Full text of Review (Federal Reserve Bank of St. Louis) : September/October 1990

View original document

The full text on this page is automatically extracted from the file linked above and may contain errors and inconsistencies.

3 A ccou n tin g fo r Changes in M anu
factu red Exports at the State
Level: 1976-86
15 T h e Pitfalls o f Exchange Rate T a r
geting: A Case Study fro m the
United K ingdom
25 Do G overn m en t D eficits M atter?
40 W h at Do Econom ic M odels T ell Us
A bout the E ffects o f the U.S.Canada Free T ra d e A greem ent?
59 M on eta ry Policy on the 75th
A n n iversa ry o f the Federal
R eserve System: Sum m ary o f a
C on feren ce Proceedings

t h i;
FEDERAL
A RESERVE
I1VNK of
ST.I/M IS

Federal R eserve B ank o f St. Louis
R eview

September/October 1990

In T h is Issu e . . .

The disparity across states in the growth of manufactured exports
suggests that the internationalization of the U.S. economy has spread
unevenly. In the first article in this Review, "Accounting for Changes in
Manufactured Exports at the State Level: 1976-86,” Cletus C. Coughlin
and Thomas B. Mandelbaum explore several factors that may have pro
duced these differences in states’ export growth.
Using a technique called shift-share analysis, the authors determine
that, in most cases, the industrial composition of a state's exports is not
a major influence on its export growth. Rather, the relationship between
export growth in a state's individual industries and national export
growth in these industries causes state export growth to differ from the
national average. Since previous research has established that capital
abundance—especially human capital abundance—is the United States’
primary source of international comparative advantage, the link between
a state's 1976-86 export growth and its growth in physical and human
capital is also examined. When differences in the industrial mix of ex
ports are eliminated, Coughlin and Mandelbaum find a positive relation
ship between a state's export growth and the growth of its human capital.
*

In the second article of this issue, "The Pitfalls of Exchange Rate
Targeting: A Case Study from the United Kingdom,” Michael T. Belongia
and K. Alec Chrystal discuss the arguments, pro and con, surrounding
the use of the exchange rate as a target objective for monetary policy.
Many of the world’s central banks considered this strategy in the 1980s,
primarily because of breakdowns in the historical relationships between
monetary aggregates and economic activity and because of perceived
misalignments in exchange rates and trade balances. Using a welldefined episode of monetary policy in the United Kingdom as a case
study, Belongia and Chrystal argue that exchange rate targeting can
have severe adverse consequences if a central bank uses monetary
policy to offset a real exchange rate change. Specifically, resisting a real
exchange rate change while trying to maintain a nominal exchange rate
target can exacerbate recession or inflation. Examining the United
Kingdom in the late 1980s, they argue that the rapid acceleration of its
inflation rate is linked directly to the Bank of England's attempt to keep
the exchange rate at a level of three DM per pound rather than let it
appreciate to a higher value.
* * *
In the third article in this issue, Daniel L. Thornton considers the
question: "Do Government Deficits Matter?” Thornton discusses alter
native views of the effects of deficits on the economy, interest rates, the

SEPTEMBER/OCTOBER 1990

trade deficit and the like, pointing out the rationale behind the conven
tional views, for example, that government deficits cause real interest
rates to be high or the trade deficit to be large. He then discusses the
alternative view, called Ricardian Equivalence, that deficits don’t matter.
Finally, he presents some evidence on the relationship between govern
ment deficits and various economic variables using annual data for 16
OECD countries for the period 1975-86. By and large, his results, consis
tent with most U.S. time-series studies, support the Ricardian view.
* * *
Studies of the historic U.S.-Canada Free Trade Agreement have pro
duced conflicting estimates of its economic effects. In the fourth article
in this R eview , “What Do Economic Models Tell Us About the Effects of
the U.S.-Canada Free Trade Agreement?” Cletus C. Coughlin examines
five studies to better understand their estimates and determine why
they contradict each other. The conflicting results emerge both from
different assumptions about how certain markets operate and the values
of some parameters and from differences in the level of detail with
respect to commodities and countries.
As Coughlin stresses, several key aspects of the agreement are not in
cluded in these models because they are difficult to quantify. These
unmeasured aspects, the author says, may be more important than the
measured ones in terms of the economic effects of the agreement. Thus,
the message of this article, which is applicable to all economic models of
trade policy changes, is "let the user beware.”
*

In the last article of this issue, Michael T. Belongia summarizes the
proceedings of a conference held at the Federal Reserve Bank of St.
Louis in honor of the Federal Reserve System's 75th anniversary. The
papers at this conference reviewed the historical performance of
monetary policy in the United States, why the Federal Reserve might
choose the federal funds rate as its operating target, whether the
Federal Reserve should be constrained by a monetary rule with a target
of price stability, how the money stock should be measured and
whether monetary actions have significant effects on real output. A
complete conference program and instructions on how to obtain the
proceedings also are included.
*

FEDERAL RESERVE BANK OF ST. LOUIS

Cletus C. Coughlin and
Thomas B. Mandelbaum

Cletus C Coughlin is a research officer and Thomas B.
.
Mandelbaum is an economist at the Federal Reserve Bank of
St. Louis. Thomas A. Poilmann provided research assistance.

Accounting for Changes in
Manufactured Exports at the
State Level: 1976-86

£ HE RAPID INTERNATIONALIZATION of the
U.S. economy in recent years has spread un
evenly across regions and states. For example,
while the real value of direct manufactured ex
ports rose 25 percent in the nation between
1976 and 1986, it actually declined slightly in
the Middle Atlantic and Upper Midwest regions
of the United States.1 Coughlin and Fabel (1988)
have demonstrated that some of the variation in
state export levels can be explained by differ
ences in their endowments of productive re
sources. According to these authors, states with
relatively more capital, both human and physical,
have higher export levels and, thus, higher
shares of U.S. exports.

varied growth in states’ export sectors. Using a
technique called shift-share analysis, we isolate
two influences on state export growth—industrial
composition and relative export growth of the
same industry at the state and national levels—
and compare their relative importance. Next, we
examine the relationship between export growth
and resource endowments at the state level to
see if we find results consistent with those of
Coughlin and Fabel.

This paper extends the research in Coughlin
and Fabel by examining the change in export
levels across states between 1976 and 1986. We
explore the factors that potentially caused the

Table 1 contains the basic export and resource
endowment data used in this study for the 48
states in the continental United States.2 A look
at the export data columns shows the tremen-

1This comparision uses U.S. Census regions. The Middle
Atlantic Census Region consists of New York, New Jersey
and Pennsylvania; the Upper Midwest is actually the East
North Central Census Region, which consists of Ohio, In
diana, Illinois, Michigan and Wisconsin. The value of direct
manufactured exports is the plant value of manufactured
exports (U.S. Department of Commerce, 1981 and 1989).
In 1986, the port value of U.S. manufactured exports was

FACTORS INFLUENCING THE
SHIFT OF EXPORTS AMONG
STATES

$182 billion; the plant value of exports, $159.4 billion, is
obtained by removing transportation and insurance costs.
2Our analysis excludes the District of Columbia, Alaska and
Hawaii because their export values are small and their ex
ports are not disaggregated by industry, a deficiency that
precludes meaningful interpretation of the shift-share
analysis that we present later.

SEPTEMBER/OCTOBER 1990

Table 1
Direct Manufactured Exports, Human Capital and Physical Capital_________
Direct Exports
1986 level
($ millions)

Alabama
Arizona
Arkansas
California
Colorado
Connecticut
Delaware
Florida
Georgia
Idaho
Illinois
Indiana
Iowa
Kansas
Kentucky
Louisiana
Maine
Maryland
Massachusetts
Michigan
Minnesota
Mississippi
Missouri
Montana
Nebraska
Nevada
New Hampshire
New Jersey
New Mexico
New York
North Carolina
North Dakota
Ohio
Oklahoma
Oregon
Pennsylvania
Rhode Island
South Carolina
South Dakota
Tennessee
Texas
Utah
Vermont
Virginia
Washington
West Virginia
Wisconsin
Wyoming

$ 1684.9
1755.8
1065.4
17216.4
1477.7
3996.4
429.5
3372.6
2826.7
502.6
7209.2
4787.4
1932.4
1835.0
1939.8
3020.3
800.6
1740.5
5513.8
10878.0
3691.9
1337.1
4267.9
101.2
753.3
167.1
892.6
3548.1
177.7
9412.4
5260.8
214.7
10653.0
1084.6
1862.7
6026.6
481.9
2398.0
212.7
2910.4
10981.5
668.5
384.0
2704.0
9862.8
983.2
3313.5
19.1

FEDERAL RESERVE BANK OF ST. LOUIS

Percent
change
1976-86

102.5%
174.7
63.6
113.3
139.8
104.1
128.0
147.5
107.2
197.6
8.3
69.3
28.8
188.9
70.6
118.3
214.1
171.7
120.3
57.9
135.6
91.6
163.1
132.1
143.6
514.3
206.6
33.4
156.8
76.9
138.9
154.1
83.9
87.4
126.0
28.1
79.3
156.4
211.4
132.3
111.2
199.0
92.2
75.1
204.9
119.9
50.0
85.4

Human Capital
1986 level
($1000/worker)

$ 91.4
97.5
63.7
135.5
117.6
122.0
195.1
98.5
74.5
82.0
123.8
131.2
100.1
88.7
102.1
125.3
88.0
128.6
115.7
164.7
116.2
69.2
107.3
31.0
82.0
100.4
86.2
127.9
88.2
134.8
66.9
68.6
131.1
111.4
90.8
107.3
89.8
80.5
73.2
87.9
121.5
102.1
90.5
80.9
141.6
155.2
100.1
56.9

Percent
change
1976-86

140.4%
114.2
380.2
173.2
128.4
138.9
133.2
124.1
177.5
118.9
158.6
134.2
206.9
137.2
423.8
164.5
138.7
143.8
131.4
200.5
175.3
197.1
112.7
-30.9
146.0
275.6
98.1
101.9
704.9
162.4
175.1
234.2
147.9
166.7
182.8
130.0
109.6
464.6
261.0
154.9
193.6
240.1
72.8
166.6
166.7
215.1
231.8
45.9

Physical Capital
1986 level
($ 1000/worker)

$ 56.3
40.0
41.1
36.3
43.2
30.3
48.4
38.6
40.2
50.3
44.3
57.9
54.8
41.8
51.3
167.4
60.1
45.7
30.4
47.4
35.6
46.5
35.7
71.4
35.9
43.9
31.6
34.5
57.0
33.7
37.6
55.7
45.6
52.1
43.7
42.1
22.6
53.6
30.0
40.5
81.5
39.1
55.7
43.7
55.0
68.1
39.9
102.1

