Rank of urbanized area
(B) 5 lst through 396th urbanized areas
Chapter 3 Big and Small Ci1ies 41

LOCALIZATION ECONOMIES AND

INDUSTRY CLUSTERS

One of the puzzling features of an urban economy is the tendency of firms produc-
ing the same product to locate close to one another. The clustering is puzzling be-
cause dispersing into separate territories would reduce competition for workers and
perhaps bring firms closer to their dispersed customers. There are sorne subtle ben-
efits from clustering, and for many industries these benefits dominate the more
obvious costs. We'll explore three types of benefits: sharing the suppliers of inter-
mediate inputs, sharing a pool of labor, and sharing information. In ali three cases,
the production cost of an individual firm decreases as the total output of the industry
cluster increases. The label for these external economies is "localization economies,"
indicating that the cost savings occur only for local firms, (i.e., firms in the cluster).
Our discussion of localization economies focuses on a firm's decision between
locating in an industry cluster or atan isolated site. We'll ignore the other facets of
the location decision, leaving the discussion of thesc other factors for later chapters.
We'll also assume initially that the firms export thcir goods outsidc the city. Later in
the chapter, we' 11 consider the decisions of firms producing products for consumption
within the city.

Sharing Input Suppliers

Sorne industry clusters occur because firms in a particular industry buy an intermedi-
ate input from the same supplier. Firms will cluster around a.common input supplier
if two conditions are satisfied:

1. The input demand of an individual firm is not large enough to exploit the scale
economies in the production of the intermediate input.
2. Transportation costs are relatively high. If demanders and supplier interact in
the design or fabrication of the intermediate input, face-to-face contact between
buyer and seller is necessary, and proximity to the input supplier is important.
Similarly, ifthe intermediate input is bulky, fragile, or must be delivered quickly,
proximity is importan!.

Vernon ( 1972) uses the Manhattan dressmaking industry to il lustrate localization

economies from sharing the supplier of an intermediate input. Because the demand
for high-fashion dresses is unpredictable, a dressmaking firm cannot commit itself
to large-scale production of a given type of dress, but must be prepared to quickly
change to another design if the first dress is unsuccessful. The firm must be small and
agile, with its manager carefully monitoring the dress market, the design process,
and the production processes. For the small and agile firm, specialized labor and
machinery-designed to produce output on a large scale-are not feasible. The
dressmaker buys sorne of its intermediate inputs from firms specializing in those
inputs.
42 Part 1 Market Forces in the Development of Cities

One of the intermediate inputs to dressmaking is buttons, a good that is subject

to large economies of scale. A large buttonmaker can use specialized labor and large
machines to produce buttons at a relatively low cost. Suppose that it takes l Osmall
dressmakers to generate sufficient button demand to exhaust the scale economies
in button production. By sharing a buttonmaker, the 1O dressmakers can get their
buttons ata lower cost. It is efficient to share a buttonmaker because there is a scale-
economy mismatch: the scale economies in dressmaking are small relative to the
scale economies in buttonmaking.
Should dressmakers cluster around the button firm, or could they share the
buttonmaker by ordering buttons out of a catalog? Dressmakers locate near the
button producer because each dress design requires a unique button design, and
dressmakers must supervise the design and fabrication of buttons for a particular
dress with face-to-face contact. In contrast with many other intermediate inputs, the
buttons for a high-fashion dress are not standardized, so they cannot be ordered out of
a catalog. PhysicaJ proximity to the input supplier is required, causing dressmakers
to cluster around the buttonmaker.
There are many other examples of clusters that result from scale economies in
the provision of interrnediate inputs.

1. Corporate headquarters produce a wide variety of outputs. An executive may

supervise the development of a new advertising campaign one week, pick a location
for a new plant the next week, and develop a strategy to fend off a lawsuit the follow-
ing week. Most corporations hire advertising firms because the scale economies in
producing advertising campaigns are large relative to the advertising demand of an
individual corporation. Corporate executives need to be close to the advertising firm
because the executives assist in the design of the advertising campaign; face time is
required. In general, the clustering of corporate headquarters allows corporations to
exploit scale economies in the production of intermediate goods (advertising, eco-
nomic consultants, legal services) for which face time is an important part of the
production process.
2. Firms producing new high-technology goods face uncertain demand for their
products. Instead of producing ali of their own components, they purchase electronic
parts from firms that can exploit scale economies in producing these parts. The firms
interact with their suppliers in the design and fabrication of the components, and must
locate close enough to facilitate frequent face time. Although a high-technology firm
<loes not need to locate close to its supplier of nuts and bolts (standardized inputs that
can be ordered from a catalog), there are substantial benefits from locating close to
its supplier of nonstandardized electronic parts. In addition, high-technology firms
exploit scale economies in product testing by sharing firms that provide testing
facilities, locating close enough to tap the testing facilities at top speed.
3. Newspapers and magazines produce outputs (text and illustrations) that change
in unpredictable ways from day to day. In producing their products, publishers
call on a wide variety of experts, including information sources and illustrators. If
each publisher uses an expert on Armenian history for one story per year, it would
be inefficient for each publisher to hire a full-time Armenian expert. Instead, the
Chaptcr 3 Big and Small Citics 43

publishers may share a single expert, sharing the cost of acquiring specialized in-
formation. Similarly, publishers occasionally need special illustrations, and call on
graphic design firms to do them. Publishers cluster around organizations that pro-
vide expert information (libraries, research institutes, universities) and illustrations
(graphic design firms).
There is evidence of clustering based on input sharing. Rosenthal and Strange
(2000a) use data on inve ntories (finished but unsold goods) to identify industries
that could benefit from input sharing. We've seen that if a firm faces uncertain and
rapidly changing demand for its produce, it will be small and agile, and will have
an incentive to share input suppliers. Such a firm is also likely to maintain a small
inventory because a larger inventory exposes the firm to the damaging effects of
sudden changes in demand. The study found that industries with relatively small
inventories are more likely to cluster.

Sharing a Labor Pool: Varying Demand for Labor

A second type of localization economy results from the uncertainty and variabi lity of
demand for labor. A firm faces two sorts of questions about its future labor demands:
l. How many workcrs will wc hire?
2. What sort of labor skills will we need in our workcrs?
If a firm is uncertain about tbe quantity of workcrs it will hire or the skills of its
workforce, it will have an incentive to cluster around other firms and draw from a
common labor pool.
Consider first the issue of the quantity of labor demanded. Consideran indus-
try with rapidly changing production proccsscs and product demands, such as the
computer industry. Every year, computer finns introduce new products; sorne are
big sellers, and othcrs fail. Eventually, the workers in the unsuccessful firms will
move to the successful oncs. A cluster of computer firms facilitates the transfcr of
workers for two reasons. First, workers in the cluster have relatively low job scarch
costs because (1) information about job openings is spread through informal chan-
nels (casual conversation at restaurants, bowling alleys. and baseball games) and
(2) prospective employers are nearby, making formal job searches relativcly easy.
Second, because of the physical proximity of employers, moving costs are relatively
low: workers can easily switch to a different firm in the samecity. Firms in the cluster
of computer firms can draw from the labor pool to quickly fil! their job vacancics
and increase production.
The television industry also benefits from labor-market economies. In a given
season, sorne television programs are hits and others are failures. When it becomes
clear which programs will be discontinued, actors and technicians move from the
unsuccessful programs to the successful ones. The concentrations of the television
industry in Los Angeles and New York facilitate the transfer of laborers from one
firm to another.
Table 3-1 provides a simple example of the effects of labor pooling. Consider
a firm that must locatc either in an isolated si te (far from other firms in the industry)
44 Part 1 Market Forces in the Development of Cities

