. ...

Urban Economics
FIFTH EDITION

Arthur O'Sullivan
Departme11t of Economics
Lewis & Clark College

LIBROS T
COMPAÑIA
Calle 19 No. 3-16 • Local 104
librosycomp@ ho tmai l .com
Telefox: 341 71 00 • Bogotá, O.C.

B McGraw-Hill
t:a lrwin
Boston Burr Ridge, IL Dubuque, IA Madison. WI New York San Francisco St. Louis
Bangkok Bogotá Caracas Kuala Lumpur Lisbon London Madrid Mexico City
Milan Montreal New Delhi Santiago Seoul Singapore Sydney Taipei Toronto
 CHA P TER 4

Where Do Firms Locate?

/ n earlier chapters we explained why cities exisl and why sorne cities are so big. In
this chapter, we'll explore the where of cities. To address the where question, we' ll
examine the location decisions of firms. When a firm chooses a particular location
for its produclion facility, the resulling concentration of employment either generates
a new city, or, more often, causes an existing city to grow. Although location theory
is usually cast in terms of where new firms locate, it applies as well to the expansion
of existing firms. The economic conditions that attract new firms to a city also make
it profitable for firms already in the city to expand their operations. In other words,
Jocation theory helps us explain both why cities arise in particu lar locations and why
sorne cities grow more rapidly than others.
Figure 4-1 provides a useful backdrop for our discussion of location decisions
and urban growth. A city will grow if the number of new jobs gained from the birth
of new firms and the expansion of existing firms exceeds the number of old jobs
lost from firm deaths and contractions. The question is whether growing cities have
relatively large job gains, or relatively small job losses. T he clear message from
Figure 4-1 is that cities differ in their job gains, not their job losses. Although most
cities have roughly the same percentage of job losses, growing cities have larger
gains from births and expansions.
The location decisions of firms are based on profit maximization. A firm's po-
tential profit varíes across space far several reasons. First, it is costly to transport
inputs and outputs, and locations with relatively low transport costs will generate
higher profits, ceteris paribus (everything else being equal). Second, sorne inputs
cannot be transported at ali, and locations with inexpensive local (nontransferable)
inputs will generate higher profits, ceteris paribus. Third, sorne firms benefit from
proximity to other firms in the same industry (localization economies) and other
firm s benefit from being in a large diverse city (urbanization economies). Fourth,
the public sector Jevies taxes and provides public goods and services, and Joca-
tions with a relatively efficient public sector will generate highcr profits, ceteris
paribus.

65
66 Part 1 Markct Forces in the Development ofCitics

FI GURE 4-1 Gross Gains and Losses in Employment in Selected Cities, 1984-1986

1 Openings and expansions

40- O Closings and contractions -------------------------- --
35-- ----- - - - - - - - - ---- - ----- - - - - - - - - - - - - - - --- --

...
30 - ~ ------ -------------------- ------- -- --- 1--

--- --
el)
e: ,__ --
"' 25-
..e:
o
~------------- - --

-
.,
E
...~
!l.
20- ,__
-- --
,...
,... - -- -- ,_ __
,... --
,... - --- -
15->-- -- -- >-- --

10 -- ,__ --
5-- --
o o oel) <> oel)
..e:
~
~
o ·g ~
..... a
~
:.e"'o. "'
.Q
-~
8.
~
"'
E ...
:::> >- ¿; o"'
:.e .,,-¡; "
o .2
i5
~
:::> e:
...~ a)
"'"'"'e: u " "'e: CI)
~ e:
o: z ~
:.e
CI)
'"
:.,; '"
CI)
:.::"' !l. ..=
So11rce: Table 2.3 in Randall W. Ebcr~~ ancl Joc A. Sione. Wage <md Adj11.i1111e111 in Local Labor Markers
(Kalamazoo. MI : Upjohn h1>1i1u1c. 1992).

TRANSFERABLE INPUTS AND OUTPUTS

A transfer-oriented firm is defined as one for which transportation cost is the dom-
inant factor in the location decision. The firm chooses the location that minimizes
total transport costs, defined as the sum of procurement and distribution costs. Pro-
curement cost is the cost of transporting raw materials from the input source to the
production facility. Distribution cost is the cost of transporting the firm's output from
the production facility to the consumer.
T he classic model of a transfer-oriented firm has fou r assumptions that focus
attention on transportation costs as the dominant location variable.

• Single transferable output. The firm produces a fixed quantity of a single

product, which is transported from the production facility to a market at point M.
• Single transferable input. The firm may use severa! inputs, but only one input
is transported from an input source (point F) to the firm's production facility.
All other inputs are ubiquitous, meaning that they are available at all locations
at the same price.
Chapter 4 Where Do Firms Locate? 67

• Fixed-factor proportions. The firm produces its fixed quantity with fixed
amounts of each input. In other words, the firm uses a single recipe to produce
its good, regardless of the prices of its inputs. There is no factor substitution.
• Fixed prices. The firm is so small that it does not affect the prices of its inputs
or its product.
Under these four assumptions, the firm maximizes its profit by minimizing its
transportation costs. The firm's profit equals total revenue (price times the quantity
of output) less input costs and transport coses. Total revenue is the same at ali
locations because the firm sells a fixed quantity of output ata fixed price. Input costs
are the same ae ali locations because the firm buys a fixed amount of each input
at fixed prices. The only costs that vary across space are procurement costs (the
costs of transporting the firm 's transferable input) and distribution costs (ehe costs
of transporting the firm's output). Therefore, ehe firm will choose the location that
minimizes its total transport costs.
The firm's location choice is determined by the outcome of a tug-of-war. The
firm is pulled toward its input source because the closer to the input source, the lower
the firm's procurement costs. On the other side, the firm is pulled toward the market
because proximity to the market reduces the firm 's distribution costs.

Resource-Oriented Firms
A resource-oriented firm is defined as a firm that has relatively high coses for trans-
porting its input. Table 4-1 shows the transport characeeristics for such a firm. The
firm produces baseball bats, using 10 tons of wood to produce 3 tons of bats. The
firm is involved in a weight-losing activity in the sense that its output is lighter than
its transferable input.
The key factors in the tug-of-w~r are the monetary weights of the firm's input
and output. The monetary weight of the input is equal to the physical weight of
the input ( 1O tons) times the transportation rate ($1 per ton per mile), or $1 O per
mi le. Similarly, the monetary weight ofthe oueput is 3 tons times $2, or $6 per mi le.
T he firm is considered a resource-oriented fi rm because the monetary weight of its
transferable input exceeds the monetary weight of its output. Although the unit cost
of transporting output is higher (because finished bats must be packed carefully,
but logs can be tossed onto a truck), the loss of weight in the production process
generales a lower monetary weight for the output.
Figure 4-2 shows the firm's transportation costs. If we use x as the distance
from the input source (the forest) to the production site (the factory), the firm's

TABLE 4-1 Monetary Weights for a Resource-Oriented Firm

Input (wood) Output (bats)
Physical weight (tons) 10 3
Transpon rate (cost per ton per mile) $1 $2
Monetary weight (physical weight times rate) $ 10 $6
68 Part 1 Markct Forces in thc Development of Cities

FIGURE 4-2 Total Transport Cost for Resource-Oriented Finn

$

100

80

60

40

20
Market (M)

2 4 6 8 10
x (Distance from forest)
Forcst (F)

Total transpon co~t (thc sum of procurement cost and distribution cost) is minimi zed at the fores1
bccause the monctary wcight of thc input ($JO) excccds the monetary wcight of thc output ($6). Thc
weight-gaining activiry locates at its source of raw materials.

procurement cost is
PC = W¡. (¡.X

that is, the monetary weight of che input (the physical weight w; times the trans-
port cost rate t;) Limes the distance between forest and factory. The slope of the
procurement-cost curve is the monetary weight of the input, so PC rises by $ 1O per
mil e, from zero at the forest to $100 at the market 1O mi les away.
The distribution costs are computed in an analogous way. If XM is the distance
between the forest and the market, the firm 's distribution cost is