Percent
change
1976-86

130.2%
95.4
148.2
129.6
118.2
113.5
78.1
100.5
136.4
136.3
125.6
119.8
159.4
136.7
150.6
204.5
188.5
102.1
146.7
122.7
152.2
191.0
146.5
67.9
108.0
94.7
136.6
89.8
158.3
131.3
127.9
170.1
108.9
181.3
102.9
113.1
106.6
159.5
142.3
121.1
150.6
110.8
259.6
133.6
99.9
85.2
143.0
114.4

dous diversity across these states in both the
level of exports and their growth over the de
cade. In 1986, for example, the value of direct
manufactured exports ranged from $19.1 million
in Wyoming to $17.2 billion in California. While
direct exports rose 91.2 percent in the continen
tal United States between 1976 and 1986, state
growth rates ranged from 8.3 percent in Illinois
to 514.3 percent in Nevada. States with relative
ly high growth rates captured larger shares of
U.S. exports over time, while those with lower
growth rates saw their export shares diminish.
These changes in export shares can be ex
amined by a technique called shift-share analysis.
This method, an accounting technique, is de
scribed in detail in the appendix. Basically, the
technique calculates each state’s net relative
change over the period; states in which exports
grew more rapidly than the national average
between 1976 and 1986 have a positive net rela
tive change and vice versa. The figures in the
first column in table 2 show these net relative
changes in exports across states. These changes
and their individual components (also in table 2)
are expressed as percentages of the export levels
that would have been achieved in 1986 had
their exports grown from 1976 to 1986 at the
national rate. Thus, for example, Arizona’s ex
ports in 1986 were 36 percent higher than if its
exports had grown at the national rate from
1976 to 1986.
The shift-share method divides a state’s net
relative change (NRC) in exports among three
components: the industrial mix effect (IME), the
competitive effect (CE) and the allocative effect
(AE). Each state's IME, CE and AE sum to its
NRC.

The Industry Mix Effect
During any period, exports of some of the na
tion’s goods will grow faster than others. Those
states whose exports are more heavily concen
trated in these faster-growing export sectors
will find their share of the nation’s total exports
rising, other things the same. The opposite rela
tionship holds true as well: states whose exports
are more heavily concentrated in goods whose
export sales are growing relatively slowly at the
national level will find their share of the nation’s
export sales declining. Discussions of regional

export growth frequently focus on the region’s
industrial mix as a key determinant of its export
performance. For example, Hervey (1986) attrib
uted the Midwest’s slow export growth through
out most of the 1970s and early 1980s to its
"traditional” industry composition.3
Table 3 shows 1976-86 annual growth rates of
U.S. exports from the 20 major industry groups.
The industries are listed in declining order of
their export growth rates over the 10-year
period. The last column in table 3 shows the
composition of U.S exports in 1976. If the com
position of a state’s exports was identical to that
of the nation's exports, its IME would equal
zero. If a state had a favorable (unfavorable)
mix of exports, that is, if it had high (low) con
centrations of its 1976 exports in industries ex
periencing rapid national export growth over
the 1976-86 decade, its IME would be greater
(less) than zero. The magnitude of IME indicates
how much higher or lower the state’s exports
were in 1986 than they would have been if the
state’s export composition were identical to the
nation’s. This value is expressed as a percentage
of the level of 1986 total state exports that
would have resulted if they had grown at the
national rate in the 1976-86 period.
The IME values listed in table 2 range from
99 percent for North Dakota to -29 percent for
Nevada. Thus, the industrial mix effect, ceteris
paribu s, contributed to a 99 percent increase in
North Dakota’s exports relative to what they
would have been otherwise, while contributing
to a 29 percent reduction in exports in Nevada.

The Competitive Effect
The CE figures listed in table 2 indicate the
influence of the relative export growth of a
state’s industries, assuming its industry mix of
exports is identical to the nation’s. A positive
(negative) CE indicates how much higher (lower)
a state’s exports were in percentage terms in
1986 solely because exports from individual
state industries grew at a faster (slower) rate
than the corresponding national industries over
the 1976-86 period. This value is expressed as a
percentage of total state exports that would
have been achieved in 1986 had they grown at
the national rate over the 1976-86 period.

3More recently, Smith (1990) concluded that a region’s in
dustrial mix was an important factor in distinguishing its
relative export performance during 1987 and 1988.

SEPTEMBER/OCTOBER 1990

Table 2
Shift-Share Components for State ---------------- Growth, 1976-86
Export ----— -- ------ 7
Net
Relative Change

20.7%
36.1
-10.6
14.0
23.4
7.7
-7 .2
31.5
4.6
57.3
-42.4
-9 .7
-30.4
56.9
-10.0
19.1
86.0
34.7
18.7
-15.9
27.5
-1 .4
40.6
50.2
29.6
209.0
64.8
-2 8 .6
36.9
-5 .4
26.8
161.5
-2 .7
0.6
21.9
-3 3.3
-2 .3
37.6
64.2
20.9
12.5
52.5
37.7
-6 .7
61.4
23.6
-20.2
7.8

Industry
Mix Effect

Competitive
Effect

15.8%
3.8
0.2
6.0
5.1
7.0
13.3
4.2
0.5
13.8
-2 .6
3.8
-2 .7
10.9
3.3
13.6
38.0
-3 .8
4.3
3.8
-2 .0
0.5
7.1
76.5
0.5
-29.0
0.3
8.9
-15.9
0.4
-3 .2
99.0
0.6
-3 .9
-6 .9
-4 .5
-12.5
-2 .4
31.6
-1 .5
6.6
-8 .7
21.8
3.1
0.9
5.5
-2 .8
7.6

30.0%
36.3
92.6
11.2
50.0
18.2
249.2
61.3
16.1
83.0
-27.6
5.0
-12.8
53.4
-11.4
8.8
181.0
206.3
24.8
-9 .3
11.4
7.4
20.0
-7 .8
46.0
1902.7
45.8
-34 .4
800.7
0.2
80.8
104.6
4.4
17.0
131.3
-22.2
-6 .5
59.7
287.4
46.3
6.9
433.0
270.7
24.2
305.6
-1 .0
-9 .8
67.5

NOTE: See appendix for definitions of components.

FEDERAL RESERVE BANK OF ST. LOUIS

Allocative
Effect

-25.1%
-4 .0
-103.4
-3 .3
-31.7
-17 .4
-269.7
-34.0
-12.0
-39.4
-12.1
-18.5
-15.0
-7 .4
-1 .9
-3 .3
-133.0
-167.7
-10.4
-10.3
18.1
-9 .3
13.5
-18.5
-16.9
-1664.7
18.8
-3.1
-747.9
-6 .0
-50.8
-42.1
-7 .6
-12 .5
-102.5
-6 .6
16.7
-19.7
-254.7
-2 3.9
-1 .0
-371.8
-254.8
-34.0
-245.0
19.0
-7 .5
-67 .3

Table 3
Growth and Composition of U.S. Exports
SIC

29
25
30
38
28
37
27
31
21
36
23
20
26
32
35
22
24
34
39
33

Industry Group

Compounded annual
growth rate,
1976-86

Petroleum and coal products
Furniture and fixtures
Rubber and miscellaneous plastics
Instruments and related products
Chemicals and allied products
Transportation equipment
Printing and publishing
Leather and leather products
Tobacco products
Electrical equipment
Apparel and textile products
Food and kindred products
Paper and allied products
Stone, clay and glass products
Machinery, except electrical
Textile mill products
Lumber and wood products
Fabricated metal products
Miscellaneous manufacturing
Primary metal industries

10.5%
10.3
8.8
8.5
8.5
8.1
7.4
7.3
7.1
7.1
6.7
6.4
6.0
5.6
5.5
3.8
3.4
3.4
2.1
1.5

Total Exports

6.7

Regardless of its export composition; a state’s
overall exports could grow more rapidly than
the national average if its individual sectors suf
ficiently outpaced the national industry average.
In other words, a state can experience rapid ex
port growth not only by exporting those goods
that grew rapidly at the national level, but also
by relatively rapid growth of exports from in
dustries displaying little national export growth.
South Carolina’s pattern of export growth ex
emplifies this possibility. The state, as reflected
in its negative IME, has an unfavorable mix of
exports. This mix is characterized by a relative
ly large export share in the textile mill products
sector, whose exports had grown slowly nation
ally, and low concentrations in the chemicals
and transportation equipment sectors, among
the more rapidly growing export sectors nation
ally. Despite this industrial mix, exports from
South Carolina grew faster than the national
average because, as the positive CE shows, it
had relatively rapid export growth in individual

4According to Esteban-Marquillas (1972), p. 252, the
allocative effect “ will show us if the region is specialized
in those sectors in which it enjoys better competitive ad

Percent of
total
1976 exports

1.4%
0.2
1.5
4.5
11.2
20.0
0.7
0.3
1.2
11.1
0.9
6.9
2.7
1.1
23.0
1.5
2.2
4.5
1.6
3.6
100.0

sectors. Exports of South Carolina’s textile mill
products, for example, grew at a 6.2 percent
annual rate between 1976 and 1986; at the na
tional level, in contrast, they grew at a relative
ly slow 3.8 percent rate.

The Allocation Effect
The allocation effect reflects differences be
tween a state and the nation in both industrial
mix and relative industry export growth. Unfor
tunately, unlike the IME and CE terms, there is
no clear-cut interpretation of the AE.4 In 43 of
the states, the AE component was negative. For
most states, then, those sectors for which 1976
exports accounted for a small share of total
state exports relative to the national export com
position tended to grow more rapidly than at
the national level between 1976 and 1986. In
addition, those sectors that were relatively large
in 1976 grew more slowly.
Returning to the South Carolina example, one
reason that the state's AE was negative stems

vantages” as evidenced by faster-than-national growth.
Since our analysis is restricted to exports rather than pro
duction, this terminology is inappropriate.

SEPTEMBER/OCTOBER 1990

from its transportation sector. Exports of
transportation equipment accounted for less
than 1 percent of the state’s exports in 1976
compared with 20 percent nationally, while the
state’s 1976-86 annual growth rate of transpor
tation equipment exports was approximately
double the national rate. This combination of
small relative size and rapid growth contributed
to the state’s negative AE.