TABLE 3-1 Expected lncome in Cluster and at Isolated Site

Switch Probability
Wage Cost of Switch Expected Net Income
lsolated site $20 $8 1/2 $16 = (l/2). 20 + ( 1/2) . (20 - 8)
Industry cluster 16 o l/2 $16 = (1/2). 16+ (1 /2). (16-0)

orinan industry cluster. Although the firm is guaranteed success this year, there is a
50 percent chance that it will fail next year. The prevailing wage at the isolated site
is $20: the firm will pay its workers $20 this year, but has a 50 percent chance of
firing them next year. If a worker is fired, he or she can find another job, but the cost
of finding a job in another city and moving there is $8. In the first row of Table 3-1,
the expected net income next year for a worker in the isolated site is $16; there is
a 50 percent chance of getting $20 andan equal chance of getting only $12, so the
average or expected net income is $16.
The alternative to the isolated site is locating close to other firms in the indus-
try. Suppose that the probability of success in the industry cluster is the same as
the probability of success at the isolated site (50 percent). The only difference is
that workers who live near the cluster have lower switching costs. To simplify the
example, suppose the cost of switching fi.rms is zero: if a firm goes out of business
next year, the worker can instantly switch firms at zero cost. The cost is zero because
workers in the cluster have information about job openings in other firms , and can
change jobs without changing their homes.
In equilibrium, workers will be indifferent between working in the cluster and
working at the isolated site. At what wage in the cluster will they be indifferent?
Given the $20 wage at the isolated site and the cost of switching jobs, the expected
income at the isolated site is only $16. Because there are no switching costs in the
cluster, a wage in the cluster of only $16 will make the worker indifferent between
the two sites. Although the worker still has a 50 percent chance oflosing her $16 job,
she can costlessly switch to another job paying $ 16. In the cluster, a $16 income is
a sure thing. Of course, if there were sorne small switching cost in the cluster, the
wage in the cluster would be greater than $16, but less than $20.
Although a wage of $16 in the cluster makes workers indifferent between the
cluster and the isolated site, firms are not indifferent between the two possible
production si tes. Faced with a wage of $20 at the isolated si te and $16 in the cluster,
the firms will choose the cluster. By clustering and providing workers with a pool
of employers, the firms generate an externa! economy-lower switching costs-that
translates into lower wages for all firms in the cluster.

Benefits and Costs of Labor Pooling

We can use Figure 3-2 to explore the trade-offs associated with labor pooling.
Considera software firm that has two equally likely outcomes in a particular year:
good times or bad times. In good times the price of its product is high, so the firm
Chapter 3 Big and Small Citi:s 45

FIGURE 3-2 Benetits and Costs of Labor Pooling

(A) (B)

Isolated Firm Firm in Cluster

s s

30 ---- -- -------- g 30

20

10 --------- ------ b 10
1
1 Di.id

120
Number of workers Numbcr of workers

An isolated firm pays different wages in good times and bad times. but hires the J>ame number of
workers in both cases. ln contr.t\t. a firm in an industry cluster pays the same wage in both cases, but
hires more workers in good times. The benefit of clustering and labor pooling is that during good
times the firm pays a lower wage and hires more workerl>. The cost is that during bad times the firm
pays a higher wage. The bencfit exceeds the cost because the lirm hires more workcrs during good
times than during bad times.

has a relatively high demand for labor, as indicated by the demand curve labeled
D good· In bad times the price of its product is low, so the firm has a relatively low
demand for labor, as indicatcd by the demand curve labeled D bad·
The left panel of Figure 3-2 shows what happens íf the fi rm chooses an isolated
site, far from other software firms. The supply of labor available to the software firm
at the isolated si te is fixed at l 20 workers. In good times the firm pays a wage of $30,
but in bad times the firm pays a wage of only $ l O. In other words, an isolated firm hires
the same number of workers in good times and bad, but pays a higher wage in good
times.
The right panel of Figure 3-2 shows what happens if the firm locates in a cluster
of other software firms and draws on a common pool of software workers. Suppose
that in any year, a good time cxperienced by one firm is balanced by a bad time
experienced by another firm, so the industrywide demand for software workers is
stable. This means that the wage is constant over time at $20. A firm in the cluster
will paya wage of $20 in good times and bad, but will hire 160 workers in good
times. compared to only 80 workers in bad times.
Consider the trade-offs assocíated with moving the firm from the isolated site
to the industry cluster with the labor pool. The benefit of moving to the cluster
is that during good times, the firm would pay a Iower wage and earn more profit.
In the right panel of Figure 3-2, the benefit of pooling is shown by the líghtly
46 Pan l Market Forces in the Development of Cities

shaded trapezoid, which includes a rectangle and a triangle. The rectangle (a height
of $1 O ($30 - $20) and a width of 120) shows the savings on labor costs for the
first 120 workers hired: The firm saves $10 on each worker, so the total savings is
$ 1,200. The triangle (a height of $1 Oanda width of 40) shows the extra benefits as-
sociated with hiring the 12lst through the 160th workers. The demand curve shows
the marginal benefit (marginal revenue product) of workers, and the gap between
the demand curve and the wage line shows the net benefit from hiring a worker
(the revenue generated by a worker minus the wage). The area of the triangle is
$200 (equal to ($10 · 40)/2)) so the area of the trapezoid (the benefit of pooling)
is $1,400.
The cost of moving from the isolated si te to the cluster is related to the fact that
during bad times, the firm will pay a higher wage in the cluster than it would in
an isolated site. During bad times the firm would pay $20 in the cluster, compared
to only $ 1Oat the isolated si te. In the right panel of Figure 3-2, the darkly shaded
trapezoid shows the cost of labor pooling. The firm in the cluster pays a higher wage
to the 80 workers it hires, generating a cost of $800 ($1 Otimes 80). In addition, the
firm hires fewer workers and thus loses an amount shown by the triangle with height
$1 Oand base 40 workers ($200). Therefore, the cost of clustering and labor pooling
is $1,000.
Are the benefits of clustering greater than the costs? Looking at the right panel
of Figure 3-2, the lightly shaded trapezoid is larger than the darkly shaded one, so
the benefits exceed the costs. In this example, the benefit is $1,400 and the cost is
$1 ,000, so the net benefit is $400. This will always be the case if the wage in the
cluster is the average of the two wages (for good times and bad times) at the isolated
si te. Geometrically, the height of the two trapezoids is the same, but the base of the
benefit trapezoid is longer, so the benefit trapezoid will always be bigger than the
cost trapezoid. The economic explanation for the net benefits of clustering is that
the cluster allows the firm to benefit from good times by hiring more workers at the
same wage, but also allows the firm to cushion the effects of bad times by hiring
fewer workers.