DC = W0 • 10 • (xM - x)
that is, the monetary weight of the output (weight w0 times the transport cost rate t0 )
times the distance from the factory to the market. The slope of the distribution-cost
curve is the monetary weight of the output, so as wc move from the forest toward
the market, DC decreases by $6 per mi le, from $60 at the forest ( 1O miles from the
market) to zero at the market.
Total transport cost is the sum of procurement and distribution costs. In Fig-
ure 4- 2, total transport cost is minimized at the forest site at $60. To see why
transpon cost is minimized here, suppose the firm started at the forest site and
then moved one mi le toward the market. Its distribution cost would decrease by $6
(the monetary weight of bats) but its procurement cost would increase by $ 1O (the
Chapter 4 Where Do Firms Locate?
~18L/OTECA r ---
1_ll\f) v'l-1\...:JIL, ,u U._¡. ¡ -.¡I_¡ ._

monetary weight of the wood), so its total transport cost would increases by $4. The
69

firm's total transport cost is minimized at the forest because thc monetary weight
of the input exceeds the monetary weight of the output. The resource-oriented firm
locates near its input source.
The bat firm is resource-oriented because it is a weight-losing activity, using
10 tons of wood to produce only three tons of bats. The cost of transporting wood
is large relative to the cost of transporting the finished output, so the firm saves on
transport costs by locating near the forest. In this case, the tug-of-war is won by the
input source because there is more physical weight on the input side. Here are sorne
other examples of weight-losing firms.

1. Beet-sugar factories locate near sugar-beet farms because one pound of sugar
beets generates only about 2.7 ounces of sugar.
2. Onion dehydrators locate near onion fields because one pound of fresh onions
becomes less than one pound of dried onions.
3. Ore processors locate near mines because they useonly a fraction ofthe materials
extracted from the ground.

Sorne firms are resource oriented because their inputs are relatively expensive
to transport. Considera firm that cans fruit, producing one ton of canned fruit with
roughly a ton of raw fruit. The finn's input is perishable, and must be transported in
refrigerated trucks, while its output can be transported less expensively on regular
trucks. Because the cost of shipping a ton of raw fruit exceeds the cost of shipping a
ton of canned fruit, the monetary weight of the input exceeds the monetary weight
of the output, and the firm will locate near its input sourcc, a fruit farm.
In general, a firm's input will be more expensive to ship if it is more bulky,
perishable, fragile, or hazardous than the output. Hoover (1975) provides some
examples of such activities:

l. Cotton baling. The input (raw cotton) is more bulky than the output (baled
cotton). The cost of shipping a ton of ftuffy cotton exceeds the cost of shipping
a ton of compacted cotton, so the monetary weight of the input is higher and the
resource-oriented cotton baler will locate near the cotton field.
2. Skunk deodorizing. The input (fully armed skunks) is more fragile and haz-
ardous than the output (disarmed skunks). The cost of shipping a ton of armed
skunks exceeds the cost of shipping a ton of disarmed ones, so the skunk de-
odorizer will locate near the skunk farm.

In general, when a firm's input is relatively bulky, perishable, fragile, or hazardous,

the tug-of-war wíll be won by the input source, not because the input is heavier, but
because it is more expensive to transport.
There are many examples of industries that locate close to their transportable
inputs (Ellison and Glaeser, 1999). The producers of soybean and vegetable oil
are concentrated in Nebraska, North Dakota, and South Dakota, close to the farms
that supply soybeans and corn. Milk and cheese producers are concentrated in South
70 Part l Market Forces in the Development of Cities

Dakota, Nebraska, and Montana, close to dairy farms. Sawmills and other wood pro-
cessors are concentrated in Arkansas, Montana, and Idaho, close to vast timberlands.

Market-Oriented Firms
A market-oriented firm is defined as a firm that has relatively high costs for trans-
porting its output to the market. Table 4-2 shows the transport characteristics for
such a firm. The bottling firm uses one ton of sugar and three tons of water (a ubiq-
uitous input) to produce four tons of bottled beverages. The firm is involved in a
weight-gaining activity in the sense that its output is heavier than its transferable
input. The monetary weight of the output exceeds the monetary weight of the input,
so this market-oríented firm will locate near its market.
As shown in Figure 4-3, the firm's transport cost is minimízed at the market.
Because the monetary weight ofthe output exceeds the monetary weíght ofthe input,
a one-mile move away from the market íncreases the distribution cost by more than

TABLE 4-2 Monetary Weights fo r a Market-Oriented Firm

Input (sugar) Output (beverage)
Physical weight (tons) 1 4
Transpon rate (cost per ton per mile) SI SI
Monetary weight (phy~ical weight times rate) SI $4

FIGURE 4-3 Total Transpon Costs for a Market-Oriented Firm

$
50

40

30

20
Procurement cost

10 -------- Markct (M)

2 4 6 8
x (Distance from input source)
Sugar plantation (F)

Total transpon cost (PC + DC) is minimized at the market becau~c the monctary weight of the output
($4) exceeds the monetary weight of the transferablc input ($1 ). Thc weight-gaining activity locatcs
at its market.
Chapter 4 Where Do Firms Locate? 71

it decreases procurement cost. Specifically, such a move increases distribution cost

by $4, but decreases procurement cost by only $ 1, for a net loss of $3. For chis
weight-gaining activity, the tug-of-war between input source and market is won by
the market because there is more physical weight on the market side.
Sorne firms are market-orienled because their output is relatively expensive to
transport. A fi rm 's output wil 1be relatively costly to transpon i f it is bulky, perishable,
fragile, or hazardous. Hoover ( 1975) provides several examples of such firms:
• For an automobile assembly firm, the output is more bulky than the inputs- the
metal, plastic, and rubber parts that become part of the assembled automobile.
Because the cost of shipping a ton of automobiles excecds the cost of shipping a
ton of component parts, the rr.onetary weight of the outpul exceeds the monetary
weight of the inputs, pulling the firm toward thc market.
• For a bakery, the output is more perishable than its inputs: bread becomes stale
in a matter of hours, while flour can be stored for months without spoiling. To
ensure timely delivery of fresh bread, the bakery locates near its market.
• A weapons producer combines harmless inputs into a lcthal output. The firm
locates near its output market to avoid transporting the hazardous (or fragile)
output long distances.
In general, when a firm's output is relatively bulky, perishable, fragile, or hazardous,
the tug-of-war will be won by the market, not because the output is heavier, but
because it is more expensive to transport.

Scale Economies in Transpor tation

Up to this point, we have assumed that the cost of transporting inputs and outputs
is proportional to the distance transported, regardlcss of distance. In fact, we expect
the average cost of transportation to decrease with distance for two reasons.
• Fixed cost. There are fixed costs associated with processing paperwork, and
loading and unloading a shipment. Thcrefore, the average cost of transportation
decreases as the distance increases.
• Line-haul economies. The transpon cost per mile decreases as the distance
increases, reflecting the efficiencies from using different modes for different
distances. A firm may use a truck for a short haul, but a train ora ship for a long
haul. The unit transport cost is lower for trains and ships. so the average cost
decreases as distance increases.
The presence of scale economies in transportation reinforces the tendency for
a firm to locate at either its input sources or its markets. In the absence of scale
economies, if the monetary weight of the output equals the monetary weight of
the input, the firm would be indifferent between the two locations and any points
in between, and could choose sorne location in the middle. But if there are scale
economies in transportation, it would not be sensible to locate anywhere between che
two points. It will be more efficient to make one long trip (transporting eitherthe input
or the output) instead of two short trips (transporting both the input and the output).
72 Pan 1 Market Forces in the Development ofCities

The Principie of Median Location

Thc classic model of the transfer-orientcd firm assumes that the firm has a single
input source anda single market. For more complex cases involving multiple inputs
or markets, we can use the principie of median location to predict where a firm
will locate: The optimum location is the median transport location, the location that
splits the total monetary weight of the firm into two equal halves. At the median
location, hal f the monetary weight comes from one direction, and half from the other
direction.
Consider the location decision of Ann, who makes and delivers pizzas to con-
sumers along a highway. The characteristics of her operation are as follows.