The Relative Influence o f IME, CE
and AE
To evaluate the contribution of industry mix,
industry growth and allocation effect for each
state, each component was ranked by its impor
tance in influencing the state’s net relative
change. Using the figures from table 2, the com
ponent with the smallest absolute value for each
state—and thus the state’s least important
factor—was ranked 1, while the state’s largest
component in absolute value—its most influen
tial component—was ranked 3. The results of
this exercise, shown in table 4, clearly indicate
that the IME was least important for most
states, while the CE was most important.5 The
IME was ranked as the least important compo
nent in 37 states, while the CE was ranked as
most important in 34 states.
The relative influence of each of the three
components also can be evaluated by comparing
each component’s percent share of the sum of
the absolute values of the three components. In
California, for example, the IME value of 6.0
represents 29.3 percent of the total effect
[{6.0/(6.0 + 11.2 + 3.3)}xl00= 29.3]. As table 4
shows, on average, IME, CE and AE account for
12.6 percent, 49.7 percent and 37.7 percent,
respectively, of the total influence on NRC.
Correlations between NRC and each of the
three components reinforce the notion that a
state’s CE is the primary influence on NRC. The
simple correlations across states between NRC
and the IME, CE and AE components were .32,
.68 and -.62, respectively. While all three coef
ficients are significantly different than zero at
the 0.5 percent level, the NRC-CE correlation is
substantially larger than the NRC-IME relation
ship, and, unlike the NRC-AE correlation, in
dicates a positive relationship.

5This result corroborates Bauer and Eberts’ (1990) finding
that a state’s growth rate of exports between 1980 and
FEDERAL
RESERVE BANK OF ST. LOUIS

Table 4
Relative Importance of Shift-Share
Components
Number of
states in which
component was
ranked
1
2
3

Industry mix effect
Competitive effect
Allocative effect

37
4
7

9 2
10 34
29 12

Average
percentage share
of total effect

12.6%
49.7
37.7

NOTE: A rank of 1 indicates a component was the smallest
in absolute value of the three components for a state,
while a 3 rank indicates it had the largest absolute
value.
The strongly negative NRC-AE relationship
suggests that, in general, those states with
faster-than-national export growth managed this
growth despite having relatively small shares of
their 1976 exports in industries in which the
state’s exports subsequently outgrew the na
tional industry average. Rather, their rapid ex
port growth was the result of faster-thannational growth of individual industries, even
though the rapid growth from these industries
tended to account for a relatively small share of
their 1976 exports. While states with rapid ex
port growth tended to have favorable industry
mixes, this factor is less important than the
relatively fast growth of state exports from
these industries.
In summary, across all states, the IME appears
to be relatively unimportant in determining
whether a state’s exports grew faster than the
national average. For the most part, it is the
relative export growth of a state’s individual in
dustries that determines whether the state’s ex
port performance is superior to the nation’s.

W H AT ARE THE SOURCES OF A
STATE’S COMPARATIVE EXPORT
ADVANTAGE?
To explain the relative export performance of
states, Coughlin and Fabel applied the Heckscher-

1986 cannot be explained by the mix of industries in the
state.

Ohlin theory of international trade. The
Heckscher-Ohlin approach highlights the impor
tance of a country’s productive resources in
determining its pattern of trade. One reason for
international trade is differences in production
costs across countries. These differences depend
on what proportions various factors of produc
tion exist in different countries (that is, the
relative factor endowments) and how the fac
tors are combined in producing different goods
(that is, the relative factor intensities).
Assuming a world consisting of two factors of
production, two goods and two countries, the
essence of the Heckscher-Ohlin theory can be
explained simply. In a two-factor world, a coun
try is relatively capital-abundant (labor-abundant)
if it has a higher (lower) ratio of capital to labor
than the other country. In a two-good world, a
product is capital-intensive if its production re
quires a higher ratio of capital to labor than the
other good. The Heckscher-Ohlin theory predicts
that a country will export the good that uses its
abundant factor intensively, while importing the
other good. For example, if the United States is
relatively capital-abundant and Mexico is
relatively labor-abundant, the United States will
export capital-intensive products and import
labor-intensive goods, while Mexico would do
just the opposite. The reason for this trade pat
tern hinges on the relative production costs. A
country should be the lower-cost producer of
goods that use relatively larger amounts of its
more abundant resource.
The Heckscher-Ohlin approach allows for
predictions about trade patterns based on a
knowledge of countries' factor supplies. Since
the services of factors of production are em
bodied in exports and imports, international
trade may be viewed as the export of the ser
vices of the country’s relatively abundant factor
in exchange for the services associated with its
scarce factor.
The preceding idea can be applied to states
within a country.6 Relative state export perfor
mance depends on state advantages; however,
the specific advantages must also be defined in

6Neither Coughlin and Fabel (1988) nor the present study
are tests of the Heckscher-Ohlin theory of trade. See
Bowen et al. (1987) for a rigorous examination. In the pre
sent case, the Heckscher-Ohlin theory provides a wellknown framework in which to analyze the factors that con
tribute to a state’s relative export performance.

the context of the world economy. For example,
if a state is relatively well-endowed with a re
source that is scarce in the United States relative
to other countries, then its resource advantage
will not necessarily translate into superior ex
port performance. Rather, the resource may
simply allow increased production of an importcompeting good. States that are better endowed
with the characteristics that are associated with
comparative advantage at the national level,
however, should display relatively better export
performance.
Numerous empirical studies suggest that the
United States’ primary source of international
comparative advantage is its abundance of hu
man capital.7 In addition, as Coughlin and Fabel
found, physical capital is a significant determi
nant of relative state export performance. To
further explain the interstate differences in ex
port growth rates, we examine the link between
states’ export growth and their changing en
dowments of physical and human capital.

The Relationship between Changes
in State Exports and Endowments
The connection between state export growth
and endowment changes is explored by testing
whether there is a statistically significant rela
tionship across states between measures of ex
port growth for the 1976-86 period and the per
cent change in human and physical capital per
manufacturing worker for the same period.
Two measures of export growth are used in
the statistical analysis: a state’s NRC and its CE.
Over any given period, a state’s export growth
relative to the nation (expressed by its NRC) is
influenced by both the export growth of its in
dividual industries (m easured by CE) and the
state’s industrial mix at the beginning of the
period. While a state’s human or physical capital
growth might be expected to stimulate the ex
port growth of its individual industries (and,
thus, increase its CE), there is no reason to
think that the state’s capital growth would be
linked to the industry mix of its exports at the
beginning of any period. Thus, a state’s capital
growth should be more closely linked to its CE

7Keesing (1966), Balassa (1979), Branson and Monoyios
(1977) and Stern and Maskus (1981) are a few of the
studies that have emphasized the impact of human capital
on U.S. international trade.

SEPTEMBER/OCTOBER 1990

than to its NRC; by definition, the former
measure is purged of the irrelevant and possibly
confounding effects of a state’s industrial mix
that is included in the latter measure.
For our analysis, a state’s human capital per
manufacturing worker is measured using the
difference between the state’s average wage for
manufacturing workers and the average wage
of unskilled manufacturing workers in the
state.® This difference, which is assumed to per
sist indefinitely, is viewed entirely as a return
to human capital. This flow of returns is con
verted to a stock of human capital by dividing
by an interest (discount) rate. Physical capital
per manufacturing worker is measured by de
preciable assets per manufacturing employee in
the state.9
Table 1 shows the 1986 levels of the capital
measures and their percent change since 1976.
Montana has the dubious distinction of having
the slowest growth in both human and physical
capital. The change in human capital ranges
from -30.9 percent in that state to 464.6 per
cent in South Carolina with a mean of 182.7
percent. The change in physical capital ranges
from 67.9 percent in Montana to 259.6 percent
in Vermont with a mean of 132.3 percent.
The relationships between state-level changes
in exports and endowments were explored by
first regressing NRC, and then CE, against the
percent change in human and physical capital in
a cross-sectional framework. The regression
analysis shows whether variations across states
in human or physical capital are closely linked
to variations in CE or NRC among states.
The results of this analysis are shown in table
5.1 Overall, neither the changes in human capital
0
nor those in physical capital explain, in a statisti
cal sense, differences in net relative change

8Following Hufbauer (1970), this method of calculating
human capital has been used frequently in international
trade studies. Average manufacturing wages for 1976 and
1986 are from the U.S. Department of Commerce (1981
and 1988). Unskilled manufacturing wages were from the
Current Population Survey-BLS Microdata File. A 10 per
cent discount rate was used for all states. This value af
fects the levels of human capital per worker, but does not
affect the statistical results.
9Data for depreciable assets are from the Annual Survey of
Manufactures. Data represent the gross book value of
depreciable assets at year’s end, 1975 and 1985.
10Nevada was excluded from the reported regressions
because an examination of the residuals indicated that it
was an outlier.
FEDERAL
RESERVE BANK OF ST. LOUIS

across states. We do, however, find that changes
in human capital endowments explain differences
in the competitive effect across states.
Specifically, we find that, ceteris paribu s, states
with larger increases in human capital endow
ments per manufacturing employee had larger
values for their competitive effect. Changes in
physical capital endowments, however, do not
explain differences in the competitive effect.
The difference in explanatory power of
human capital between the two regressions is
not surprising. A state’s relative export growth
is affected by a variety of factors besides
changes in resource endowments. A list of
reasonable determinants includes resource
changes in the rest of the world, demand
changes in both the United States and the rest
of the world and promotional expenditures by
state governments.1
1
By focusing on the competitive effect, some of
the potentially confounding effects associated
with a state’s industry mix are eliminated. For
example, foreign demand shifts toward certain
industries would result in rapid export growth
(and large positive NRCs) in states that happened
to have relatively large export concentrations in
those industries. Conversely, in states that had
relatively small shares in the rapidly growing in
dustries, we might find NRCs that are negative
even though many of their industries may have
experienced faster-than-national export growth.

CONCLUSION
A shift-share analysis reveals that the differing
growth of state exports relative to the national
average was due primarily to the "competitive
effect,” that is, faster-than-national or slowerthan-national export growth in individual indus-

"Evidence is presented in Coughlin and Cartwright (1987)
and Coughlin (1988) that export promotion expenditures by
state governments alter the export performance of states.
We also recognize that the rest of the world does not con
sist of identical countries, a fact that creates numerous
empirical issues. States export their products to different
mixes of foreign countries. Thus, each state’s exports are
affected by specific foreign supply and demand changes
to varying degrees. Primarily because of the volatility of
exchange rates in the 1980s, the different regional effects
of exchange rate changes is a topic that has received in
creasing attention. See Cox and Hill (1988) and Carlino et
al. (1990) for attempts to identify the differential output ef
fects across states of exchange rate changes.