Sharing a Labor Pool: Matching

Consider next the issue of uncertainty about what sort of job skills a firm will re-
quire. Although a firm probably knows what sorts of labor skills it needs this year,
uncertainty about the future demand for its products and future production tech-
nology causes uncertainty about the firm's future needs for labor skills. A large
labor pool near an industry cluster provides a wider variety of skills for firms
to tap.
To illustrate the notion of labor matching, considera workforce with five types
of workers, that is, those having five types oflabor skills. The firm doesn 't know what
type of worker will be most appropriate for next year's production, but must pick a
location now and then hire workers that provide the best match of labor skills. The
firm must choose between an isolated site and the industry cluster. The difference is
that the cluster has more workers and thus a greater variety of labor skills.
Chapter 3 Big and Small Cities 47

TABLE 3-2 Labor Market Matching in an Isolated Location

Type 1 Type 2 Type 3 Type 4 Type5
Actual types X
Output produced S4 $5 S6 SS S4

TABLE 3-3 Labor Market Matching in an lndustry Cluster

Typel Type2 Type3 Type4 Type5
Actual type~ X X X
Output produced $6 $5 $6 $5 $6

Table 3-2 provides a simple example of labor-skill matching. The isolated

location has a single type of labor, type 3. The output of a worker depends on
the gap between the firm 's ideal type and the actual type of worker. If the firm dis-
covers next year that its ideal labor type is 3, there is a perfect match and the worker
produces $6 worth of output. But if the firm's ideal type turns out to be 1, the best
the firrn can do at the isolated site is to hire a type 3, and such a worker will produce
only $4 worth of output. Similarly, ifthe firm's ideal type turns out to be 2, the output
of the type 3 worker at the isolated site will be $5.
Since a firrn doesn't know what its ideal labor type wi ll be, it must compute its
expected output. Let's assume that each type is equally likely to be the ideal type.
The expected output at the isolated site is
1 1 1 1 1 $24
Expected output =
5($4) + S($5) + 5($6) + 5($5) + S($4) = 5 = $4.80

Table 3-3 shows an example for a firm locating in the industry cluster. The
presence of other firms in the industry increases both the nurnber and variety of
workers. There are three types of labor available in the cluster ( 1, 3, and 5), so the
firm is likely to get a better ski lls match. lf the firm's ideal type turns out to be
l , 3, or 5, there will be a perfect match and the worker will produce $6 worth of
output. If the ideal type turns out to be 2 or 4, the worker will produce $5 worth of
output.
Each of the five possible ideal types is equally likely, so the expected output in
the cluster is
I 1 1 1 l $28
Expected output =
5($6) + 5($5) + 5($6) + 5($5) + 5($6) = 5 = $5.60
The lesson from Tables 3-2 and 3-3 is that clustering increases productivity
by providing a better match between firms and workers. Firms in an industry with
rapidly changing skills requirements and large variation in the skill levels of its
workforce have an incentive to form an industry cluster. The evidence of clustering
to exploit labor pooling comes from Dumais, Ellison, and Glaeser (1997), who show
that industries with similar labor requirements form industry clusters, and clustering
48 Pan 1 Market Forces in the Development of Citics

is more prevalent in industries with volatile markets. Rosenthal and Strange (2000a)
provide additional evidence for clustering based on labor pooling.

Sharing Information: Knowledge Spillovers

A third type of localization economy arises from the sharing of information among
firms in the same industry. As we saw in Chapter 1, a disproportionate amount of
innovation occurs in cities, especially the largest cities. A city provides opportunities
for interactions among peoplc with common interests and thus promotes creative
thought.
A cluster of firms in the same industry promotes innovation by bringing together
people producing similar goods with similar production technology. In the words of
Alfred Marshall (1920),

When an industry has chosen a locality for itsclf. it is likely to stay there long; so
great are the advantages which people following the same skilled trade get from near
ncighborhood to one another. The mysteries of the trade become no mysteries; but are
as it were in the air, and children learn many of them unconsciously. Good work is
appreciated, inventions and improvementS in machincry, in processes and the general
organization of the business have their merits promptly discussed; if one man starts a
new idea, it is taken up by others and combined with suggestion of their own; and thus
it becomes the source of new ideas.

The opportunity to exchange ideas occurs in both formal and informal settings. A
cluster of computer makers produces a large concentration of computer scientists and
engineers who can exchange ideas while they work (e.g., the suppliers ofintermediate
inputs interact with the designers of new products) and play (e.g., the workers from
different firms "talk shop" while eating or jogging).
There is evidence that knowledge spillovers encourage industry clustering.
Dumais, Ellison, and Glaeser (2001) show that knowledge spillovers increase the
number of new plant births, with the Jargest effect on industries that employ college
graduares. Their results suggests that knowledge spillovers are more important in
determining the locations of firms in idea-oriented industries. Rosenthal and Strange
(2000a) show that industries that are more innovative (as measured by innovations
per dollar of goods produced) are more likely to form clusters.
There is also evidence that the strength of knowledge spillovers depends on
the structure of the industry. Rosenthal and Strange (2000a, 2000b) show that
knowledge spillovers are strongest in industries with many small, competitive firms.
Saxenian ( 1994) compared knowledge spillovers in the electronics industries in
Silicon Valley to the spi llovers in the electronics industries along Route l 28 in the
Boston area. Silicon Valley is a network of specialized companies, each of which fo-
cuses on a narrow niche within the industry. Each new product uses inputs from many
firms. This network system creares an atmosphere that encourages collaboration, ex-
perimentation, and shared learning among companies. In contras!, the firms along
Route 128 are less interdependent, leading to less extensive know ledge spillovers.
The knowledge spillovers in Silicon Val ley played a large role in its emergence as
the center of the electronics industry.
Chapter 3 Big and Small Cities 49

URBANIZATION ECONOMIES

A second type of externa] scaleeconomy occurs ifthe production cost of an individual

firm decreases as the total output ofthe urban area increases. Urbanization econo_mies
differ from localization economies in two ways. First, urbanization economies result
from the scale of the entire urban economy, not simply the scale of a particular
industry. Second, urbanization economies generate benefits for firms throughout the
city, not just firms in a particular industry.
Although urbanízation economies occur at the metropolitan level racher than the
índustry level, they are similar to localization economies in the sense that they arise
for the same reasons.

1. lntermediate inputs. Firms from different industries share the suppliers of

intermediate inputs, allowing the realization of scale economies in the provision
of business services such as banking, insurance, real estate, hotels, building
maintenance, printing, and transportation. Given the large scale economies in
public services, there are benefits from sharing highways, mass transil systems,
schools, safety services, and fire protection.
2. Labor pooling. Large cities provide citywide labor-market pooling. lf fluctu-
ations in the labor demands of different industries are not correlated, workers
in a declining industry can easily switch to a growing industry. Given the lower
search and moving costs in large cities, firms can more easily increase or de-
crease their workforces.
3. Sharing information. J ustas knowledge spills over from one firm in an industry
to another firm, it spills across firms in different industries. The knowledge
spillovers lead to innovation in product design and production methods.

EVIDENCE OF EXTERNAL ECONOMIES

There is a large volume of economics literature examining the extent of externa]

economies, both localization and urbanization economies. In searching for evidence
of localization economies, researchers focus on the effects of industry concentration
on (1) worker productivity, (2) the number of new production plants (plant births),
and (3) growth in industry employment. If there are locahzation economies, we
expect industry clusters to generate higher productivity, more births, and more rapid
employment growth. In contrast, ifthere are urbanization economies, what matters is
the size of the urban economy, not the size of a particular industry, so expect larger
cities to generate higher productivity, more plant births, and faster employment
growth.