1. Ali inputs (labor, dough, toppings) are ubiquitous (available at all locations for
the same price), so input transport costs are zero.
2. The price of pizzas is fixed, and each consumer along the highway demands one
pizza per day.
3. Ann bears the delivery cost, which is $2 per pizza per mile traveled. Each pizza
delivery requires a separare trip.

Ann will pick the location that minimizes her total delivery cost.
Figure 4-4 shows the distribution of consumers along the highway. Distances
are measured from the western end of the highway (point W). There are 2 customers
at point W. 8 customers at point X (one mi le from W), 1 customer at Y (two miles
from W) , and 1O customers at Z (nine miles from W). Because each customer buys
one pizza and the delivcry cost is $2 per pizza per mile, the monetary weight of a
particular location (the sales volurne times the delivery cost per pizza per mile) is
twice the number of consumers at that location.
Using thc principie of median location, Ann will minimize her transport cost at
the median location. Point Y divides the monetary weights into two equal halves,
so it is the median location. For the locations to the west of Y, the monetary
weight is $1 O (equal to $2 for W plus $8 for X); for thc locations to the east,
the monetary weight is $10 (equal to $10 for Z). The median location divides Ann's
customers into two equal halves; she has 1O customers to the east, and JO to the
west.

FIGURE 4-4 Pizza Oclivery and the Principie of Median Location

w X y s z

Distancc from IV o 2 3 9
Number of consumers 2 8 10
Monclary weighl $4 $16 $2 $20

Ann Jocale!> her pizza parlor al poinl Y bccause il i' thc median location: she delivers JO pizzas to
consume1~5 lO the wesl of Y, and JO lO consumers lO the east of Y. A move f'rom Y to S would
dccrease delivcry cosl for the 10 consumers at point Z (savings of $20). bul increase dclivcry costs
for the 11 consumers al points W. X. and Y (increasc in cost of $22). A movc in the opposite
direction would also increasc total transpon costs.
Chapter 4 Whcrc Do Fiml'> Locate? 73

FI GURE 4-5 Median Location in thc Largc City

Location~ L

Dcmand 4 4 4 4 17

Thc median location is in the largc city (l). Any movc to the left of point L will increase rravcl CO'>l5
for thc majority of con~umcrs. incrca~ing total transponation costs.

To show that the median location minimizes total transport costs, suppose that
Ann starts at the median location, and then moves to point S, one mi le east of Y. As
she moves to the east. the good news is that she spends less on delivery to point Z:
she saves $2 per trip to Z, saving a total of $10 in eastward dclívery costs. The bad
news is that westward costs íncrease: shc pays $2 more per tríp to points W , X, and
Y; sin ce there are 11 customcrs to the west of S, her wcstward del ívery costs increase
by a total of $22 ($4 for W, $16 for X, and $2 for Y). Since the increase in westward
costs cxceeds the decrease in eastward costs, the move from Y to S increases totaJ
delivery costs. Thc same is true for a movc in the opposite direction; if Ann moves
from Y toward W, her delivery costs will increase.
The median location minimizes total transport costs because it splits pi1.1.a con-
sumers into two equaJ parts. As Ann moves eastward away from Y, shc moves
furthcr away from 11 customers, but moves closer to only 1O customers. Similarly,
a westward movc will cause her to move closer to 1O customers, but furthcr from
11 customers. In general, any move away from the median location will increase
delivery costs for the majority of consumers, so total delivery cost increases. It is
important to note that the distance between the consumers is irrelevant to thc firm's
location choice. For examplc, if the Z consumers were Iocated 100 miles from W
instead of 9 miles from W, the median location would still be point Y. Total delivery
costs would still be minimized (ata higher leve!, of course) at point Y.
The principie of median location provides another explanatíon of why large
cities become larger. Consider a firm that delivers its product to consumers in five
different cities. In Figure 4-5, there is a large city at location L, and four small cities
at localions S 1 , S2, S3, and S4. The firm sells 4 units in each small city. and 17 units
in the large city. The median location is in the large city, even though the largc city is
at the cnd of the line. A one-mile move wcstward from L would dccrease transport
costs by $ 16 (as the firm moved closer to consumers in the small cities), but would
increase transport costs by $17 (as the firm moved away from consumers in the large
city). The lesson from this example is that the concentration of demand in large cities
causes large cities to grow.

Transshipment Points and Port Cities

· The principie of median location also cxplains why sorne industrial firms locate
at transshipment points. A transshipment point is defincd as a point at which a good
is transferred from one transport mode to another. At a port, goods are transferred
from trucks or trains to ships; ata railroad terminal, goods are transfcrred from trucks
to the train.
74 Part 1 Marke1 Forces in the Development of Cities

FIGURE 4-6 The Sawmill Locates at the Port

1npul so urce A
Monelary weigh1 = S15

Outpul markel

M
Monetary weigh1 =SIO

1npu1 so urce 8
Moneiary weight = S 15

The firm locales i1s sawmill al 1he pon ( P ) because il is the median transpon location. A move
from P toward either A or 8 would increase output transpon cosis by $10 without afTecting
input transpon costs. A move from P toward M would increase input transpon costs by $30
but decrease outpu1 transpon costs by only $10.

Figure 4-6 shows the location options for a sawmill. The firm harvests logs
from locations A and B, processes the logs inlo Jumber, and then sells lhe Jumber in
an overseas market at point M. Highways connect points A and B to the port, and
ships travel from the port to point M. The sawmill is a weight- losing activity: The
monetary weighls of the inputs are $15 for point A and $15 for point B, and the
monetary weight of the output is $ 1O.
Where will the firm locate its sawmill? Although there is no true median location,
the port is the closest to a median location. If the firm starts at the port ( P), it could
move either toward one of its input sources orto its market.
1. Toward input source A. A one-mile move from P toward point A will cause
offsetting changes in the costs of transporting logs from the two input sources:
the cost of logs from A would decrease, but the cost of logs from 8 would
increase. At the same time, the cost of transporting output would increase by
$ 1O. G iven the offsetting changes in input transport costs and the increase in
output costs, the port location is superior to locations between P and A. The
same argument applies for a move from P toward B.
2. To market (M). Unless the firm wants to operate a ftoating sawmill, it would
not move to points between the port and the overseas market at M. It could,
however, move ali the way to the market. A move from P to M would decrease
output transport cost by $ 1O (the monetary weight of output) times the distance
between M and P , and increase input transport cost by $30 (the monetary weight
of the inputs) times the distance. Therefore, the port location is superior to the
market location.
Although the sawmill is a weight-losing activity, it will locate at the port, not at
one of its input sources. The port location is efficient because it provides a central
collection point for the firm 's inputs.
Chapter 4 Where Do Firms Loca1e? 75

There are many examples of port cities that developed as a resuh of the loca-
tion decisions of industrial firms. Seattle started in 1880 as a sawmill town: Firms
harvested trees in western Washington, processed the logs in Seattle sawmills, and
then shipped the wood products to othcr states and countries. Baltimore was the
nation's first boomtown: Flour milis processed wheat from the surrounding agricul-
tura! areas for expon to the West lndies. Buffalo was the midwestern center for flour
milis, providing consumers in eastern cities with flour produced from midwestern
wheat. Wheat was shipped from midwestern states across che Great Lakes to Buffalo,
where it was processed into ftour for shipment, by rail, to cities in the eastern United
States. In contrast with Baltimore, which exported its output (flour) by ship, Buffalo
imported its input (wheat) by sh ip.