1
1

Table 5
State Export Growth and Change in Endowments1
Dependent
Variable

Independent Variables
Constant

Net Relative
Change

-1.41
(-0.07)

Competitive Effect

-36.29
(-0.49)

Human Capital

Physical Capital

-0.01
(-0.22)

0.18
(1.14)

0.69*
(4.07)

-0.06
(-0.11)

R2
-0.02
0.25

NOTE: The value of the t-statistics are in parentheses. An asterisk denotes significance at the 5 per
cent significance level using a two sided hypothesis test. R is the adjusted coefficient of
determination.
1Human capital is the percent change of a state’s human capital per worker between 1976 and
1986, while physical capital is the percent change of a state’s physical capital per worker
between 1976 and 1986. The dependent variables are measures of export growth. See the ap
pendix for additional details.
tries in the state. The industrial composition of
exports in a state and the concentration of a
state's exports in industries that grew relatively
rapidly in the state were both found to be less
influential in determining why a state’s export
growth diverged from that of the nation. Thus,
our analysis suggests that a state’s industrial
structure does not always provide useful infor
mation in accounting for its export growth.
Since previous research has established that
capital abundance—in particular, human capital—
is the United States’ primary source of interna
tional comparative advantage, the link between
a state’s 1976-86 export growth and its change
in physical and human capital abundance was
examined. No link was found between a state’s
export growth relative to the nation (that is, its
net relative change) and the growth of either its
human or physical capital. When differences in
industrial mix among states were eliminated,
however, a positive association was found be
tween a state’s export growth and the growth
of its human capital. In other words, a state’s
competitive effect was related to its human
capital growth.

REFERENCES
Balassa, Bela. “ The Changing Pattern of Comparative Ad
vantage in Manufactured Goods,” Review of Economics and
Statistics (May 1979), pp. 259-66.
Bowen, Harry P., Edward E. Learner, and Leo Sveikauskas.
“ Multicountry, Multifactor Tests of the Factor Abundance
Theory,” American Economic Review (December 1987), pp.
791-809.

Branson, William H., and Nikolaos Monoyios. “ Factor Inputs
in U.S. Trade,” Journal of International Economics (May
1977), pp. 111-31.
Bauer, Paul W., and Randall W. Eberts. “ Exports and
Regional Economic Restructuring,” Regional Science
Perspectives (No. 1, 1990), pp. 39-53.
Carlino, Gerald, Brian Cody, and Richard Voith. “ Regional
Impacts of Exchange Rate Movements,” Regional Science
Perspectives (No. 1, 1990), pp. 89-102.
Coughlin, Cletus C. “ The Competitive Nature of State
Spending on the Promotion of Manufacturing Exports,” this
Review (May/June 1988), pp. 34-42.
Coughlin, Cletus C., and Phillip A. Cartwright. “An Examina
tion of State Foreign Export Promotion and Manufacturing
Exports,” Journal of Regional Science (August 1987),
pp. 439-49.
Coughlin, Cletus C., and Oliver Fabel. “ State Factor En
dowments and Exports: An Alternative to Cross-Industry
Studies,” Review of Economics and Statistics (November
1988), pp. 696-701.
Cox, W. Michael, and John K. Hill. “ Effects of the Lower
Dollar on U.S. Manufacturing: Industry and State Com
parisons,” Federal Reserve Bank of Dallas Economic
Review (March 1988), pp. 1-9.
Esteban-Marquillas, J.M. “A Reinterpretation of Shift-Share
Analysis,” Regional and Urban Economics (No. 3, 1972), pp.
249-61.
Hervey, Jack L. “ Midwest Exports Decline,” International Let
ter, Federal Reserve Bank of Chicago (April 1986).
Hufbauer, G. C. “ The Impact of National Characteristics and
Technology on the Commodity Composition of Trade in
Manufactured Goods,” in Raymond Vernon, ed., The
Technology Factor in International Trade (National Bureau of
Economic Research, 1970), pp. 145-231.
Keesing, Donald B. “ Labor Skills and Comparative Advan
tage,” American Economic Review (May 1966), pp. 249-58.
Kochanowski, Paul, Wayne Bartholomew, and Paul Joray.
“The Shift-Share Methodology: Deficiencies and Proposed
Remedies,” Regional Science Perspectives (No. 1, 1989),
pp. 65-88.

SEPTEMBER/OCTOBER 1990

Smith, Tim R. "Regional Export Growth: Lessons from the
State-Level Foreign Trade Data,” Regional Science Perspec
tives (No. 1, 1990), pp. 21-38.
Stern, Robert M., and Keith E. Maskus. “ Determinants ot the
Structure of U.S. Foreign Trade, 1958-76,” Journal of Interna
tional Economics (May 1981), pp. 207-24.

U.S. Department of Commerce, Bureau of the Census. 1977
Census of Manufactures, (GPO, September 1981), pp. 1-15
through 1-27.
________ 1986 Annual Survey of Manufactures: Geographic
Area Statistics (GPO, July 1988), pp. 3-5 through 3-11.
________Exports from Manufacturing Establishments, 1985
and 1986 (GPO, January 1989).

Appendix
Using Shift-Share Analysis To Analyze
State Export Growth
Two important factors that determine
whether a state's foreign exports grew at a dif
ferent rate than the national average over a
given period are the state’s industrial mix of ex
ports compared with the national mix (the in
dustrial mix effect) and the differential growth
rate of exports from individual state industries
relative to their national counterparts (the com
petitive effect). Shift-share analysis enables these
two factors to be separated and evaluated. The
Esteban-Marquillas (1972) shift-share model
makes the competitive effect completely in
dependent of industry mix by calculating a
third factor, called the allocative effect, which
accounts separately for the covariance between
the industry mix and the competitive effect
(Kochanowski, et al. 1989).1
Let Sj, and Sin denote proportions of total
direct exports represented by the ith industry
in state s and the nation, n, respectively; Gs and
G„ are the 1976-86 growth rates of total exports
in s and n, respectively; Gis and Gin the 1976-86
growth rates of exports in the ith industry in s
and n, respectively; and Es the 1976 level of
direct exports in state s.2
For the 1976-86 study period, the difference
between the state's actual 1986 exports and
what its 1986 exports would have been if state
exports had grown at the national rate between
1976 and 1986 is called the Net Relative Change
(NRC). In symbols,
(1) NRC, = ES ,- ES
G
G„.

'Esteban-Marquillas (1972) and Kochanowski, et al. (1989)
show that the traditional shift-share model fails to isolate
the competitive and industry mix effects.
2Direct export data for some industries in some states were
not disclosed by the U.S. Department of Commerce to en
FEDERAL RESERVE BANK OF ST. LOUIS

This is equivalent to:
(1’) NRC, = I (E,SuGis - E.Si.GJ,
where the summation in this equation, as well
as those in the following equations, are over all
manufacturing industries. Exports were not
reported for some industries for one of the re
quired years, so the two equations yielded dif
ferent values of NRC for some states. Equation
V was used for our calculations.
A state’s export growth relative to the nation,
as reflected in its NRC, is due to its industrial
mix effect (IME) and its competitive effect (CE)—
which identifies the extent that exports of indi
vidual state industries grew at rates different
from their national counterparts. There is an
additional factor, called the allocative effect
(AE), which can be interpreted as a measure of
the degree to which a state’s exports were con
centrated in industries at the beginning of the
study period that grew faster than the national
industry average. Thus, for a given state,
(2) NRC = I IMEj + 1. CE, + 1. AE,.
The industry mix effect is measured by first
calculating what the state’s 1986 exports would
have been if, given its actual 1976 industrial
mix of exports, a state’s exports for each in
dustry grew at the national industry rate. The
IME is the difference between this hypothetical
level and the level of 1986 exports the state
would have had if (1) it had the same export

sure confidentiality. To impute this data, which accounted
for less than 3 percent of the 1976 or 1986 continental
U.S. direct exports totals, other available indicators of
state export activity, such as total export-related shipments
and export-related employment, were used.

Values Used In Shift-Share Example (dollar amounts in millions)
Nation N

State S
1986
1976
Exports Exports

Total
Industry 1
Industry 2

$10
4
6

$17.6
5.6
12.0

1.76

S,s

G,s

0.4
0.6

1.4
2.0

mix as the nation and (2) its exports had all
grown at the corresponding national rate. A
state’s IME is calculated by the following:
(3) IME = I E.(Su-S in .
)Gin
The competitive effect, which examines the
differential industry growth rates of state vs.
national exports, is calculated by first
calculating the level of exports that the state
would have achieved in 1986 if each of its in
dustry’s exports had grown at its actual rate,
but assuming that the state had an industrial
structure identical to the nation. The CE is
simply the difference between this level and the
state export level that would have existed in
1986 if the state’s industrial mix of exports and
export growth had been identical to the na
tion’s. Thus, the competitive effect is calculated
as:
(4) CE = I E,Sln ,- G in
(G1
).
Finally, the allocative effect is calculated as
follows:
(5) AE = J. E,(S,, - Sin
)(GiI - Gin
).
The allocative effect indicates the degree to
which a state’s exports are concentrated in in
dustries whose exports have grown more rapid
ly than at the national level.
To facilitate meaningful comparisons among
states, each state's shift-share components re
ported in table 2 were divided by the level of
state exports that would have resulted in 1986
if such exports had grown at the national rate
between 1976 and 1986. That is, each state’s
components are divided by actual 1986 exports
minus the state's net relative change. Thus, these
“normalized” results indicate the percentage
deviation of actual 1986 state exports from the

1986
1976
Exports Exports

$100
20
80

$158
30
128

1.58

s,N

GiN

0.2
0.8

1.5
1.6

level that would have resulted from export
growth at the national 1976-86 rate.