Evidence of Localization Economies

Consider first the effect of industry concentration on worker productivity. Henderson
( 1986) measured localization economies as the elasticity of output per worker with
respect to industry output, defined as the percentage change in output per worker
so Pait 1 Market Forces in the Development of Cities

divided by the percentage change in industry output. Far the electrical machinery
industry, the localization elasticity is 0.05, meaning that a 10 percent increase in
the output of the electrical machinery industry increases output per worker by about
0.50 percent. The elasticity estimates for other U.S. industries range from 0.02 far
the pulp and paper industry to 0.11 for the petroleum industry. Nakamura (1985)
provides evidence of localization effects for Japanese cities.
The Mun and Hutchison (1995) study examines agglomeration economies in the
office sector, using data from the city ofToronto. Their conclusion is that agglomera-
tion effects in the office sector are much more powerful than those in manufacturing.
For example, a 1Opercent increase in the number of office firms increases the produc-
tivity of office firms by about 2.7 percent (i.e., the elasticity of average productivity
with respect to the size of the office industry is 0.27). The increase in productivity is
larger in cases when the increase in office employment occurs in central locations.
In addition, the agglomeration effects are largest in the area of the city where the
increase in office employment occurs.
Consider next the implications of industry clusters for the location of new pro-
duction facilities. Carlton ( 1983) examines the location choices of firms in three
industries: plastics products, electronic transmitting equipment, and electronic com-
ponents. He found that the number of plant births increases with the total output of
the industry, indicating that industry concentration promotes births. His computed
elasticity of births with respect to industry output is 0.43: A 10 percent increase
in total output increases the number of births by 4.3 percent. More recently, Head,
Reís, and Swenson (1995) show that Japanese corporations locate their new plants
close to other Japanese plants in the same industry. Rosenthal and Strange (2000b)
examine establishment births in severa! industries, and show that births in a particu-
lar industry are more numerous in locations close to concentrations of employment
in the same industry.
Consider next the effects of industry concentration on employment growth.
Henderson, Kuncoro, and Turner (1995) examine employment growth in selected
mature industries, and conclude that growth is more rapid in areas that start with
large concentrations of the particular industry. Rosenthal and Strange (2000b) show
that the number of jobs in new establishments is sensitive to nearby employment in
the industry. Table 3-4 shows their results for the software industry. As shown in the
first row of numbers, if we add 1,000 jobs in the software industry within a one-mi le
radius of a particular zip-code area, the number of new jobs within the area increases

TABLE 3-4 Localization Economies for the Software Tndustry

Jobs in New Establishments in
Add 1,000 Software Jobs in Zip-Code Area lncrease by
1-mile radius around zip-code area 1J.7 jobs
Ring 1- 5 miles from the zip-code area 0.8
Ring 6-10 miles from the zip-code area 0.8
So11rce: Roscnthal, Stuan, and William Strangc, ··Geography, Industrial Organization, and
Agglomeration," Working Paper (2000).
Chapter 3 Big and Small Cities 51

by 11.7 jobs. The job gains are much smaller for software employment more distant
from the zip-code area. If we add 1,000 software jobs in a ring from 1 to 5 miles
from the zip area, software employment in the area increases by only 0.8 jobs.
How local are localization economies?The Rosenlhal and Strange study (2000b)
shows that the benefits associated with localizaLion economies (sharing input sup-
pliers, a labor pool, and information), fall rapidly with distance. They compute the
percentage drop-off of the localization effect for a one-mile move away from a clus-
ter. On average, Lhe localization effect peters out at a rate of about 50 percent per
mi le. The rapid attenuation of the localization economies results explains the local
in " localization economies."

Evidence of Urbanization Economies

There is sorne mixed evidence conceming urbanization economies. The consensus
among urban economists is that the doubling of a ciLy's size will increase labor pro-
ductivity by 5 to 1Opercenl (Milis, 2000). This could be a result of pure urbanization
economies (external economies of scale that cross industry boundaries) or cou ld re-
ftect the fact that large cities also have large concentrations of individual industries
subject to localization economies. Carlino ( 1987) examined 19 manufacturing in-
dustries in 68 metropolitan areas, and found evidence of urbanization economies
in 13 of the 19 industries. Nakamura (J 985) found some evidence of higher labor
productivity in larger cities, while Henderson ( 1988) did not.
In recent years, empirical studies of urbanization economies ha ve focused on the
effects of a city's economic diversity on plant births, relocations, and employment
growth. At the heart of urbanization economies are the benefits associated with
producing in a large diversified city with many opporlunities to draw on resources
used by other industries. Using data from France, Duranton and Puga (2000) show
that a small majority of new plants (59 percent) are located in areas that are relatively
diverse. This provides sorne evidence that diversity promotes plant births. Note that
it is in conftict with the results of Rosenthal and Strange (2000b), which suggests
that specialization promotes plant births, but diversity does not.
Data on plant relocations tells an interesting story about the life cycle of firms.
Using French data, Duran ton and Puga ( 1999) show that in 72 percent of ali plant
relocations, the plant departed from a relatively diverse area and arrived in a rel-
atively specialized area. For the most innovative sectors, the shares of relocations
from diverse to specialized areas are as follows: 93 percent for research and develop-
ment; 82 percent for pharmacology and cosmetics; 76 percent for business services;
73 percent for printing and publishing, 72 percent for aerospace; and 69 percent for
electrical equipment. These results suggest that diversity provides a favorable envi-
ronment for starting up production, but as the product and the production process
mature and stabilize, localization economies become more important.
How does urban diversity affect employment growth? Two studies (Glaeser,
Kallal, Scheinkman, and Schleifer (1992), and Henderson, Kuncoro, and Turner
( 1995)) suggest that diversity promotes employment growth, especially in new
and innovative industries. Hanson (2001) concludes that long-run industry growth
52 Pan 1 Market Forces in the Dcvelopment of Cities

is higher in cities with a wider variety of industries, suggesting that urban eco-
nomic diversity promotes growth. Note that these results conflict with studies that
find specialization-large concentrations of employment in a particular industry-
promotes employment growth. It seems that the issue of which is better for employ-
ment growth--<liversity or specialization-has not been resolved.
There is no doubt that economic diversity in cities promotes innovation in prod-
uct design and production. Glaeser, Kallal , Sheinkman, and Shleifer (1992) pro-
vide evide nce that knowledge spillovers contribute to the growth of cities, with the
spillovers occurring mainly between industries rather than within a particular in-
dustry. Harrison, Kelley, and Gant ( 1996) and Kelley and Helper ( 1999) show that
diversity in local employment increases the rate of adopting new production pro-
cesses. Feldman and Audretsch ( 1999) show that local diversity has a strong positive
effect on the development of new products, while narrow specialization has a neg-
ative effect. As documented by Fuj ita and Ishii ( 1998), electronic firms locate their
research and development operations and trial plants in major metropolitan areas
(Tokyo, Kyoto, and Boston), while mass-production facilities are located in smaller
cities or rural areas.