ENERGY ORIENTATION

What is the role of energy in the location decisions of firms? ln the first half of the
nineteenth century, energy was a local input, defined as an input that could not be
transported from one location to another. The waterwheel was the first device used
to generate nonanimal mechanical energy. The watcrwheel was turned by waterfalls
and fast-moving streams, providing powcr for production racilities located along
rivers and streams. Textile manufacturers built factories along small backcountry
streams in New England, with waterwheels powering their machinery. Among the
cities that dcveloped as a resu lt of the waterwheel are Lowell, Lawrence, Holyoke,
and Lewiston.
The development ofthe steam engine in the second half ofthe nineteenth century
made energy a transportable input. The steam engine could be operatcd anywhere,
with the only constraint being the availability of coal to fuel the engine. Sorne energy-
intensive manufacturers located near the coal mines in Pennsylvania. Others located
along navigable waterways and shipped coal from the mines to their factories. In
New England, textile firms shifted from backcountry watcrfall sites to sites along
navigable waterways. Production shifted to the Fati River-New Bedford area along
the south coast of New England. The development of the railroad gave coal users
another transport option, causing the development of production si tes along raíl lines.
The development of electricity affected the location patterns of manufacturcrs
Because electricity can be transmitted over distances of severa! hundred miles, a firm
can use the energy generated from water power without locati ng along a backcountry
stream. Similarly, a firrn can use coal resources without shipping the bulky fuel frorn
the mine to its factory. In general, the development of electricity decreased the
irnportance of energy considerations in location decisions.
For so me activities, the availability of cheap energy is sti 11 an important location
factor (Ellison and Glaeser, 1999). Sorne intensive users of electricity (the production
of primary aluminum and chlorine), locate in the Pacific Northwest to exploit cheap
electricity from hydroelectric power. The producers of clay ti les and fertilizer are
intensive users of natural gas, and locate in Arkansas, Louisiana, and Texas, where
natural gas is cheap. The production of cernent and lime requires large amounts
76 Part 1 Market Forces in the Development of Cities

of coal, and firms in these industries locate in Montana, Nevada, and Wyoming to
exploit cheap coal.
How will the deregulation of electricity markets affect the location decisions
of energy-oriented industries? In the regulated market of the I 990s, there was sub-
stantial variation in the price of electricity across states, a result of regulations that
restricted the ftow of electricity across state lines. Deregulation is expected to free
producers to "wheel" power from state to state, and the free ftow of electricity is
expected to reduce the spatial variation in the price of electricity. As a result, fewer
firms will choose locations based on the price of electricity.

LOCALIZATION AND URBANIZATION ECONOMIES

As we saw in Chapter 3, sorne firms are subject to localization economies and

urbanization economies. A firm that benefits from localization economies will be
pulled toward other firms in the industry. For example, the corporate headquarters
of different firms are pulled toward each other because in a cluster they can share
advertising firms and firms supplying intermediate inputs that require face-to-face
contact between buyers and sellers. Firms in the cluster indirectly exploit the scale
economies in producing the intermediate input, and can tap the input suppliers at
top speed. Similarly, firms that benefit from labor pooling will locate close to one
another to draw workers from a common pool. Firms that benefit from knowledge
spillovers will locate close to one another to share the knowledge that leads to
improved products and production processes.
Urbanization economies refer to the cost savings associated with producing
in a large city. A firm that benefits from urbanization economies will be pulled
toward large cities, with their inexpensive intermediate goods and services, extensive
knowledge spillovers, and large labor pool. Urbanization economies lead to self-
reinforcing growth, as firms locate in large cities because so many other firms are
already located there.

LABOR MARKETS AND LOCATION CHOICES

What is the role of labor in location choices? On average, labor is responsible for
about three-fourths of the cost of production, so firms' location decisions are sensitive
to the cost of labor and labor productivity. Labor is a local input in the sense that it
is impractical for workers to commute outside metropolitan areas.

Transport Cost versus Labor Cost

We can modify our model of a firm's location choice to incorporate the cost of labor.
To simplify matters, suppose the firm gets its inputs and sells its output at the same
location (T). The productivity of labor is the same at ali locations, but the price of
labor (the wage) decreases as the firm moves away from location T . In Figure 4-7
the firm 's total cost is the sum of labor costs and transport costs, which is minimized
Chapter 4 Where Do Firms Locate? 77

FIGURE 4-7 Transpor1 Costs versus Labor Costs

50

40

30

20

10

T 2 4 6 8 10
Disiance from market and inpu1s

(A) Firm locates clo\e to market and inputs

50

40

T 2 4 6 8 10
Disiance from market and inpu1s

(B) Firm loca1es far from market and inputs

When inputs and ou1pu1s are rclatively heavy and costly 10 transport, thc firrn
minimizes the sum of transport and labor costs by locating close 10 the 111arke1 and
inputs (Panel A. with cost = S~O as shown by point i). A decrease in the physical
weight or unit freight cost caus:s the firm to locate far from the market and inputs.
where labor costs are lower (Panel B. with total cost = $10 as shown by point j).

at point T. In this case, transport costs are high relative to the variation in labor
costs, meaning that the transport-cost curve is relatively steep. As a result, the forces
pulling the finn toward location Tare stronger than the forces pulling the firm toward
locations with lower wages and labor costs, so the tug-of-war is won by Jocation T.
78 Part 1 Markct Forces in the Development ofCitics

In the last several decades, transport costs have become less important in firms'
location decisions. In many industries that have traditionally been resource-oriented
or market-oriented, firms now locate far from their input sources and markets. The
changes in locational orientation resulted from innovations in transportation and
production that have decreased transport costs.
• Transportation technology. The developmcnt of fast ocean ships and con-
tainer technology decreased shipping costs, while improvements in railroads
and trucks lowered the cost of overland travel. Faster and more efficient aircraft
have decreased the cost of afr travel.
• Production technology. Improvements in production techniques decreased the
physical weight of inputs. For example, the amount of coal and ore required to
produce one ton of steel has decreased steadily, a result of improved produc-
tion methods and the use of scrap metal (a local input) instead of iron ore (a
transportable input).
Panel B of Figure 4-7 shows the effects of a decrease in the unit transport costs
and the physical weights of inputs. These changes flauen the transport-cost curve,
shifting the cost-minimizing location to a location 1O miles from the market and
inputs, where labor costs are lower. The decrease in transport costs causes the firm
to switch from transfer orientation to labor orientation.
As unit transport costs and input weights decrease, firrns are more likely to base
their location decisions on access to inexpensive local inputs rather than access to
transportable inputs. A recent example is the movement of the assembly operations
of many U.S. manufacturers to sites along the Mexican border. The steel industry
has moved from the castern United States, with its rich coa! and ore deposits, to
Brazil, Korea, and Mexico- far from both raw materia Is and steel markets. Similarly,
manufacturers have moved from the United States to Asia and Mexico, far from U.S.
markets, because the savings in labor costs are greater than the increase in transport
costs.

Labor in the Long Run

In the long run, workers can move between cities. In the typical year, about 6 percent
of the U.S. population move across county lines, and about 3 percent move across
state borders (Black, 2000). A firm that locates in one city could hire workers cur-
rently living in another city with the expectation that the workers will transfer to the
firm's city. According to Bartik ( 199 1), the vasl majority of new jobs in cities are
filled with people who rclocate from other cities: When the number of jobs in a city
increases by 1,000, on average, only 230 jobs go to current residents; the rest go to
workers who transfcr from other cities.

Labor and Natural Amenities

Consider a nation that has two regions-with warm weather in the North and cold
weather in the South- and workers prefer warm weather. To achieve locational
Chaptcr 4 Where Do Firms Locaie? 79

equi librium for workers, wages in the South must be lower than wages in the North;
otherwise, there would be an incentive for workers to move to the South, getting
the same wage and better weather as well. In equilibrium, the wage for a given skill
leve] will be lower in the South, and northern workers will be compensatcd for bad
weather by a higher wage.
The lower wage caused by weather will encourage firm s to locate their produc-
tion facil ities in the South. In this case, the local nature of weather and workers'
preference for warm weather makes labor a local input. The firm 's location depends
on the location choice of its workforce: instead of workers following firms. fim1s
follow workers. Graves ( 1979) and Porell ( 1982) provide em pirical support for this
phenomenon. There is evidence that weather has played an important role in location
decisions and urban growth. In the past few decades, the most rapidly growing urban
areas are ones with warm, dry weather (Black and Henderson, 1999; Glacser, Kolko,
and Saiz, 2001 ).
Natural amenities appear to be most important for firms that employ high-
income workers. Since the demands for these amenities are income-clastic, high-
income workcrs are attractcd to locations with amenities, and the firms hiring these
people follow. For example, research and development firms employ engineers and
computer scientists. who place a high value on good weather anda clean cnvironmenl.
One explanation for the shi ft of employment from northern states to the southern and
western states is that rising income has increased the dcmand for natural amenities,
causing workers to move to areas that provide these amenities.