A Numerical Example
An example may clarify the application of the
shift-share technique. Suppose 1976 and 1986
exports in state S and nation N are as shown in
the table. Es, the 1976 base period export level
in S is $10 million. For each of the two in
dustries, growth rates are simply the ratio of
1986 to 1976 exports. For example, GlS = 1.4
(5.6/4). Export shares are each industry’s share
of total 1976 exports. For example, SlS = o.4
(4/10). S’s net relative change can be calculated
using equation 1:
(1) NRC = EsGs-E sGn.
Substituting data for S and N yields:
10(1.76)-10(1.58) = $1.8 million.
Alternatively, NRC can be calculated using equa
tion 1’:
(1’) NRC = X (EsS,sGls—
EsSiN ).
GiN
Substituting data for S and N yields:
for i = 1: 10(0.4X1.4)-10(0.2)(1.5) = $2.6 million
for i = 2: 10(0.6)(2.0)- 10(0.8)(1.6) = -0.8
Total NRC = $1.8 million
The industrial mix effect for S is found using
equation 3:
(3) IME = I Es(SiS- S iN
)GiN.
Substituting data for S and N yields:
for i= 1: 10(0.4-0.2)1.5 =
for i = 2: 10(0.6-0.8)1.6 =

$3.0 million
-3.2

Total IME = -$0.2 million

SEPTEMBER/OCTOBER 1990

S's competitive effect is calculated using equa
tion 4:
(4) CE = I E8 (GiS-G iN
SiN
).
Substituting data for S and N yields:
for i= l: 10(0.2)(1.4-1.5) = -$0.2 million
for i = 2: 10(0.81(2.0-1.6) =
3.2
Total CE =

$3.0 million

Finally, S’s allocative effect is computed using
equation 5:
( A E = I Es(sis- SiN)(Gi G iN)
5)
S.
Substituting data for S and N yields:
for i = l: 10(0.4-0.2)(1.4-1.5) = -$0.2 million
for i = 2: 10(0.6-0.8)(2.0-1.6) = -0.8
Total AE = -$1.0 million
As equation 2 shows, a region’s NRC is the
sum of its IME, CE and AE [$1.8 million =
(-$0.2 million) + $3.0 million + (-$1.0
million)]. These results indicate that S’s 1986 ex
ports were $1.8 million higher than if they had
grown at the national rate between 1976 and
1986. S’s 1976 exports were relatively more con
centrated than were the nation’s exports in in
dustry 1, a comparatively slow-growing industry
at the national level. As indicated by IME, this
unfavorable industry mix caused S’s 1986 ex
ports to be $0.2 million below the level it would
have achieved if its 1976 export mix had been
identical to that of the nation.
S’s CE indicates that its 1986 exports were
$3.0 million higher because exports of its in
dustries grew faster than the corresponding in
dustries at the national level. Although export

FEDERAL RESERVE BANK OF ST. LOUIS

growth of S’s industry 1 was slightly slower
than the national rate, this influence was more
than offset by industry 2’s substantially fasterthan-national growth; since industry 2 ac
counted for an 80 percent share of national ex
ports and, therefore, was weighted more heavi
ly than industry 1 in computing total CE, S’s CE
was positive.
The AE value, which reflects differences bet
ween S and N in both industry mix and relative
industry growth, was negative. This result
reflects S’s relatively higher- (lower-)thannational export concentration in 1976 in in
dustry 1 (industry 2), in which its exports grew
slower (faster) than the national average in the
1976-86 period. To summarize, this example
shows that S’s exports grew faster than the na
tion’s exports, despite S’s unfavorable mix of ex
port sectors; this occurred because its in
dustries’ exports grew faster than exports of the
corresponding industries at the national level.
To ease comparison among states, each of S’s
shift-share components is expressed as a percen
tage of S’s 1986 level of exports that would
have resulted if S’s 1976 exports had expanded
at the national rate between 1976 and 1986.
This normalizing factor is S’s actual 1986 export
level minus its NRC or, in the current example,
17.6-1.8 or 15.8. In percentage terms, the nor
malized components are NRC, 11.4 percent;
IME, -1.3 percent; CE, 19.0 percent and AE,
-6.3 percent. S’s 1986 exports were 11.4 per
cent greater than if they had grown at the na
tional rate from 1976 to 1986. Although the
state industry mix depressed 1988 exports by
1.3 percent from the level that would have ex
isted had other things been equal, its relatively
fast growth of individual industries, expressed
in CE, allowed S’s exports to grow more rapidly
than did exports at the national level.

Michael T. Belongia and
K. Alec Chrystal

Michael T. Belongia is an assistant vice president at the
Federal Reserve Bank of St. Louis, and K. Alec Chrystal is the
National Westminster Bank Professor of Personal Finance in
The Business School, City University London. Lynn D. Dietrich
and Lora Holman provided research assistance.

The Pitfalls of Exchange Rate
Targeting: A Case Study from
the United Kingdom
"My advice has been for Britain to retain its system of flexible exchange
rates and to stay out of the present arrangements of the ERM.... It would
not be in Britain’s or, I believe, Europe's interest to join the present halfbaked system."
Alan A. Walters

INCE THE BREAKDOWN of the Bretton
Woods System of fixed exchange rates in the
early 1970s, the search for a new monetary
order has continued. In theory, the system that
was adopted nearly 20 years ago seems ideal: it
allowed exchange rates to float freely and each
country to pursue an independent domestic
monetary policy while the exchange rate com
pensated for differences in economic conditions
across countries. The actual behavior of nominal
exchange rates during this floating period, how
ever, has been extremely volatile in the short
run; moreover, over longer periods, prolonged

'Typically, factors such as relative rates of economic perfor
mance, foreign-domestic interest rate differentials, relative
inflation rates and changes in a nation’s trade deficit or
surplus come under the umbrella of economic “ fundamen
tals,” linked by economic theory to exchange rate move
ments. See Coughlin and Koedijk (1990) for a review

swings in real exchange rates have occurred,
which appear hard to explain in terms of eco
nomic fundamentals.1
Although, in one sense, this volatility is prim a
fa c ie evidence that the major currencies have
floated freely, the desirable characteristics of
floating exchange rates have been offset, in the
minds of many observers, by substantial disrup
tions in trade flows that such currency fluctua
tions are thought to have caused. In the United
States, for example, the strong appreciation of
the dollar in the early 1980s and the large, per-

of movements in the real exchange rates of the major cur
rencies and tests of alternative approaches to explain
them.

SEPTEMBER/OCTOBER 1990

sistent U.S. current account deficit that devel
oped during the same period are often cited as
the adverse consequences of unbridled freefloating exchange rates.2 At less aggregated
levels, the loss of jobs and market share by
firms in the automobile and steel industries are
viewed by some observers as a consequence of
freely floating exchange rates.
In this article, we discuss the benefits of using
monetary policy to peg the value of the exchange
rate at some desired level and analyze the
mechanics and likely effects of resorting to ex
change rate targeting as an approach to conduc
ting monetary policy. We first outline a simple
model of exchange rate determination and use
this framework to argue that exchange rate
pegging could produce desirable monetary policy
actions only when the real economy is stable.
This proposition is illustrated with a case study
of exchange rate pegging in the United Kingdom
and the economic consequences associated with
this policy.

SOME FUNDAMENTAL EXCHANGE
RATE DISTINCTIONS
Before discussing the mechanics of exchange
rate determination, one must understand the
difference between real and nominal exchange
rates. Such an understanding will explain why
pegging the value of the exchange rate is attrac
tive to some policymakers.
The nominal exchange rate is the relative
price of one currency in terms of another. For
example, if it takes two dollars to buy one Brit
ish pound, the exchange rate could be quoted
as $2.0/£; alternatively, it could be quoted as
0.5£/$. Many newspapers quote the exchange
rate both ways.
The nominal exchange rate tells us nothing,
however, about the “purchasing power” of one
currency vis-a-vis another. The real exchange
rate is useful for this purpose; it is obtained by
multiplying the nominal exchange rate by the
ratio of the domestic and foreign price levels.
The real exchange rate (RER) can be written as

2See, for example, Destler and Henning (1989).
3Among the problems in making this calculation are the
choice of a conceptual measure for P and the potential
measurement errors associated with this choice. The CPI,
PPI and indexes of unit labor costs have been used to
calculate real exchange rates, sometimes with significantly
FEDERAL
RESERVE BANK OF ST. LOUIS

p
RER = E ( ^ ), where E is the nominal exchange
rate expressed as units of foreign currency per
domestic currency unit, P is the domestic price
level and P* is the foreign price level. Unlike
the nominal exchange rate, the real exchange
rate is not observed in financial markets; in
stead, it must be approximated by a calculation
similar to this.3
Distinguishing between real and nominal ex
change rates would be pointless if the nominal
exchange rate and the prices of goods and ser
vices in different countries all responded to a
common influence at the same rate and by the
correct magnitudes so that no relative prices
were changed, even in the short run. As a gen
eral rule, however, nominal exchange rates ad
just more quickly than prices to actual or ex
pected changes in economic variables.4 Because E
adjusts more quickly than does ( ^ ) to econom
ic factors, the real exchange rate will mimic, at
least temporarily, movements in the nominal ex
change rate. These differences in the speed of
price adjustments are important because, until
both the nominal exchange rate and prices ad
just fully to a change in another variable, a na
tion’s pattern of trade and trade balance may be
disrupted.
Consider, for example, some good news about
the U.S. economy that leads to a 10 percent ap
preciation of the dollar but no immediate
response in either U.S. or foreign prices. This
event causes a 10 percent appreciation of the
real exchange rate. This change immediately
makes foreign goods 10 percent less expensive
to U.S. citizens and U.S. goods 10 percent more
expensive to foreigners. Eventually, the increased
demand for foreign goods and reduced demand
for U.S. goods (through a variety of mechanisms)
will result in higher prices of foreign goods and
lower prices of U.S. goods; in the interim, how
ever, foreign exports to the United States will
rise and U.S. exports to the rest of the world
will decline. Thus, while U.S. consumers will en
joy a temporary boost in their purchasing
power, some U.S. firms will be harmed tem-

different results. See Batten and Belongia (1987) for a
discussion of factors that may affect the measurement of
the real exchange rate.
Contracts, both for goods prices and wages, often are
cited as a reason for sluggish price level adjustments.

porarily by a decline in their domestic and ex
port sales. After domestic and foreign goods
prices adjust fully to the news that caused the
dollar appreciation, the real exchange rate will
have returned to its original level and neither
U.S. consumers nor U.S. exporters will be better
or worse off.