The Incubation Process

Localization and urbanization economies are responsible for the inc ubation process.
A large city provides a nurturing environment that helps firrn s and industries in the
early stages of product devclopment. In the early life of many industries, demand
is unpredictable and production techniques are unsettled, and small firms grope
around to refine their products a nd design production processes. In a large city with
many suppliers of intermediate inputs, many labor pools, and extensive knowledge
spillovers, firms in the groping stage can draw on the varied resources of the city.
Once the firm commits to a product design and a production process, it can emerge
from the city incubator a nd switch to large-scale production at a site with lower
land and labor costs. This is also known as the product cycle theory of industrial
location: Products are developed by small firms in areas that allow sharing of inputs
and knowledge spillovers. When the product becomes standardized, the firm can
produce it on a larger scale in its own production facilities.
Vernon (1972) describes the classic example of incubation, the radio industry.
In the l 920s, the market for radios was unpredictable, and firms were experimenting
with alternative production processes. Mass production was impractical: the firms
didn' t know what kind of radio consurners wanted, and hadn ' t developed efficient
production techniques. Radio firms were srnall, agite, and nervous, and clustered in
New York City to exploit localization and urbanization econornies. In the 1930s and
l 940s, products were standardized, and firms developed efficient methods of mass
production. Firms then moved their assembly facilit ies to the Midwest, which had
lower wages and better access to the national rnarkets. Tube producers migrated to
the Northeast and the South, where labor costs were lower.
The computer industry is also subject to the incubation process. In the 1970s
and early 1980s, the demand for computers was unpredictable and production
Chapter 3 Big and Small Cities 53

techniques were being developed. Small computer firms clustered in the Silicon
Valley in California and along Route 128 in Boston to exploit localization economies.
The standardization of products and production processes allowcd firms to set up
large production facilities outside the clusters. Why did the clusters survive the exo-
dus ofproduction facilities? New products continue to be developed, and firms con-
tinue to cluster to exploit localization economies. Most firms maintain their research
and development facilities in the clusters and set up branch plants for manufacturing
and assembly.

DIFFERENCES IN CITY SIZE

As shown in Figure 3-1, there is substantial variation in city sizes. In this chapter,
we've seen why sorne types of firms cluster in cities. How does the clustering of firm s
lead to the sort of distribution of city sizes shown in Figure 3-1? We'll answer this
question in three steps, looking at the roles of localization economies, urbanization
economies, and locally consumed products.

Localization Economies
Industries vary in the strength of their localization economies. Sorne industries ex-
perience substantial cost savings from input sharing, labor pooling, and information
spi llovers, while others cxperience relatively small cost savings. The stronger the
localization economies, the greater the incentive for firms to forman industry cluster
with large concentrations of employment. The variation in thestrength oflocalization
economies across industries is one factor in generating cities of different sizes.
Suppose there are three industries subject to localization economies, each of
which employs 120 workers. Industry L has strong localization economies, causing
all firms in the industry to form a single cluster; ali 120 workers locate in one city.
In contrast, industry M has moderate localization economies, and its workforce is
spread over two cities, with 60 workers in each city. Finally, industry S has weak
localization economies, and its firms are spread out over four cities, with 30 workers
in each city. In this example, variation in localization economies translates into
variation in employment levels across cities.
• There is a single large city with 120 workers.
• There are two medium-size cities, each with 60 workers.
• There are four smal l cities, each with 30 workers.

Urbanization Economies
Consider next the role of urbanization economies, the spillover benefits that result
from the scale of theentire metropolitan area notjust a particular industry. Firrns in an
industry subject to urbanization economies are attracted to large metropolitan areas,
and this tcnds to make large cities larger. In other words, urbanization economies
amplify any underlying differences in total employment.
54 Pan 1 Marke1 Forces in Lhe Developmenl of Ci1ies

FIGURE 3-3 Varialion in City Size

600
c:::::J Employmeni in local indu,try
c:::::J Employmcnt in industrie, subjcct to urbanization economies
e:;;¡ Employment in industries subjcct to localization economics

500

400 400

E
"'
E
-2o.. 300
E
Ul

200

80
120 120

100

60

o Ci1y L Ci1y M- I City M-2 Ci1y S-1 Ci ty S-2 CiiyS-3 Ci1y S-4

Industries subjec1 10 s1rong localization economies form large clus1ers, causing large concentra1ions
of jobs. Tndus1ries subjec1 10 weaker localiunion economics form small clus1ers and employmen1
concenLrations. Urbaniza1ion economies amplify difference~ in employmem because large ci1ies
provide grea1er urbanita1ion economies. Local employmem amplifies differences in cily sizc bccause
larger ci1ies have a wider varie1y of consumer goods.

We can use Figure 3-3 to show the combined effect of localization and urban-
ization economies in our seven hypothetical cities. The lower dark areas show core
employment in industries subject to localization economies, and the lighter areas
show employment resulting from urbanization economies. The large city (L), which
has 120 jobs in an industry subject to localization economies, gains 80 jobs from
urbanization economies, while each of the medium cities (M) gains only 20 jobs.
Each medium city has less to offer than the large city in terms of productivity
gains from urbanization economies, so the increase in employment from urbaniza-
tion economies is smaller. If we assume each small city (S) isn't Jarge enough to
generate urbanization economies, it won't attract any firms subject to urbanization
economies. In general, urbanization economies amplify differences in employment
Chapter 3 Big and Small Citics 55

because the larger the city, thc greater the attractiveness of the city to firms subject
to urbanization economies.

The Role of Consumer Goods

So far our discussion of the size distribution of cities has neglected the consumer
side of the urban world. We have implicitly assumed that ali workers in the city
produce goods for export from the city, but don't consume any products locally.
In fact, workers in a city spend sorne of their income on consurner products pro-
vided within the city: they buy food, get haircuts, rent housing, use medica] ser-
vices, and go to sporting events. Thís spending on local consumer products sup-
ports jobs in grocery stores, barbershops, hospitals, housing construction, and sports
teams. The jobs supported by local consumer spending is often labeled local
employment.
Sorne products are available in ali cities, large, medium, and small. If a product
has large per capita demand relatíve to scale economies in production, even a small
city wi ll generate enough demand to support a firm producing the product. For
example, it takes just a few thousand people to support a barber, so even a small
city will have at least one barber. Similarly, the per capita demand for groceries is
large relative to scale economies in grocery selling, so even a small city will have a
grocery store and sorne grocery clerks. Of course, a larger city has more hair to cut
and more people to feed, so will have more barbers and grocery clerks.
Large cities have a wider variety of local goods and services. For sorne products,
firms need a large population base (a large city) to generate sufficient demand to
realize the scale economies associated with producing the good. For example, the
per capita demand for opera is small relative to the economies of scale in producing
an opera, so it may take a mi Ilion or more people to support an opera company. As a
result. you can find opera companies in large cities, but not in small ones. Similarly,
the per capita demand for brain surgery is small relative to the scale economies in
performing brain surgery, so brain surgeons operate only in large cities. In a large
city, you can get anything you could get in a small city (e.g., haircuts and groceries),
and you can also get ali sorts of products that are not available in small cities (e.g.,
opera and brain surgery). In fact, people who live in small cities travel to larger cities
to buy products that are unavailable in their own cities. In contrast, consumers in
large cities can buy anything they want in their own city, so consumer travel from
large cities to small ones is rare.
The wider variety of consumer products in larger cities amplifics any differ-
ences in population. Consider two cities that start out with employment in expo11
production. If the Jarger city has twice as many export workers as the small one, what
happens to total employment when we introduce local products and local workers?
The small city would gain barbers and grocery clerks, while the Jarge city would gain
twice as many barbers and clerks, and also gain opera singers and brain surgeons.
In general, when we add local products to the economy, ali cities gain jobs, but the
larger city gains more jobs because it has a large enough population to support firms
producing products with relatively small per capita demand.
56 Part 1 Market Forces in the Development ofCities

We can use the simple model depicted in Figure 3- 3 to show the effects of local
employment on differences in city size. Suppose we start with export employment:
City L initially has 200 export jobs (120 from localization economies and 80 from
urbanization economies), while each of the M cities has 80 jobs, and each of the
S cities has 30 export jobs.
• Small city: One local job per export job. To provide the products that can be
supported even in a small city, we add one local worker (barbers and clerks) for
each export worker.
• Large city: Two local jobs per export job. We add enough barbers and clerks
to serve the larger population, and also add we add opera singers and brain
surgeons.
• Medium city: 1.5 local jobs per export job.
Figure 3- 3 shows how urbanization economies and local employment am-
plify any initial differences in employment. The large city started out with four
times as many jobs in industries subje¡::t to Iocalization economies as the small city
(120 versus 30), but the large city gains 80 jobs from urbanization economies and
gains more jobs related to local products, so it reaches an employment level that
is 10 times that of the small city (600 versus 60). Similarly. the medium city ini-
tially has twice as many jobs as the small city, but the combination of urbanization
economies and gains in local employment increases the gap.