Learning and Location

What is the role of learning on firms' location choices? As we saw in Chapter 1, cities
promote learn ing by faci li tating contact bctween workers of different ski ll levels,
causing learning by imitation. A city that provides better learning opportunities will
attract workers who are eager 10 learn as well as firms 1hat are eager 10 hire such
workers. Productivity is higher in places 1ha1 provide bettcr learning opportunities,
attracting firms to those places.
Learning opportunities in c ities depend on the average education lcvel of the
city's workers. Rauch ( 1993) shows that an increase in a ci ty's average education
level provides a better environment for learning, and increases the productivity of
individual workers. In other words, a worker learns more and is thus more productive
if he or she is suJTounded by better educated people. Onc implication of this result
is that when a particular worker moves from a city with a low average education
leve! to one with a high education level, the worker becomes more produclive and
earns a higher wage. Specifically, a one-year increase in 1he average education level
increases a worker's productivity by 2.8 percent.
Human capital is not distributed uniformly across space. As we saw in Chapter 1,
college graduales are overrepresented in c ities, especially la rge ones. Black (2000)
computes an index of dissimilarity for college graduates in U.S. citics. The index
shows the percentage of college graduates chat would have 10 switch cicies in arder to
cnsure an even distribution of college graduates across cities. For example. suppose
80 Par1 1 Markel Forces in the Development ofCi1ics

that for the nation as a whole, 25 percent of the metropolitan workers are college
graduates. A dissimilarity index of 12 would indicate that 12 percent of ali graduates
would have to switch cities to ensure that in every city 25 percent of workers would
be college graduates. For a set of 318 metropolitan areas, the dissimilarity index
increased from 10.2 in 1940, to 11.6 in 1970, to 14.9 in 1990. In other words, the
concentration of college graduates in cities is substantial and growing.
The u neven distribution of human capital has important implications for location
decisions and urban growth. As we saw earlier, cities with above-average education
levels will have larger productivity gains, and will attract more firms and thus grow
more rapidly.

LOCAL PUBLIC SERVI CES AND TAXES

Local governments inftuence business location decisions because the government

provides public services and imposes taxes to pay for them. The public sector pro-
vides sorne of the inputs used directly by firms (e.g., water, roads, ports, and scwers)
and provides goods consumed by households (e.g., education, parks, roads, transit
systems, public safety). A city can attract households and firms by providing efficient
public services at a relatively low cost in terms of taxes.

Taxes and Location Decisions

In the last few decades, there have been dozens of studies examining the effect of
local taxes on location decisions and urban economic growth. Bartik ( 1991) draws
sorne general conclusions from these studies. There is solid evidence that local taxes
have a strong negative effect on regional business growth: a high-tax city wiU grow
ata slower rate than a low-tax city, everything e/se being equal. Of course, one of thc
items included in everything e/se is public services: A high-tax city will grow ata
slower rate if firms pay higher taxes without receivi ng better public services in return.
We can distinguish between two types of location decisions, intercity decisions
(choosing a city or metropolitan area) and intracity decisions (choosing a site within
a city or metropolitan area). We can summarize the effects of taxes on location
decisions by computing the elasticity of business activity with respect to tax liabili-
ties, defined as the percentage change in business activity divided by the percentage
change in tax liabilities. For intercity location decisions, the elasticity is between
-0. JO and -0.60-a IO percent increase in taxes in a particular metropolitan area
decreases business activity in the metropolitan arca between 1 percent and 6 per-
cent. For intracity location decisions, the elasticity is between -1.0 and -3.0. If
an individual municipality increases its taxes by 1Opercent, business activity in the
municipality decreases between 1Opercent and 30 percent.
The elasticity for the intracity decision is larger because different locations
within a metropolitan area are better substitutes than locations in different metropoli-
tan areas. Two other results from recent empirical studies are worth noting. First,
Chapter 4 Where Do Firms Locate? 81

it appears that manufacturers are more sensitive than other types of firms to tax
differentials within and between metropolitan areas. This is sensible because manu-
facturers are likely to be oriented toward the national markel and thus have a widcr
range of Iocation options. Second, metropolitan areas with relatively high taxes on
capital (in the form of taxes on business property) tend 10 repel capital-intensive
industries and attract labor-intensive industries.

Public Services and Location Decisions

T here is solid evidence that the provision of local public services has a strong
positive effect on regional business growth. If two cities differ only in the quality
of their local public services, the city with the better public services will grow at
a faster rate. Similarly, if a city improves its public services, it will grow faster,
everything else being equal. It's important to note that one of the items included in
everything else is local taxes: the high-service city grows at a faster rate because
firms will receive better public services without paying higher taxes. The public
services that have the largest positive effect on business growth are education and
infrastructure.
How would simultaneous increases in taxes and spending on public services
affect location choices and business activity? Studies by Helms ( 1985) and Munnell
( 1990) suggest that the effect of a tax increase depends on how the extra tax revenue
is spent. If the extra tax revenue is spent on local public services (infrastructure,
education, or public safety), the tax/expenditure program increases the relative at-
tractiveness of the city and promotes economic growth. In contrast, if the extra tax
revenue is spent on redistributional programs for the poor, the tax decreases the
relative attractiveness of the jurisdiction and decreases the growth rate.

Subsidies and Incentive Programs

Many cities try to attract new firms by offering special subsidies. Levy ( 1981) de-
scribes severa! types of subsidy programs, including the following:

1. Tax abatement. In sorne cities, new firms are exempt from local property taxes
for a fixed period (e.g., 10 ycars). Sorne cities offer tax abatements to all new
developments, while others offer abatements only to firms that are particularly
sensitive to tax differentials.
2. Industrial bonds. Sorne cities issue tax-free industrial bonds to finance prop-
erty developments. The local government uses the revenue from the bonds to
purchase the land and leases the property to a private firm. Because the interest
income from industrial bonds is not subject to federal taxes, the bond buyer
accepts a relatively low interest rate (e.g., 8 percent instead of 12 percent).
Therefore, the lessee pays less than the markel interest rate on the money bor-
rowed to finance the project. The use of industrial bonds was sharply curtailed
by the Tax Reform Act of 1986.
82 Part 1 Market Forces in the Development ofCities

3. Government loans and loan guarantees. Sorne cities loan rnoney directly
to developers, and others guarantee loans frorn private lenders. In both cases,
developers borrow rnoney ata relatively low interest rate: Either the city charges
an interest rate below the rnarket rate, or the city decreases the risk associated
with a prívate loan, allowing the developer to borrow prívate rnoney ata relatively
low interest rate.
4. Site development. Sorne cities subsidize the provision of land and public ser-
vices for new development. The city purchases a site, clears the land, builds
roads and sewers, and then sells the site to a developer at a fraction of the cost
of acquiring and developing the si te.

Wassrner (1994) explores the effects of local econornic-development programs

on business activity in the Detroitmetropolitan area. Municipalities in the area differ
in their use of industrial development bonds and their property-tax abaternent pro-
grarns for cornrnercial and industrial property. Wassrner shows that these programs
have a positive effect on business activity in only 5 of 16 cases. For the other 11 cases,
the prograrns have either no effect on business activity, ora negative effect. Severa!
studies of enterprise zones (areas of a city where firms pay low tax rates, receive
subsidies for worker training, and are exempted from local regulations) suggest that
such zones are not very effective in luring firms to their areas (Boarnet and Bogart,
1996; Papke, 1993 and 1994; Dowall, 1996).
There are man y anecdotes about firms shopping around for the best tax-abatement
deal. The classic example is a firm that secretly decides to locate in a particular city
but then goes to an inferior city and asks for a tax-abatement package. With the
"bid" of the inferior city in hand, the firm then asks the preferred city to match
the tax package. The preferred city rnatches the tax package, and the firrn locates
in that city, which it had intended to do even in the absence of any tax incentives.
A recent study of property-tax abatement (Anderson and Wassmer, 1995) provides
evidence that this sort of tax shopping occurs frequently. The authors suggest that
state policymakers should develop policies to discourage this sort of cornpetition
between cities.