The Exchange Rate as a P olicy
Target
Typically, policymakers have made the nominal
exchange rate a primary target for policy actions
when real exchange rate changes were judged
to have had significantly adverse effects on
their country’s domestic and export sales. Econ
omists have investigated the exchange rate-export
relationship from two perspectives: (1) How
much does a change in the level of the real ex
change rate affect trade and (2) How much does
exchange rate variability affect trade? The
responsiveness of export sales to changes in the
level of the real exchange rate has been esti
mated for a variety of commodities; in general,
economists have found significant effects.
From the policy standpoint, however, moving
the real exchange rate to a new level has not
been a frequent topic of policy discussions.5 In
stead, the issue has been one of stabilizing the
nominal value of the exchange rate and, specifi
cally of reducing the adverse trade effects of
exchange rate variability associated with a freefloating exchange rate system. Indeed, the Euro
pean Monetary System (EMS), a cooperative
agreement among members of the European
Community, was created in 1978 to reduce ex
change rate volatility because it was widely felt
that intra-European Community trade was being
impeded by the costs of this volatility. It has
never been clear, however, what those costs
are.6 Moreover, arguments that volatile ex
change rates impede trade because of the "in

5While the September 1985 Plaza Accord may seem to be
one exception to this discussion, subsequent meetings to
that agreement have tended to focus on stabilizing the ex
change rate around some target value.
6See Ungerer, et al. (1986), pp. 17-18, for a discussion of
the ambiguity associated with arguments of how exchange
rate variability might affect trade.
7For a review of the evidence, see Farrell et al. (1983) who
conclude that, generally, the relationship is not supported
by the empirical evidence. DeGrauwe (1988), however,
finds some evidence suggesting a negative association.
8See Meltzer (1990) for a review of the alternative
arguments.

creased uncertainty” they generate have found
weak or no empirical support.7 Finally, even
economic theory does not demonstrate that
reducing exchange rate variability will con
tribute to improved economic performance.8
Whether a real exchange rate that is too high
or too volatile really produces short-run damage
to the export sector, the presumption that such
damage does occur is the main reason behind
efforts to peg the value of the nominal ex
change rate.

A SIMPLE MODEL OF THE EX
CHANGE RATE
Any discussion of the implications of attemp
ting to peg the value of exchange rate must
begin with a simple notion of why exchange
rates change in the first place—that is, with a
model of the factors that influence the ex
change rate. Suppose that there are two coun
tries: the home country and the rest of the
world, whose economic variables are denoted
by an *. The nominal exchange rate between
these two currencies can be written as (all
variables, except interest rates, are in
logarithms):9
(1) e = (m *-m ) - h (i*-i) - k (y *-y ) + s
where:
e = the nominal exchange rate in units of
foreign currency per unit of domestic
currency;
m - nominal money supply;
i = nominal interest rate;
y = real GNP;
k = the income elasticity of real money
demand;
h = the interest semi-elasticity of real money
balances;1
0
s = a “shift” factor that reflects the impact on
the exchange rate of all factors other than
those shown.

9This exchange rate model, based on the standard
monetary approach to the balance of payments, is taken
from Dornbusch (1980). The model assumes that “ un
covered interest parity” holds, which means that, at every
point in time, the interest rate differential (i*-i) is equal to
the expected change in the exchange rate. Thus, any
shock that affects the expected path of the exchange rate
will be reflected in the interest differential.
10An elasticity, such as k or h, represents the percentage
change in one variable (e) in response to a 1 percent
change in some other variable [(i*-i) or (y*-y)]; h is a semi
elasticity because the interest rate terms are not expressed
in logarithms.

SEPTEMBER/OCTOBER 1990

The equation states that a country's currency
will depreciate (e will decline) if domestic money
growth accelerates, domestic nominal interest
rates decline, or domestic real economic growth
slows relative to changes in the same variables
in the foreign country; moreover, there also are
exogenous shocks that can adversely affect the
exchange rate independent of the three in
fluences just described. In this model, policy
makers can affect the nominal exchange rate
either through monetary policy, which affects
m, or through policies that affect domestic in
terest rates or output; their influence on e, of
course, depends on relative changes in these
variables. The shift factor, s, presumably is
beyond the ability of policymakers to control.
Equation 1 can be modified to express the
real exchange rate as follows:
(2) rer = (m *-m ) - h (i*-i) - k (y*-y)
-(p * -p ) + s,
where p* and p are the logarithms of the
foreign and domestic price levels, respectively.
Thus, equations 1 and 2 show that the nominal
and real exchange rates are affected by essen
tially the same variables. And, empirical work
shows that the nominal and real exchange rates
typically move together in the short run, sug
gesting that aggregate price levels and their dif
ferential adjust more slowly than the other fac
tors discussed above.1
1

Policy Levers and the Exchange
Rate
A comparison between equations 1 and 2
highlights a key conclusion for any policymaker
concerned with exchange rate targets. While
changes in domestic monetary policy that affect
m relative to m* could move the nominal ex
change rate in equation 1 permanently to a new
level, monetary policy can change the real ex
change rate only if the nominal exchange rate
and aggregate price level differential adjust at
different speeds. Even in this case, the change
in the rea l exchange rate will be only temporary.
In equation 2, for example, an increase in the
domestic money supply (m) reduces rer while
an increase in the domestic price level (p) in
creases it. If one believes that changes in the
domestic price level ultimately are proportional

"See, for example, Mussa (1986).
FEDERAL
RESERVE BANK OF ST. LOUIS

to changes in the domestic money supply—a
standard interpretation of the quantity theory
of money—equation 2 shows that these effects
will offset each other, at least in the long run;
thus, the real exchange rate, but not the nomi
nal rate, will return to its former level. Indeed,
this "netting out” effect could occur even in the
short run if people anticipated the monetary
policy change and responded to it by quickly
raising prices. The conclusion from equation 2,
then, is that monetary policy actions have no
permanent effects on the real exchange rate;
any effect is strictly temporary.
Given equations 1 and 2, we can now con
sider two possible “sources” of exchange rate
changes. Broadly speaking, the two sources are
nominal factors that originate from a change in
the money supply and real factors that originate
outside of the monetary sector of the economy.
What difference does the source of the change
in the exchange rate make?
Suppose that the source of the exchange rate
movement is a nominal one—that the domestic
money stock is growing rapidly relative to that
in the foreign country. By itself, this would pro
duce some domestic inflation and a continuing
fall in the currency’s nominal value. If the mone
tary authority were pegging the nominal value
of the exchange rate, however, it would be
forced to buy back the "excess” money with
foreign reserves or otherwise tighten monetary
policy. As long as the foreign money supply
growth were unchanged, pegging the exchange
rate would have had the beneficial effect of for
cing the monetary authority to slow money
growth and domestic inflation.
In contrast, suppose some supply-side im
provement boosts domestic real GNP, causing
the nominal exchange rate to rise. The effect on
the real exchange rate depends on how much
the effect of a lower domestic price level (or
slower inflation rate) offsets the effect of the
higher real income level (or growth rate). If the
monetary authority resists this rise in the nomi
nal exchange rate, however, the result will be
faster money growth and a higher unnecessary
inflation.
The problem for policymakers is that only the
nominal exchange rate can be observed in world
financial markets and, as such, exchange rate
targets also are expressed in nominal terms. For

a variety of reasons, however, policymakers have
trouble distinguishing whether changes in the
nominal exchange rate are due to nominal or
real sources. Moreover, they are generally ig
norant as well as to whether these changes are
permanent or temporary.

many are intervening similarly to support the
value of a third currency (the dollar). Pepper
(1990) has argued, however, that Nigel Lawson,
then Chancellor of the Exchequer, used the
Louvre Accord as an opportunity to introduce a
nominal exchange rate target of DM 3.0/£.1
4

An early example of the problem associated
with this confusion was experienced by the
United Kingdom in 1977. The United Kingdom
had recently started producing oil and was
beginning the transition from net importer to
substantial exporter of oil. Both a nominal and
real appreciation of the currency would be ex
pected under the circumstances. The monetary
authorities, however, initially resisted the nomi
nal appreciation by selling pounds and buying
massive amounts of foreign currencies.1 This
2
strategy was later abandoned because of its
unacceptable implications for domestic infla
tion.1 We now turn to a more recent episode
3
of attempted exchange rate pegging, which fol
lowed the same pattern as the earlier episode.

Although data on intervention in specific cur
rencies are not available, the actual pattern of
net foreign exchange reserves at the Bank of
England, shown in table 1, is broadly consistent
with Pepper's view. The one-year interval of a
nearly stable DM 3.0/£ exchange rate is assoc
iated with frequent, and often massive, foreign
exchange interventions. Instead of steady inter
ventions in one direction, as one would expect
from an effort to support the value of the dollar,
the amount of intervention varies widely across
months; it even switches direction, with reduc
tions in foreign exchange reserves in some
months.

A CASE STUDY OF THE UNITED
KINGDOM SINCE 1987
In figure 1, the daily DM/£ exchange rate is
plotted over the most recent three-year period.
Overall, there are four striking aspects of these
data: a large appreciation of pound against DM
in February 1987, an extremely stable rate at
close to DM 3.0/£ between February 1987 and
March 1988, another large appreciation of the
pound in March 1988 and a general and sharp
depreciation of the pound throughout 1989.
Each of these episodes is discussed briefly below.
The appreciation of the pound from DM 2.747/£
on February 3, 1987, to DM 2.95/£ by March 18
is associated with the Louvre Accord. This
agreement pledged cooperative monetary policies
among the G-7 countries to strengthen what was
then a weakening value of the dollar. This
agreement presumably implied relatively restric
tive monetary policy in the United States and
relatively expansionary monetary policies among
the other G-7 members. In theory, there is no
reason for the pound to appreciate against the
DM if both the United Kingdom and West Ger

12U.K. foreign exchange reserves rose from SDR2.3 billion
in 1976 to SDR16.1 billion in 1977, a seven-fold increase.
13See Chrystal (1984) for a thorough discussion of this
episode.

The data in figure 1 confirm this view as well.
They show that, while the DM/£ exchange rate
was flat over the period, the pound appreciated
steadily and substantially against the dollarrising from $1.52 on February 3, 1987, to $1.90
by January 1988. Thus, the evidence in both the
table and figure is consistent with the view that
the Bank of England varied the amount of in
tervention to keep the DM/£ rate near a rate of
DM 3.0/£ while paying less attention to the
movement in the $/£ rate.
Although the United Kingdom apparently
directed its monetary policy to achieve an ex
change rate target of 3.0DM per pound through
out most of 1987, it is unclear why this objec
tive was chosen. In view of the earlier discus
sion, countries direct monetary policy to ex
change rate objectives when exports are weak
and unemployment in export industries is ris
ing. In the United Kingdom, however, exports
were strong at the time this policy strategy was
adopted and, overall, the United Kingdom econ
omy was performing quite well: inflation had
declined substantially, real growth was strong
and the government was running an increasing
budget surplus.

14ln the United Kingdom, the Chancellor of the Exchequer
has ultimate responsibility for monetary policy. The Bank
of England is under the Chancellor’s direct control.