AGGLOMERATIVE ECONOMIES IN MARKETING:

SHOPPING EXTERNALITIES

This section explains how agglomerative economies in marketing affect the devel-
opment of cities. The previous section of the chapter explained how agglomerative
economies in production cause industrial firms to cluster, resulting in the develop-
ment of Jarge industrial cities. This section explains how agglomerative economies
in marketing cause trading firms to cluster, resulting in the development of smaJI
market-based cities and retail concentrations within cities.
A shopping externality occurs if the sales of one store are affected by the Jocation
of other stores. Suppose that an isolated drugstore relocates next to a grocery store.
If both stores experience an increase in sales, they generate shopping externalities:
each store attracts consumers to the retail cluster, generating benefits for the other
store. These shopping externalities cause firms selling related products to form retail
clusters. Sorne retail clusters cause the development of market cities. Other clusters
occur within large cities, generating downtown shopping areas, malls, and shop-
ping centers. There are two types of products that generate shopping externalities:
imperfect substitutes and complements.

Imperfect Substitutes
Two goods are imperfect substitutes if they are similar but not identical. Suppose
that you've decided to buy a new sports car, but haven't decided whether to buy a
Chap1er 3 Big and Small Ci1ies 57

Ford, a Chevrolet, or a Honda. The sports cars offered by the threc companies are
similar, but differ in performance, shape, color, and gadgetry. Because the differ-
ences between the cars are subtle, you must do your comparison shopping in person,
spending time and money traveling to the three car dealers. If the three dealers fonn
a cluster, they decrease the cost of comparison shopping, attracting consumers to
the cluster. There are many goods that are imperfect substitutes. Sorne examples are
clothes, shoes, jewelry, and electronic equipment. For these goods, the clustering of
firms selling similar products decreases shopping costs and attracts potential buyers.
Sorne retailers cluster in the city center, while others cluster in shopping centers and
malls.
Figure 3-4 illustrates the notion of agglomerative economies from comparison
shopping. Suppose that there are two auto dealers in a particular city, and they are
initially located far from one another. Each dealer has a fixed supply of 50 cars per
week, so the supply curves at each location are vertical at 50 cars. In the initial
equilibrium, each dealer faces the demand curve D' and sells its 50 cars ata price of
$8,000. Suppose that dealer M moves to a si te next to the other dealer. The total supply
of cars al the location of the original dealer doubles, from 50 cars per week to 100
cars. [f the customers of dealer M follow the dealer to its new location,'the demand
for cars also doubles: the demand curve shifts to the right by 50, from D' to D" .

F IGU RE 3-4 Agglomerative Economies from Comparison Shopping

$ lnilial Supply in
supply el usier

~ 9.000
<.>
'o 8,000
~
·e
o..

D*
D"

D'

50 100 110
Number of cars per week

In 1hc initial equi librium wi1h 1wo alllo dealers locmcd far from one another, cach dcaler
supplies 50 cars al a price of $8,000. If dealer M rcloca1es next to dealer S. the supply of cars
at location S doubles, from 50 10 100. The demand more than doubles bccausc !he el usier
allracts comparison shoppers who would otherwisc not patronize either dealcr.
58 Part 1 Market Forces in the Development of Citics

The shi ft of the demand c urve from D" to o• incorporates shopping extemalities.
The clustering of auto sellers decreases the shopping costs of consumers who en-
gage in comparison shopping. Therefore, the cluster attracts consumers who would
otherwise not patronize either of the dealers. These additional consumers cause
another rightward shift of the demand curve, from D" to D*. In this example, an
additional 10 consumers patronize the two dealers. lf the dealers continue to supply
only 50 cars each, the eq uilibrium price increases from $8,000 to $9,000. Of course,
another option is to increase the number of cars sold. In either case, the extra shift
of the demand c urve frorn cornparison shoppers generares benefits for the dealers in
the cluster, resulting in a higher price or a larger volume.

Complementary Goods
Clustering also occurs when firms sell complementary goods. Complementary goods
are often purchased on the same shopping tri p. For example, if a consumer purchases
pants and shoes on the same trip, his travel cost will be lower ifthe pants store is near
the shoe store. T he shoe store wi 11 benefit from the presence of the pants store beca use
together they provide one-stop shopping for consumers. Because of the bencfils of
one-stop shopping, firms selling complementary goods cluster in shopping centers,
malls, and city centers.

Retail Clusters
The agglomerativc economies associated with shopping generate severa! types of
retail clusters. For sorne goods, retailers form retail strips along arterial streets, for
example, auto rows. Other retailers cluster in shopping centers, malls, and city cen-
ters. These clusters provide both a mix of imperfect substitutes (different jewelry
stores) and complements (food and drugs, clothing and shoes), allowing both com-
parison shopping and one-stop shopping. Retaile rs that choose isolated locations
instead of clusters sell goods that are not subject to shopping externalities.

INNOVATIONS IN TELECOMMUNICATIONS
AND THE FUTURE OF CITIES

Will improvements in tclecommunications make cities obsolete? Cities provide phy-

sical proximity and thus facilitate face-to-face communication among the producers
of goods and services. Recent developments in telecommunications have gene rated
new options for communication, including electronic mail, faxes, the Internet, and
teleconferencing. According to sorne noted futuri sts, because these new technologies
are substitutes for face-to-face communication, cities will shrink and may even
disappear. Alvin Toffter ( 1980) suggested that telecommunications technology wi 11
allow people to escape the city to live in "electronic conages" in rural areas. Nicolas
Negroponte ( 1995) predicts that "the transmission of place itself will start to become
possible through new virtual reality and telecommunications technology" resulting
in a decline in the importance of special places like cities.
Chaptcr 3 Big and Small Citics 59