Professional Sports, Stadiums, and Jobs

In San Francisco in 1997, billboards around the metropolitan area read, "Build the
Stadium-Create the Jobs." That was the carnpaign slogan for the proponents of a
plan to use $100 mi Ilion of public money to help pay for a new football stadium
for the San Francisco 49ers (Noll and Zirnbalist, 1997). San Francisco is one of
many cities that have subsidized the construction of faci lities for professional sports.
Between 1989 and 1997, thirty-one new stadiums were built, atan average cost of
about $150 mi Ilion. Are sports stadiums effective tools of econornic development?
Do they create jobs?
The logic behind the job-creation effects of sports stadiums is straightforward.
To attract a professional sports tearn (orto retain an existing team), a city can build
a stadium. Like other organizations, a professional team sells a product and hires
Chapter 4 Where Do Fim1s Locate? 83

workers, including athletes, groundskeepers, ticket takers, accountants, and media

folks. In addition, sorne of the money earned by the team 's employees is spcnt in the
local economy, leading to an increase in employment in restaurants, dental offices,
and hardware stores. The question is: How many additional jobs does a professional
team cceate?
Despite the hyperbole of stadium proponents, the job-creation effects of sta-
diums are modest. The stadium for the Arizona Diamondbacks cost $240 million,
but increased total employment in the area by only 340 jobs. This figure includes
both the direct effect (people hired by the team) and the indirect effect (additional
jobs outside the team as money is spent and rcspent in the local economy). In other
words, the cost per job was $705,882. For other teams in Denver, Kansas City, and
San Diego, the employment gains were between 128 and 356 jobs. A study of cities
that host professional teams showed positive (and small) effects in only one-quarter
of cases (Baade and Sanderson, 1997). In about a fifth of the cases, the presence of
a sports team actually decreased total employment.
Why are the employment effects of sports teams so small? Most of the money
consumers spend on professional sports evcnts comes at the expense of local goods
such as movies and restaurant meals. When a sports team comes to town, a large
fraction of the money spent on the team is diverted from these other consumer
products. For example, there may be more popcom sellers in the stadium, but that
means there are fewer popcorn sellers in movie theaters. Similarly, a sports event
provides a different place to drink beer. To the extent that consumers switch from
movies and other local goods to sport events, the employment effects of sports teams
will be small. In fact, it could even be negative ifthe money spent on local consumer
goods is more labor-intensive than spending at sporting events.
T he real power of sports teams to incrcase employment comes from their ability
to attract money from outside the metropolitan area. When someone travels from
Peoría to Chicago to see the Bu lis, the $50 he spends on tickets, souvenirs, and food
adds money to the Chicago economy-and subtracts it from the Peoria economy. As
a resul t, total spending and employment in the Chicago economy increases. However,
because most of the money spent on sporting events comes from local consumers,
sports teams don't create many jobs.
AJthough professional sports teams have small employmcnt effects, that doesn 't
necessarily mean they are bad investments. One of the benefits of a team is the thrill
to fans of following a team as it competes on a national stage. In a city of 5 million
people, $250 mi Ilion invested in a stadium works out to $50 per person. lf the per
capita benefit over the life of the stadium is at least $50, the stadium makes economic
sense as a consumption good.

CASE STUDIES ANO EMPIRICAL RESULTS

This section discusses sorne of the facts on the location choices of firms. We'll look
at case studies of the location decisions of companies in the semiconductor industry,
Japanese and American auto firms, Mexican garment manufacturers, and the carpet
84 Part 1 Market Forces in the Development ofCities

industry. We'Jl also discuss empirical evidence concerning the effects of labor costs
and unions on location decisions.

The Semiconductor lndustry

The semiconductor industry illustrates sorne of the complexities of locational choices
in modern industry. The industry's output is light and compact relative to its value,
so transport costs are relatively unimportant in location decisions. What matters are
localization economies and access to different types of labor.
As explained by Castells ( 1988), the making of semiconductors involves several
distinct operations, requiring three different types of labor:
l. Research and development. Engineers and scientists design new circuits and
prepare the circuits for implantation into sil icon chips.
2. Wafer fabrication. Skilled technicians and manual workers make the chips
holding the circuits.
3. Assembly into components. Unskilled workers assemble the chips into elec-
tronic components.
Many semiconductor firms split their operations into three parts. Research and
development occurs in the Silicon Valley to exploit localization economies gener-
ated by the large concentration of semiconductor firms. The Silicon Valley is also
considered a desirable location by engineers and scientists. Advanced manufactur-
ing (wafer production) is typically located outside the Silicon Valley. For example,
National Semiconductor has manufacturing facilities in Utah, Arizona, and the state
of Washington; lntel has plants in Oregon, Arizona, and Texas; and Advanced Mi-
cro Devices has a plant in Texas. These facilities are located in areas that provide
(1) a plentiful supply of skilled manual laborers, (2) an environment attractive to
engineers and technicians, and (3) easy access, by air transportation, to the Silicon
Valley. The firm's assembly facilities are typically overseas, in locations such as
Southeast Asia that provide a plentiful supply of low-skilled workers.

Japanese Automobile Firms

A recent study of Japanese firms in the automobile industry (Smith and Florida,
1994) explores the relative importance of various locational attributes. Under the
just-in-time inventory system, a firm links the delivery of intermediate inputs with
its production schedule, the idea being that the inputs should arrive just in time for
production. This inventory system requires geographic proximity to input suppliers
to both save on shipping time and facilitate communication between firms. The study
suggests that the major factor in the location of Japanese-affiliated automotive sup-
pliers in the United States is proximity to Japanese-affiliated automotive assemblers.
In other words, agglomeration is the key feature in the location choices of such firms.
The study of Japanese firms provides sorne other insights into Iocational choices.
Japanese-affiliated automotive suppliers prefer locations with relatively large popu-
lations. a high density of manufacturing activity, an educated workforce, and better
Chapter 4 Where Do Firms Locate? 85

transportation. Furthermore, these firms tend to pick locations with relatively high
wages, reflecting their willingness to pay higher wages in exchange for higher labor
productivity. Finally, Japanese-affiliated manufacturers tend to locatc in areas with
relatively high concentrations of minority workcrs.

Mexican Garment Industry

Hanson (J 996) describes thc locational patterns of firms in the Mexican garment
industry. Until the l920s, most garments in Mcxico were made by housewives and
neighborhood tailors, who used factory-made cloth to produce custom garments for
family members and local patrons. The first garment factories, built by Lebanese and
Jewish immigrants, were located in downtown Mexico City. For severa! decades, the
industry was highly concentrated in the Federal District (the area containing Mexico
City). with about 60 percent of Mexico's garmentjobs there.
During the l 970s and l 980s, garment employment in the Federal District de-
creased in relative and absolute terms. Thc district's share of national garmenl
employmcnt dropped from 60 percent in 1965 to 33 percent by 1985, with jobs
moving to states with lower wages (Aguascalientes, Jalisco, Mexico, Nuevo Leon,
and Puebla). Each state specialized in a particular type of garment, and garment em-
ployment in each state is concentrated in onc or two cities. For example, the city of
Aguascalientes is the employment center for children 's outerware, with over 44 per-
cent of the national employment in that sector.
Although many garment jobs moved out of the Federal District, the district
retained its role as the ccnter for marketing and dcsign. In 1980, about 70 percent
of the wholesale trade in garments, textiles, and leather goods was conducted in thc
Federal District, suggesting that the bulk ofmarketing and design still happens there.
Most of the jobs that moved to the outlying areas were low-skill assembly jobs. In
other words, there is a hierarchy of functions, with the activitics subjecl to largc
agglomerative economies (marketing and design) concentrated in one location, and
activities with smaller agglomerative economies (assembly) dispersed to outlying
areas.