SEPTEMBER/OCTOBER 1990

Figure 1
The DM/Pound and $/Pound Exchange Rate
DM/Pound

$/Pound

Weekly Data

3.30

2.00

3.20

1.90

3.10

1.80

3.00

1.70

2.90

1.60

2.80

1.50

1.40

2.70
1987

Table 1
Changes in U.K. Foreign Exchange
Reserves (millions of dollars)
February 1987
$ 287
March
1,785
2,912
April
4,760
May
-230
June
499
July
-457
August
380
September
6,699
October
November
31
3,737
December
January 1988
38
-2 5
February
2,225
March
514
April
814
May
SOURCE: Datastream

FEDERAL
RESERVE BANK OF ST. LOUIS

1988

1989

Why, then, peg the pound to the DM? To con
dense a long and complicated series of events
leading up to this decision, the monetary ag
gregates in the United Kingdom—as in other in
dustrialized countries—began to behave in the
1980s in ways that were perceived to be er
ratic.1 In particular, the traditional linkages be
5
tween monetary actions and changes in either
output or inflation appeared to weaken or evap
orate. Thus, if the monetary aggregates could
not be relied upon for guidance in conducting
monetary policy, becoming a “shadow” member
of the EMS’s Exchange Rate Mechanism (ERM)
and pegging the pound to the DM might have
been viewed as a plausible alternative for
monetary targeting.1 Indeed, one advantage of
6

15This shift in the behavior of the U.K. monetary aggregates
was exacerbated by the abolition, in June 1980, of a quan
titative control on bank liabilities (known as the corset) and
other regulatory changes.
16Much of this discussion is drawn from Pepper (1990).
Although circumstantial evidence seems to support this
view, to our knowledge, no official statements exist that ex
plicitly establish a DM 3.0/£ exchange rate target.

Figure 2
Three-Month Eurocurrency Rates for Germany
and the United Kingdom

the EMS from the perspective of its high infla
tion members is that low and stable inflation in
Germany will discipline their internal policies—
especially monetary policy—if their currencies
are tied to the DM. As we shall see, however, it
is not enough just to peg your currency to the
DM; you must peg it at the correct rate as well.
Early in 1988, the pound again began to ap
preciate strongly, both in real and nominal
terms, because of a widening U.K.-German in
terest rate differential on one hand and several
well-publicized events that presaged stronger
U.K. real growth on the other hand. As figure 2
shows, the U.K.-German interest differential
widened considerably between December 1987
and February 1988. On the real side, the end of
a ban on overtime work by miners on March 7,
the announcement of a large oil discovery on
March 8 and a large tax-reducing budget on
March 16 are consistent with reductions in the

values of either the (i* - i) and (y* -y ) terms in
equation 1 and, therefore, appreciations of the
pound. Each of these events would tend to raise
both the real rate of return on physical assets
and real output for the United Kingdom in
equation 1 so that (if there were no immediate
change in U.K. price levels) the real interest
rate and real output differentials would decline
and the pound would appreciate.
What could be done about this rise in the ex
change rate? From a theoretical standpoint, the
term (m *-m ) is an obvious policy lever in equa
tion 1. And, because the money stock can be
controlled by the central bank, this is an effec
tive lever with which to achieve a reduction in
its currency’s nominal value, should that be the
desired policy result. What is required is a
faster growth rate of its money stock relative to
the money growth in the nation against which
the exchange rate has appreciated.

SEPTEMBER/OCTOBER 1990

Figure 3
Three-Month Moving Average of U.K. MO Growth

How does this relate to the recent U.K. ex
perience? Figure 3 plots the growth rate of MO
(the U.K. monetary base), the monetary variable
targeted by the Bank of England in the late
1980s, and the target ranges that the Bank had
established for its policy objective. As the figure
shows, while the target range for MO was being
adjusted downward, actual MO growth was in
creasing on average.1 Moreover, a sharp in
7
crease in MO growth occurred early in 1988,
when the pound appreciated substantially above
the level of 3.0 DM that had prevailed for about
a year.

Figure 4 depicts the inevitable consequences
of this action. As growth in MO expanded, the
U.K. inflation rate, with a lag, also accelerated.
Recalling from equation 2 that faster money
growth in the United Kingdom will be associ
ated with a decline in the (nominal) DM/£ ex
change rate, the data in figures 1-3 show that
the point at which the pound began to decline
in value coincides closely with the time at which
U.K. money growth began to increase and U.K.
interest rates fell. Thus, while expansionary
monetary actions by the Bank of England
resulted in a reduction in the DM/£ nominal ex-

17The effects of this monetary expansion were reinforced by
four incremental reductions of 0.5 percentage points each
in the U.K. base lending rate between March 9 and May
11. Thus, while expansionary U.K. monetary policy was
helping to reduce the pound's value by decreasing the
value of the (m*-m) term in equation 2, further downward
pressure on the pound was coming from a widening of the
(i*-i) term in equation 2. This sharp decline in U.K. interest
rates and its impact on the German-U.K. interest differen
tial during the March-May interval also are shown clearly
in figure 2. The base rate, set by the Bank of England,
was reduced from 11 percent to 9 percent. See Pepper
(1990).

Another factor in these U.K. interest rate movements is
the “ excess credibility” problem. The idea is that, under
normal circumstances, an investment abroad carries the
risk of an adverse exchange rate movement in addition to
all of the usual risks. When a country is believed to be
pegging its exchange rate, however, the exchange rate
risk is reduced and, for a given foreign-domestic interest
differential, additional capital will flow into the country that
is pegging. Thus, once investors perceived and believed in
a continuation of the U.K.’s attempt to peg the pound,
financial capital flowed into the United Kingdom and tended
to reduce U.K. interest rates.

FEDERAL RESERVE BANK OF ST. LOUIS

Figure 4
Four-Quarter Moving Average of U.K. Inflation Rate

change rate, that effort also increased the U.K.
inflation rate from 4.1 percent in 1987 to 7.8
percent in 1989 and 9.7 percent as of May
1990.1
8
The final episode, the sharp decline in the
pound during 1989, reflects the increase in the
U.K. inflation rate, relative to the German infla
tion rate, that was produced by the excessive
U.K. monetary expansion. U.K. inflation, as mea
sured by the CPI, had averaged near 4 percent
between 1985-87. With the monetary expansion
of 1987-88 the inflation rate accelerated to a
peak of 12 percent in January 1989 and stood
at 9.7 percent as of May 1990. German inflation,
on the other hand, has remained relatively cons
tant, near an average of 4 percent. As equations
1 and 2 indicate, domestic monetary expansion
and the associated inflation will reduce a cur
rency's nominal value with no long-run effect
on its real value.

This result has led some observers to note the
irony of how an attempt to peg the exchange
rate to a low-inflation country (Germany) can
result in higher domestic inflation. The problem
with this reasoning, of course, is not the act of
pegging itself but, rather, selecting the wrong
value for the exchange rate target. If, for exam
ple, the United Kingdom had chosen to pursue
an exchange rate objective of, say, DM 3.3/£,
the real appreciation of the pound might have
been accommodated without resorting to an off
setting domestic monetary expansion. But, by
establishing the target at a level too low for the
fundamental economic differences that then
prevailed between the United Kingdom and Ger
many, the monetary expansion and subsequent
inflation were necessary results of maintaining
a DM 3.0 objective. Unfortunately, this type of
error is clear only in hindsight. The fundamen
tal problem for policymakers is how to deter-

18See Pepper (1990) for more detail on this episode.

SEPTEMBER/OCTOBER 1990

mine the "correct” value for the exchange rate
target in advance.

Belongia, Michael T. “ Prospects for International Policy Co
ordination: Some Lessons from the EMS,” this Review (Ju
ly/August 1988), pp. 19-29.

CONCLUSIONS

Chrystal, K. Alec. “ Dutch Disease or Monetarist Medicine?:
The British Economy under Mrs. Thatcher,” this Review
(May 1984), pp. 27-37.

Volatile movements in both exchange rates
and trade flows in the 1980s have prompted
many calls for nations to join cooperative
agreements that would peg bilateral nominal ex
change rates at some target level. As this paper
points out, however, nominal exchange rates
may change because of changes in underlying
real economic conditions. To maintain target
values for nominal exchange rates, domestic
monetary policy must pursue either an expan
sionary course that ultimately will produce only
an increase in the domestic inflation rate or a
contractionary stance that will exacerbate an
underlying economic downturn. Moreover,
whatever effects these monetary actions have
on the nominal exchange rate, they will have
only transitory effects on trade or other real
magnitudes; monetary actions will have no per
manent effect on the fundamental real causes of
the exchange rate change.
In the European context there are two impor
tant lessons for designing steps toward Economic
and Monetary Union (EMU). The first lesson is
that a system of mutuallly pegged currencies,
such as the current ERM, has obvious dangers.
Structural changes in one or another of the
member countries may cause unnecessary infla
tion or deflation if real exchange rate adjust
ments are resisted. The second lesson is that, if
a common currency were to be established,
obstacles to real market adjustments must be
eliminated.

REFERENCES
Batten, Dallas S., and Michael T. Belongia. “ Do The New
Exchange Rate Indexes Offer Better Answers to Old Ques
tions?” this Review (May 1987), pp. 5-17.

FEDERAL RESERVE BANK OF ST. LOUIS

Coughlin, Cletus C., and Kees G. Koedijk. "What Do We
Know About The Long-Run Real Exchange Rate?” this
Review (January/February 1990), pp. 36-48.
DeGrauwe, Paul. “ Exchange Rate Variability and the Slow
down in Growth of International Trade,” IMF Staff Papers
(March 1988), pp. 63-84.
Destler, I.M., and C.R. Henning. Dollar Politics: Exchange
Rate Policymaking in the United States (Washington: In
stitute for International Economics, 1989).
Dornbusch, Rudiger. “ Exchange Rate Economics: Where Do
We Stand?,” Brookings Papers on Economic Activity
(1:1980), pp. 143-85.
Farrell, Victoria, et al. “ Effects of Exchange Rate Variability
on International Trade and Other Economic Variables: A
Review of the Literature,” Staff Study No. 130 (Board of
Governors of the Federal Reserve System, December
1983).
Meltzer, Allan H. “ Some Empirical Findings on Differences
between EMS and non-EMS Regimes: Implications for Cur
rency Blocs,” Cato Journal (Fall 1990, forthcoming).
Mussa, Michael. “ Nominal Exchange Rate Regimes and the
Behavior of Real Exchange Rates: Evidence and Implica
tions,” in Karl Brunner and Allan H. Meltzer, eds., Real
Business Cycles, Real Exchange Rates and Actual Policies,
Carnegie-Rochester Conference Series on Public Policy
(Autumn 1986), pp. 117-214.
Pepper, Gordon. “ Money, Credit and Inflation” (London:
Institute of Economic Affairs, 1990).
Lingerer, Horst, Owen Evans, Thomas Mayer and Philip
Young. The European Monetary System: Recent
Developments, Occasional Paper No. 48 (International
Monetary Fund, December 1986).
Walters, Alan A. “A Life Philosophy,” The American Econ
omist (Fall 1989), pp. 18-24.