Marshall McLuhan (J 964) suggested that the city "as a form of major dimension
must inevitably dissolve like a fading shot in a movie." The prediction that cities
will decline or disappear is based on the notion that telecommunication is a true
substitute for face-to-face communication. According to Gaspar and Glaeser ( 1998),
however, the two forms of communication may be complementary. Man y production
relationships involve both face-to-face interactions and telecommunications via the
telephone, e-mail, faxes, or the Internet. An improvement in telecommunications
has two effects on the number of face-to-face contacts.
• In a given relationship, some face-to-face encounters will be replaced by telecom-
munication: An investment banker could e-mail a client instead of meeting her
for a brief conversation; the banker could fax a contract instead of delivering it
by hand.
• Easier communication may increase the number of relationships. An investment
banker could consult with more people in the process of evaluating the merits
of a particular project; the banker could handle more projects.
As long as each relationship requires sorne face-to-face communication, the net effect
on the number of these contacts is ambiguous. For example, i f the investment banker
reduces the number of face-to-face contacts per relationship from five to four but
increases the number of relationships from six to eight, the number of face-to-face
contacts will increase from 30 to 32.
There is sorne evidence that face-to-face communication and telecommunication
are not pure substitutes. lf they were substitutes, we'd expect that the widespread
use of faxes and electronic mail would decrease the demand for business travel
as business workers substitute electronic communication for face-to-face contact.
Between 1985 and 1995, a period during which faxes and e-mail were first developed
and then became ubiquitous, the volume ofbusiness travel actually increased by more
than 50 percent (Gaspar and Glaeser, 1998). Of course, other changes during this
period may have contributed to the growth of business travel, and it is difficult to
discern the contributions of different factors. But the facts on business travel provide
suggestive evidence that the two forms of communication are complements rather
than true substitutes.
More evidence concerning the complementarity of face-to-face contact and
telecommunication comes from data on the use of telephones, the first telecom-
munication device. If the two types of communication are complements, face-to-
face contact will increase the demand for telecommunication, and people who are
physically closer-and presumably see each other more often-will also call each
other more often. In the l 970s, more than 40 percent of phone calls were made
to places within a two-mile radius, and more than 75 percent were made to places
within a six-mile radius. In Japan, the shorter the distance between two points, the
more telephone calls that are made between those points. Also in Japan, there is
a positive relationship between phone usage and urbanization: people who live in
Jarge cities (population exceeding 100,000) spend more time on the phone than
people who live in small cities or rural areas. One interpretation of these facts
is that physical proximity-and thus the ease of face-to-face contact-generates
60 Pan 1 Markct Forces in the Development of Cities

greater demand for telecommunications beca use the two forms of communication are
complements.
A more subtle issue concems the effect of telecommunications on the complex-
ity of interactions in cities. Easier communication may allow greater specialization
in design and production, increasing the need for face-to-face communication to
coordinate the specialized efforts of the various people involved. In addition, easier
communication may increase the complexity of the information and ideas trans-
mitted between people, requiring more face-to-face contacts to communicate these
ideas.
Gottman ( 1977) discusses the effects of the telephone (the first telecommunica-
tion device) on cities:

The telephone provides, when needed, quasi-immediatc verbal communication between

all the interdependcnt units at minimum costs... 11 would have been very difficult for ali
thesecomplex and integratcd networks (in cities) to work in unison without the telephone,
which made possible thc constant and efficient coordination of ali the systems of the
large modern city. Thc tclephone helped to make thc cities bigger and more exciting.

The same logic applies to more recent innovations such as faxes, e-mail, and the
Internet. What are the facts on innovations in telecommunications and urbanization?
During the recent period of rapid development of telecommunications, urbanization
actually increased. Based on recent trends, the proportion of the global population
living in cities is expected to increase from its current leve! of 45 percent to 61 percent
by the year 2025. During this period, the proportion of population living in cities is
expected to increase from about 74 percent to about 84 percent in North America,
South America, and Europe; from about 34 percent to 54 percent in both Africa and
Asia; and from 70 percent to 75 percent in the Pacific region.
It appears that innovations in telecommunicati on technology will not cause
cities to disappear. Sorne interactions among producers require face-to- face contact,
so there will always be a need for cities and the physical proximity they provide.
The number of face-to-face contacts will be affected by innovations in telecommu-
nications, but both the direction and the magnitude of the changes are unknown.
Although telecommunication is a good substitute for sorne types of face-to-face
contact, easier communication may increase the need for face-to-face contact as
(l) more relationships are developed, and (2) relationships become more complex.
If the net effect is more rather than less face-to-face communication, innovations in
telecommunications cou ld cause cities to grow rather than shrink.

SUMMARY

l. Localization economies (lower production costs as industrywide output in-

creases) occur because firms in an industry cluster benefit from sharing ( 1) the
suppliers of intermediate inputs, (2) a labor pool, and (3) information.
2. Urbanization economies occur if the production cost of a particular firm de-
creases as the total output of the urban area increases.
Chap1er 3 Big and Small Cili!S 61

3. The incubation process results from localization and urbanization economies.

A large city provides a nurturing environment that helps firms and industries in
the early stages of product development.
4. Differences in city s ize are caused in part by diffcrences in localization econo-
mies across industries. Any underlying differences in city size are amplified by
urbanization econom ies and the provision of local goods and services.
S. lnnovations in telecommunication technology will not cause cities to disappear
because sorne activities require face time, so there will always be a need for
cities and the physical proximity they provide.
6. A shopping externality occurs if the sales of a particular store increase as other
retailers move closer to the store. These agglomerative economies in marketing
cause the clustering of retailers. Comparison shopping causes the clustering of
firms selling imperfect substitutes. One-stop shopping causes the clustering of
firms selling complementary goods.

EXERCISES AND DISCUSSION QUESTIONS

l. Most of the dresses made in the United States are produced by large firms in
suburban areas, not by small firms in city centers. Is this consistent with the
notion of industry externalities and clusters? Hint: Is there a difference between
a Kmart dress and a dress produced by a small dressmaker?
2. The conventional wisdom for urban economic development is: "Don't pul ali
your eggs in one basket. Diversify the economy." To explain the idea of diver-
sification, consider old McDonald, who must carry a dozen eggs from the barn
to the house. The ground between the barn and the house is slippery, so there
is a 50 percent chance that McDonald will slip on a given trip and break ali the
eggs in his basket. Consider two strategies: a one-basket strategy (a single trip
with ali 12 eggs) anda two-basket strategy (two trips, with 6 eggs per trip).
a. List ali of the possible outcomes under each of the strategies.
b. What is the expected number of delivered (unbroken) eggs under each
strategy?
c. What are the trade-offs between the two strategies? lf you were McDonald,
which strategy would you adopt?
d. What are the lessons for economic development strategies?
3. According to the conventional wisdom concerning urban economic develop-
ment, a city should develop a di verse economy with a Jarge number of industries.
Evaluate the merits of this conventional wisdom in light of the empirical evi-
dence concerning the magnitudes of localization and urbanization economies.
4. Considera 10-firm industry that produces computer equipment, a set of goods
with rapidly changing demand and production technology. The industry has
the following characteristics: [i] The 1O fi rms produce computer equipment
using labor and raw materials; [ii] Raw materials are ubiquitous (available at
all locations at the same price); [iii) Each firm produces one new product per
year, and each product becomes obsolete after ayear; [iv] Only 3 of the 1Onew
62 Par! 1 Marke1 Forces in the Development of Cities

products will be successful (sel! more than a trivial amount); [v] The monetary
and time costs of switching a worker from one firm to another are zero, regardless
of the spatial distribution of firms.
a. Will the firms in the industry form a cluster? Why or why not?
b. How would your answer to (a) change if workers incur moving costs when
they switch from one firm to another?
S. Mr. Wizard, a regional planner, recently made the following statement: "If my
assumptions are correct, ali cities in this region will eventually be identical.
They will be the same size and will sell the same set of goods."
a. Assuming that Mr. Wizard's reasoning is correct, what are his assumptions?
b. Are Mr. Wizard's assumptions realistic?
6. Considera firm that must choose between an isolated production si te anda site
in an industry cluster. The following table shows the marginal product of labor
for different number of workers. Good times and bad times are equally likely.
In bad times, the price of output is $5, and in good times the price is $10. In the
isolated site the supply of labor is fixed at five workers. In the cluster, the wage
equals the expected wage at the isolated si te.
a. Which location should the firm choose?
b. Defend your answer with a completely labeled graph.