Carpet Manufacturing and Localization Economies

As an example of localization economies, consider DaJton, Georgia, the preemi-
nent carpct manufacturing ccnter of the Unitcd States (Krugman, 1991). In J 895,
Catherine Evans of Dalton used an outdatcd technique-known as tufting or
candlcwicking-to makc a bedspread as a wedding gift. The gift was a big hit,
and in the next few years, Ms. Evans made a few tufted items for her friends and
even sold a few. After she discovered a few production tricks such as a technique for
locking the tufts onto the backing, she and her neighbors Jaunched a local handicraft
industry, producing handmade tufted items for sale.
Immediately after World War Il, a machine for producing tuftcd carpets was
developed, making tuftcd carpel a cheaper alternative to woven carpet. New carpet
makers sprung up in the Dalton area because that's where they could find workers
86 Part 1 Market Forces in the Development of Cities

with knowledge and experience in tufting. Support firms located nearby, supplying
dyes, backing, and other intermediate inputs. Sorne of the old fi rms that had produced
woven carpets went out of business because they were underpriced by tufting firms,
while others moved from the Northeast to the Dalton area and switched to the new
technique. By the middle of the l 950s, Dalton was the carpet capital of the nation.
Of the top 20 carpet manufacturers in the United States, 6 are Located in Dalton, and
13 others are located nearby.

GM's Satum Plant

In J985, General Motors announced that its new Saturn plant would be located in
Spring Hill, Tennessee, a crossroads community about 30 miles from Nashville. The
plant, which cost severa! billions of dollars to build, employs about 3,000 workers
and supports thousands of jobs in the Nashville area. Why did GM choose Nashville
over the dozens of other communities that sought the plant?
A case study by Bartik, Becker, Lake, and Bush (1987) explored differences in
transportation, labor costs, and taxes for alternative locations. The first step in the
case study addressed the issue of transportation costs and market access. The authors
estimated the costs of shipping finished cars from seven alternative production si tes to
GM's markets in the continental United States. Based on the distribution of projected
sales in 2000, the location offering the best market access was Terre Haute, Indiana.
As shown in Table 4-3, the variation in transportation cost among the seven sites
was relatively small. The gap between Terre Haute and the site with the highest
transportation cost (Kalamazoo) was only $ 17 per car. Nashville was ranked fifth,
with transport costs only $13 per car higher than Terre Haute.
The second step of the study incorporated labor costs into the analysis. A labor
contract betwcen GM and its unionized workers stipulated that the Saturn workers
will be paid the same wage regardless of the plant's location, so GM would not
decrease its own wage bill by locating in Tennessee, a state with low wages. How-
ever, GM would purchase a large fraction of its inputs from nearby suppliers, who

TABLE 4-3 Case Study of GM's Saturn Plant

Estimated Cost (S per car)

Labor Costs for State and Local Taxes Total Measured

Transpon Costs Local Suppliers (before subsidies) Cost
Nashville. TN $426 $159 $11 8 $703
Leltington. KY 423 186 106 715
St. Louis. MO 419 172 134 725
Bloomington, lL 417 202 162 781
Kalamazoo, MI 430 244 116 790
Tcrrc Haute, IN 413 209 168 790
Marysville, OH 427 219 169 815
So11rre: Tímothy J. Banik. Charles Bcckcr. Steve Lake, and John Bush. ··Satum and Stntc Econo mic Oevelopmcn1:· Fornm
for Applied Researrh a11d Public Policy (Spring 1987). pp. 2~.
Chapter 4 Where Do finns Locate? 87

would pay low wages and pass on the savings to GM. As shown in Table 4-3, the
variation in labor cost was relatively large. The difference between the lowest-cost
site (Nashville) and the highest-cost site (Kalamazoo) was $85 per car.
The third step was to estimate taxes for the seven possiblc sitcs. Table 4-3
shows the tax cost per car in the absence of special subsidies and exemptions. The
difference between the site with the highest taxes (Marysville) and the site with the
lowest laxes (Lexington) was $63 per car. Nashville was ranked third, with tax costs
per car $12 higher than Lexington.
This case study suggests that in the absencc of special tax treatment, Nashville
had the lowest total cost. The most important factor was the lower labor costs in
Nashville. Although the city had a slight disadvantage to Tcrre Haute in terms of
market access, this disadvantage was more than offset by Nashville's lower labor
costs.
The state ofTennessee offered three types of inducements to GM. The first was a
$30 mili ion highway project that connected the plant with Interstate 65. The second
was a subsidized job-training program for Saturn workers, worth about $4 per car.
The third was a program of property tax subsidies, worth about $30 per car. As
shown by the numbers in Table 4-3, these subsidies would not have been necessary
in the absence of special subsidies from other states. Other states did offer special
subsidies, and Tennessee responded in kind to stay competitive in the bidding for
the plant.

Effects of Wages and Unions

How do wages affect firms location decisions? Carlton ( 1983) examines the re-
lationship between the number of new firms in a metropolitan area and several
variables, including local wages. He looks at firm births in three industries: plastics
products, electronic transmitting equipment, and electronic components. These in-
dustries have relatively low transport costs, so they are oriented toward local inputs,
not toward output markets or natural resourccs. His results suggest that for the elec-
tronic components industry the elasticity of the number of plant births with respect
to the metropolitan wage is - 1.07. ln other words, a 10 percent increase in the wage
decreases the number of births by 10.7 percent. Bartik (199 l) examines and sum-
marizes the results of dozens of studies and concludes that the long-run elasticity of
business activity with respect to the wage is between -1.0 and -2.0. This means
that a 1O percent decrease in the metropolitan wage will increase business activity
between l O percent and 20 percent.
How do unions affect location decisions and the volume of business activity?
One of the effects of unions is to increase wages, so a highly unionized metropolitan
area is likely to have relatively high wages. Since the elasticity of business activity
with respect to the wage is relatively large, to che extent that unions increase wages,
they decrease business activity. Unions may have other (nonwage) effects that influ-
ence business location decisions. For example, unions may affect labor productivity.
According to Bartik ( 1991 ), empirical studies of the nonwagc cffects of unions
have generated mixed results. Although most studies found that the presence of
88 Part 1 Market Forces in the Development of Cities

unions decreases business activity, many studies suggested that the negative effects
are rather small.

SUMMARY

l. Cities develop around the concentrations of employment generated by firms, so

the location choices of firms determine the location of cities and urban growth.
2. A resource-oriented firm has relatively high transport costs for its inputs, so it
locates near its input source.
3. A market-oriented firm has relatively high transport costs for its output, so it
locates close to its market.
4. A market-oriented firm with multiple inputs and outputs will Jocate at its median
transport location . The median location is often in large cities, providing one
reason for the growth of big cities.
S. Sorne firms are oriented toward local inputs. Energy-intensive firms are drawn
toward locations with cheap energy, while firms that benefit from agglomeration
economies are pulled toward industry clusters or large cities.
6. In the last severa! decades, the relative importance of transport costs has de-
creased, reflecting decreases in unit transport costs and the physical weight of
inputs.
7. Natural amenities attract workers, and the resulting lower wages attract firms .
8. Large cities provide learning opportunities that increase productivity and attract
firms, causing urban growth.

EXERCISES AND DISCUSSION QUESTIONS

l. Comment on the following from the owner of a successful plywood mili: "Firms
don 't use location theory to make location decisions. 1 chose this location for
my plywood mili because it is close to my favorite fishing spot."
2. Depict graphically the effects of the following changes on the bat firm's cost
curves (shown in Figure 4-2). Explain any changes in the optimum location.
a. The cost of shipping bats increases from $ 1 per ton to $4 per ton, while the
cost of shipping wood remains at $1 per ton .
b. The forest at point F burns down, forcing the firm to use wood from point
G which is 1O miles west of point F (20 miles from the market).
c. The firm starts producing bats with wood and cork, using three tons of wood
and two tons of cork to produce three tons of bats. Cork is ubiquitous (avail-
able at ali locations for the same price).
3. Why do breweries typically locate near their markets (far from their input
sources), while wineries typically locate near their input sources (far from their
markets)?
4. The building of wooden ships was a weight-losing activity, as evidenced by the
piles of scrap wood generated by shipbuilders. Yet shipbuilders located in ports,
far from their input sources (inland forests) . Why?
Chapter 4 Where Do Fim1s Locate? 89

S. Considera firrn that delivers video rentals to its customers. The spatial distribu-
tion of customers is as follows: 1O videos are delivered to location W, 1O miles
due west of the city ccnter; 50 videos are delivered to the ci ty center; 25 videos
are delivered to E, 1 mi le due east of the city center; and 45 videos are delivered
to point F, 2 rn!les east of the city center. Production costs are the same at ali
locations.
a. Using a graph, show where the firm should locate. Explain your location
choice.
b. Suppose that point W is in a valley and point F is at the top of a mountain.
Therefore, the unit cost of easterly transport (shipments from west to east)
is twice the unit cost of westerly transport. If production costs are the same
al ali Iocations, where should the firm locate? Explain.
6. Figure 4-4 shows the location choice of Ann 's pizza firm. Discuss the effecls
of the following changes on Ann's location choice.
a. A tripling of the distance between Y and Z (from 7 miles to 21 miles).
b. A tripling of the number of customers at point W. Instead of two customers
at W there are six.
c. Ann stops delivery service, forcing consumers to travel to the pizza parlor.
7. In Figure 4-6, the weight-losing firm is located at point P (the port). If the
monelary weight of location 8 is $27 instead of $15, will the firm still locale at
point P?
8. There is sorne evidence that people have become more sensitive to air pollulion.
In other words, people are willing to pay more for clean air. lf this is true, what
influence will it havc on the location decision of firms?
9. Consider a firm lhat uses one transferable input to produce one output. The
monetary weight of the output is $4, and the monetary weight of the input is $3.
The distance between M (the market) and F (the input source) is 1O miles.
a. Suppose that produclion costs are the same at ali locations. Using a diagram
like the one in Figure 4-2. explain where the firm will locate.
b. Suppose that the cost of land (a local input) increases as one approaches the
market. Specifically, suppose that the cosl of land is zero al F but increases
al a rate of $2 per mi le as the firm approaches M. Depict the location choice
of lhe firm graphically.
10. Chapter 2 discusses the incubator process. When industries mature. they move
from single-activity clu ters to areas with lower land and labor cosls. Explain
this process in terms of changes in the orientation of firms as they mature.
11. Suppose that country L has a plentiful supply of labor (and low wages) but a
relatively Iow supply of raw materials. In contrast, H has a plentiful supply of
raw materials, but a relatively low supply of labor (and high wages). The two
countries are separated by a mountain range that makes travel between the two
counlries prohibitively costly. Suppose that a weight-losing product is initially
produced in H (close to the supply of raw materials). Suppose that a tunnel is
bored through the mountain, decreasing the costs of shipping raw materials and
output between the two countries. Assume that labore rs do not migrate from
one country to the other.
90 Part 1 Market Forces in the Development of Cities

a. How will the tunnel affect the location choices of weight-losing firms?
b. How will the tunnel affect wages in the two countries?
c. How might this analysis be used to explain (1) the shift in manufactur-
ing from the United States to East Asían countries and (2) the narrow-
ing of the wage differential between the United States and East Asian
countries?

REFERENCES
1. Anderson, John E., and Robert W. Wassmer. "The Decísion to 'Bid for Business':
Municipal Behavior in Granting Property Tax Abatements." Regional Science
and Urban Economics 25 (1995), pp. 739-57.
2. Baade, Robert A., and Allen R. Sanderson. "The Employment Effects ofTeams
and Sports Facilities." Sports, Jobs and Taxes, eds. Roger Noll and Andrew
Zimbalist. Washington: Brookings, 1997.
3. Bartik, Timothy J. Who Benefitsfrom State and Local Economic Development
Policies? Kalamazoo, MI: W. E. Upjohn Institute, 1991.
4. Bartik, Timothy J.; Charles Becker; Steve Lake; and John Bush. "Saturn and
State Economic Development." Forumfor Applied Research and Public Policy,
Spring 1987, pp. 29-40.
5. Black, Duncan, and Yernon Henderson. "A Theory of Urban Growth." Journal
of Política! Economy 107 (1999), pp. 252-84.
6. Black, Duncan. "Local Knowledge Spillovers and Inequality." Working Paper,
University of Californ ia, Irvine (2000).
7. Boarnet, Marlon, and William T. Bogart. "Enterprise Zones and Employment:
Evidence from New Jersey." Journal of Urban Economics 40 (1996), pp. 198-
215.
8. Carlton, D.W. "The Location and Employment Choices of New Firms." Review
of Economics and Statistics 65 (1983), pp. 440-49.
9. Castells, Manuel. "The New Jndustrial Space: Information Technology Manu-
facturing and Spatial Structure in the United States." In America 's New Market
Geography, eds. George Sternlieb and James Hughes. New Brunswick, NJ:
Center for Urban Policy Research, 1988.
1O. Dowall, David. "An Evaluation ofCalifornia's Enterprise Zone Programs." Eco-
nomic Development Quarterly 1O( 1996), pp. 352-68.
11 . Eberts Randall W., and Joe A. Stone. Wage and Adjustment in Local Labor
Markets . Kalamazoo, MI: Upjohn Institute, 1992.
12. Ellison, Glen, and Edward Glaeser. ''The Geographic Concentration oflndustry:
Does Natural Advantage Explain Agglomeration?" American Economic Review
89 ( 1999), pp. 311-16.
13. Glaeser, Edward L., Jed Kolko, and Albert Saiz. "Consumer City." Journal of
Economic Geography 1 (200 1), pp. 27-50.
14. Graves, Phillip E. "A Life-Cycle Empírica! Analysis of Migration and Climate
by Race." Journal of Urban Economics 6 ( 1979), pp. 135-47.
Chapter 4 Where Do Firm~ Locate? 91

15. Greenwood, Michael. "Human Migration: Theory, Models, and Empirical Stud-
ics." Joumal of Regional Science 25 ( 1985), pp. 521-44.
16. Hanson, Gordon H. "Agglomeration, Dispcrsion, and the Pionccr Firm." Journaf
of Urban Economícs 39 ( 1996), pp. 255- 81.
17. Helms, L. Jay. "The Effect of State and Local Taxes on Economic Growth: A
Times Series-Cross Section Approach." Review of Economícs cmd Statistics 68
( 1985), pp. 574-82.
18. Hoover, Edgar M. Regional Economics. New York: Knoph, 1975.
19. Krugman, Paul. Geography and Trade. Cambridge, MA: MIT Press, 1991.
20. Levy, John M. Econo111ic Development Programs for Cítíes, Counties, and
Towns. New York: Praegcr Publishers, 1981.
21. Munnell, Alicia H. " How Does Public Jnfrastructure Affect Regional Economic
Performance?" New Engfand Economic Review, 1990, pp. 11 - 33.
22. Noll, Roger G., and Andrew Zimbalist. "Build the Stadium-Create the Jobs!"
Sports, Jobs and Taxes, cds. Roger Noll and Andrew Zimbalist. Washington:
Brookings, 1997.
23. Papke, Leslie. "Tax Policy and Urban Development: Evidence from the Indiana
Enterprise Zonc Program." Journal of Pubfíc Economícs 54 ( 1994), pp. 37-49.
24. Papke, Leslie. "What Do We Know About Enterprise Zones?" In Tax Policy and
the Economy, ed. James Poterba. Cambridge, MA: MIT Press, 1993.
25. Porell, Frank W. " Jntermetropolitan Migration and Quality of Life:· Joumaf of
Regional Science 22 ( 1982), pp. 137-58.
26. Rauch, J .E. "Productivity Gains from Gcographic Concentration of Human
Capital: Evidence from the Cities." Journaf of Urban Economics 34 ( 1993),
pp. 380-400.
27. Smith, Donald F., and Richard Florida. "Agglomeration and Industrial Location:
An Econometric Analysis of Japanese-Affiliated Manufacturing Establishments
in Automotive-Related Industries." Joumal of Urban Eco110111ics 36 (1994),
pp. 23-41.
28. Wassmer, Robert W. "Can Local Incentives Alter a Metropolitan City's Eco-
nom ic Developmcnt?" Urban Studíes 3 1 ( 1994), pp. 1251 - 78.