Number of workers l 2 3 4 5 6 7 8
Marginal product 10 9 8 7 6 5 4 3

7. Suppose that the outputs of beauty shops and pet-grooming salons are comple-
mentary, providing one-stop shopping for personal and pet maintenance. Betty
Beehive is thinking about moving her beauty shop from an isolated location to a
vacant building next to Peter's pet-grooming shop. In making her decision, she
rnakes the following assumptions: [i] If Betty moves, she will keep ali of her
current customers (20 people per week), and attract 25 percent of Peter's cur-
rent customers. [ii] Peter currently has 60 customers per week. [iii] Excluding
rent, Betty's profit per beauty treatment is $1 O. [iv] The weekly rent at the new
location is $200 higher than Betty's current rent.
a. lf Betty moves her beauty shop, will her profits increase or decrease?
b. Suppose that if Betty makes the move, 50 percent of her original customers
will switch from George's grooming salon to Peter's. lf Peter's profit per
treatment is $8, how much would he be willing to pay Betty to make the
move? Will the payment be enough to induce Betty to make the move?
8. Considera city with two auto sellers, a Toyota dealer anda Honda dealer. lni-
tially, the distance between the two sellers is three miles. The Toyota dealer wants
to relocate to a site adjacent to the Honda dealer and submits a rezoning request
to the city council. The Honda dealer responds to the rezoning request with the
following statement: "One of the lessons from Econ 100 is that an increase in
supply will decrease price. If the Toyota dealer moves to the site adjacent to my
dealership, the local supply of cars will increase and I'll have to cut my prices to
Chap1er 3 Big and Small Cilics 63

sell the same quantity of cars." Critically appraise the Honda dealer's statement.
lf the statement is incorrect, what's wrong with the reasoning? Illustrate your
answer with a graph.

REFERENCES
1. Carlino, Gerald A. "Productivity in Cities: Does City Size Matter?'' Business Re-
view ofthe Federal Reserve Bank of Philadelphia (November/December 1987),
pp. 3-12.
2. Carlton, D.W. "The Location and Employment Choices ofNew Firms." Review
of Economics and Statistics 65 (1983), pp. 440-49.
3. Dumais, Guy, Glen Ellison. and Edward Glaeser. "Geographic Concentration
as a Dynamic Process." National Bureau of Economic Research Working Paper
6270, November 1997.
4. Duranton, G., and D. Puga. "Diversity and Specialization in Cities: Why, Where
and When Does It Matter?" Urban Studies (2000).
5. Duranton, G., and D. Puga. "Nursery Cities: Urban Diversity, Process Inno-
vation, and the Life Cycle of Products." CEPR Discussion Paper No. 2376
(1999).
6. Feldman, Maryann P., and David B. Audretsch. "lnnovation in Cities: Science-
Based Diversity, Specialization and Localized Competition." European Eco-
nomic Review 43 ( 1999), pp. 409-29.
7. Fujita, Masahisa, and Ryoichi Ishii. Global Location Behavior and Organi-
zational Dynamics of Japanese Electronics Firms and Their Impact on Re-
gional Economies. ln Alfred D. Chandler, Jr., Peter Hagstrom, and Órjan Solvell
(eds.), The Dynamic Firm: The Role of Technology, Strategy, Organization and
Regions. Oxford: Oxford University Press, 1998, pp. 343- 83.
8. Gaspar, Jess, and Edward L. Glaeser. "lnformation Technology and the Future
of Cities." Journal of Urban Economics 43 ( 1998), pp. 136- 56.
9. Glaeser, Edward L.; Hedí D. Kallal; Jose A. Scheinkman; and Andrei Shleifer.
"Growth in Cities." Joumal of Political Economy 100 ( 1992), pp. 1126-52.
l O. Gottman, J. "The Telephone and the Structure of the City." In The Social lmpact
ofthe Telephone, ed. I. De Sola Pool. Cambridge, Mass.: MIT Press, 1977.
11. Graham, Stephen. ''Telecommunications and the Future of Cities: Debunking
the Myths." Citíes 14 (1997), pp. 21-29.
12. Hanson, Gordon. "Scale Economies and the Geographic Concentration of In-
dustry." Journal of Economic Geography 1: 255-276 (2001 ).
13. Harrison, Bennett; Maryellen R. Kelley; and Jon Gant. Specialization versus
Diversíty in Local Econornies: The Implications for Innovative Prívate-Sector
Behavior. Cityscape 2 (1996), pp. 61 - 93.
14. Head, K.; J. Ries; and D. Swenson. "Agglomeration Benefits and Location
Choice." Journal of International Economics 38 (1995), pp. 223-48.
15. Henderson, J.V. "Efficiency ofResource Usage and Cíty Size." Journal of Urban
Economics 19 (1986), pp. 47-90.
64 Pan 1 Markct Forces in 1hc Dcvclopment of Cities

16. Henderson, J.V. Urban Development: Theory, Fact, and lllusion. New York:
Oxford University Press, 1988.
17. Henderson, J.V. ; Kuncoro, A.; and M. Turner. "Industrial Development and
Cities." Journal of Política/ Economy 103 (1995), pp. 1067-8 1.
18. Jacobs, Jane, The Economy of Cities New York: Random House, 1969.
19. Kelley, Maryellen R., and Susan Helper. Firm Size and Capabilities, Regional
Agglomeration, and the Adoption of New Technology. Economics oflnnovation
and New Technology 8 ( 1999), pp. 79- 103.
20. Marshall, Alfred. Principies of Economics. London: Macmillan, 1920, p. 352.
21. McLuhan, H. Understanding Media-The Extension of Man. London: Sphere,
1964.
22. Mills, Edwin. "A Thematic History of Urban Economics." Brookings-Wharton
Papers on Urban Affairs (2000), pp. 1-52.
23. Mun, Seil, and Bruce G. Hutchinson. "Empirical Analysis of Office Rent and
Agglomeration Economies: A Case Study of Toronto." Journal of Regional
Science 35 (I 995), pp. 437-55.
24. Nakamura, R. "Agglomeration economies in urban manufacturing industries: a
case of Japanese cities," Journal of Urban Economics 17 ( 1985), pp. 108-24.
25. Negroponte, N. Being Digital: The Roadmap to the lnformation Superhighway.
London: Hodder and Stoughton, 1995.
26. Rosenthal, Stuart, and William Strange. "The Determinants of Agglomeration,"
Working Paper (2000a).
27. Rosenthal, Stuart, and William Strange. "Geography, Industrial Organization,
and Agglomeration," Working Paper (2000b).
28. Saxenian, Annalee. Regional Advantage: Culture and Competition in Silicon
Va/ley and Route 128. Cambridge, Mass.: Harvard University Press, 1994.
29. Toffler, A. The Third Wave. New York: Morrow, 1980.
30. Vernon, Raymond. "Externa! Economies." In Readings in Urban Economics,
ed. M. Edel and J. Rothenberg. New York: Macmillan, 1972.