Daniel L. Thornton

Daniel L. Thornton is an assistant vice president at the Federal
Reserve Bank of St. Louis. Lynn D. Dietrich provided research
assistance.

Do Government Deficits
Matter?

H HERE HAS BEEN considerable concern
about the size of the federal deficit and the
rather modest success of the Gramm-RudmanHollings Act to reduce it. The traditional, that is,
Keynesian, view of deficit spending in macro
economics was that it could smooth out fluctua
tions in economic activity due to gaps between
saving and investment that were primarily the
result of exogenous shifts in investment. From
this perspective, deficit spending was seen as
both desirable and necessary to offset cyclical
fluctuations in economic activity that were char
acteristic of capitalist, free-market economies.
Recent discussions of government deficits,
however, have focused on their alleged adverse
effects. It is now common to blame deficit
spending for high real rates of interest and the
large and persistent trade deficit. Moreover,
there is widespread concern that large and per
sistent federal deficits will produce stagnant

'Recently, there have been three excellent surveys of the
empirical investigations of Ricardian equivalence. See
Barro (1987), Bernheim (1987) and Aschauer (1988). Also,
see Evans (1988), Koray and Hill (1988) and Leiderman
and Razin (1988) for more recent work that is not sum
marized in the three surveys. For some very recent work
on this issue, see Feldstein and Elmendorf (1990),
Modigliani and Sterling (1990) and Kormendi and Meguire
(1990). Most of the empirical evidence against the Ricar
dian view is at the microeconomic level using crosssectional or panel data.
2The reader is cautioned that the evidence presented here
can be considered suggestive for only a couple of

economic growth and result in renewed infla
tionary pressures.
An alternative view, called Ricardian equiva
lence, however, sees deficit spending as a har
binger of neither good nor ill. According to this
view, deficit spending cannot offset fluctuations
in economic activity due to exogenous shifts in
either private saving or investment; nor, can it
be blamed for high real interest rates or the
large trade deficit. Moreover, it has no influence
on the outlook for economic growth or infla
tion. Macroeconomic tests of the effects of
deficits on the economy using U.S. time-series
data have generally favored the Ricardian view.1
The purpose of this article is to review both
views of deficit spending to determine which
view of the relationships between deficit spen
ding and various associated macroeconomic var
iables is supported by evidence from 16 OECD
countries over the period 1975-86.3

reasons. First, strictly speaking, Ricardian equivalence is a
proposition about what happens when deficit spending is
substituted for taxes at an unchanged expenditure level.
Second, in a rational expectations framework, only unan
ticipated changes in deficits should affect macroeconomic
variables, and there is no plausible way to isolate the
unanticipated component of deficits from these annual
data. Third, the deficit measures used here are not
cyclically adjusted. Hence, they are endogenous, at least
in part.

SEPTEMBER/OCTOBER 1990

THE CONVENTIONAL VIEW OF
DEFICIT SPENDING
Total saving has two components: public and
private saving. Public saving measures the
government’s surplus or deficit position: sur
pluses represent positive public saving, while
deficits represent public dissaving. In the Keynes
ian view, deficits can be used to offset gaps
between saving and investment, thereby stabiliz
ing output around its potential (full-employment)
level. For example, if private saving is too high
relative to investment to achieve potential out
put, a government deficit will reduce the amount
of total saving and close the gap between saving
and investment. Conversely, if private saving is
too low relative to investment, a government
surplus can increase total saving. Hence, cyclical
swings in economic activity arising from move
ments in the saving/investment gap can be
shortened by changing public saving appropri
ately to maintain total saving at a level consis
tent with potential output.

Deficits and Private Saving
The above analysis is based on the assumption
that public and private saving are separate and
distinct activities.3 In the extreme, they are
viewed as totally independent: changes in public
saving have no direct effect on private saving
and vice versa.4 In this case, total saving will
change one-for-one with changes in public sav
ing. In the less extreme case, total saving is
simply positively correlated with public saving.5

Deficits and Interest Rates
In the conventional view, increases in the
deficit cause the interest rate to rise.6 Since the

3The following discussion is based on the standard IS/LM
aggregate demand-aggregate supply model of the
macroeconomy, where deficit spending is treated as
exogenous.
“Also, deficits can have an indirect effect on saving if the
increase in the deficit spending raises real income and,
hence, saving. Likewise, if a shift in the saving or invest
ment functions cause a change in income, the deficit will
respond endogenously because both expenditures and
taxes are related to the level of income. Hence, the
assumption that the series are independent applies to their
autonomous components.
5Strictly speaking, the above conclusion holds as long as
public and private savings are not perfect substitutes.
Note, however, fiscal policy is used to offset changes in
private savings, there may be a negative correlation bet
ween public and private savings even, if at a more
abstract level, the series are unrelated. That is, there may
be a policy reaction function where deficits respond,
presumably with a lag, to changes in private savings.

FEDERAL RESERVE BANK OF ST. LOUIS

interest rate is the price of credit, it is deter
mined by the supply and demand for credit. Ac
cording to the traditional view, an increase in
the deficit increases the demand for credit
relative to the supply and, consequently, in
creases the interest rate.7

Deficits and the Trade Deficit
The effect of deficits on credit demand can
also affect the relationship between government
deficits and the trade deficit. The effect of
deficit changes on interest rates depends on the
slope of the credit supply curve—the steeper
the slope, the larger the effect on interest rates.
Among other things, the slope of the credit
supply curve depends on the degree of open
ness of the nation’s money and capital markets
to the rest of the world. In general, the more
open these markets are, the flatter is the supply
of credit.
To see this more easily, let's recast the discus
sion in terms of the demand for securities. In
this framework, the price of bonds and, hence,
the interest rate are determined by the supply
and demand for bonds. The bond and credit
markets are related inversely: those who supply
credit are demanding bonds while those who
demand credit are supplying bonds. If financial
markets are open and competitive, the demand
curve for bonds seen by individual bond sup
pliers, including the government, is flat at the
market price of securities. In other words, no
matter how many bonds that individuals, firms
or the government may supply, individually,
they see no effect on the market price of bonds
and, hence, on the interest rate. All market par
ticipants, including the federal government, are

6The discussion here abstracts from the possible effects of
deficits on the rate of inflation, so the hypothesized effect
in this section is on the real interest rate.
7Most standard IS/LM aggregate demand-aggregate supply
models assume that there is no direct effect of the in
crease in the deficit on private savings; however, this
assumption is not necessary to obtain the usual implica
tions. All that is required is that the direct effect on private
savings (if the effect is to increase it) be less than the in
crease in the deficit. Also, the effect on the interest rate
would be small if the monetary authority monetized debt to
keep real interest rates low. While this is a controversial
issue, the evidence for the U.S. suggests that the Federal
Reserve has not monetized the debt. See Thornton (1984).
This discussion abstracts from the effect of the deficit on
interest rates via a deficit-induced increase in the rate of
inflation.

open, there may be no (zero) correlation be
tween government deficits and interest rates,
but there may be a positive correlation between
the government and trade deficits.

Figure 1
Price Level and Output Determ ination
P rice
L evel

The strength of this correlation depends on
the degree of openness of financial markets. If
financial markets are only partially open or ap
proximated by the competitive model, the
government deficit will be positively correlated
with both the interest rate and the trade deficit.

A gg re g a te S u p p ly
A g g re g a te S u p p ly *

Deficits and Econom ic Activity

A g g re g a te D em and*

A g g re g a te Dem and

O u tp u t

"small” relative to the total market, that is, each
is a "price-taker.”8 Thus, if the economy is suffi
ciently open, the money and capital markets are
competitive and the government is a price-taker,
changes in deficit spending will have no effect
on the interest rate. Increases or decreases in
the supply of bonds will be fully absorbed at
the current market price.
This does not mean, however, that deficit
spending has no effect on the economy. In the
above case, the deficit will be matched by
foreign claims on U.S. assets. That is, there will
be a capital inflow which, given the balance-ofpayments identity, must be matched by a trade
deficit. Hence, if the economy is "sufficiently”

8This argument also applies to the Federal Reserve’s con
trol over real interest rates in an open economy, and may
have implications for the Fed’s control over interest rates
in a “ closed” economy as well.
9The “ long run” tends to mean different things in
macroeconomics depending on the context in which it is
used. In the present IS/LM aggregate demand-aggregate
supply framework, it means the period of time for all
wages and prices to adjust to exogenous shocks. If all
wages and prices adjust instantaneously, the aggregate
supply curve would be vertical for all time periods. Hence,
the positively sloped aggregate supply curve in this model
results from some “ imperfections” which keep wages and
prices from adjusting instantaneously to shocks.
10Bernheim (1989) and others call this the neoclassical view.
In the neoclassical framework, crowding out necessarily

The effect of deficits on economic activity can
be illustrated in terms of the aggregate de
mand/aggregate supply model. The traditional
view asserts that increases in the deficit will in
crease the demand for goods and services, as il
lustrated in figure 1. The extent to which the
increase in aggregate demand affects output or
prices is determined (among other things) by
the slope of the aggregate supply curve. The
steeper the slope of the aggregate supply curve,
the larger the effect on the price level and the
smaller the effect on output. If the aggregate
supply curve were vertical, the deficit would af
fect only the price level.
Economic theory and empirical evidence sug
gest that the aggregate supply curve is upwardsloping in the short run.9 If this is true, there
should be a positive correlation between deficits
and both prices and output.

Adverse Effects o f Deficits on
Output
In addition to the positive effects of deficits
on output, it is often suggested that deficits can
affect output adversely because of their impact
on interest rates. In a closed economy, some
argue, the deficits raise the real interest rate
and, thereby, “crowd out” private investment.1
0

results from a full-employment assumption. The only way
that the government can spend more is for the private sec
tor to spend less. That is, interest rates must rise to in
crease private savings by the amount of the decrease in
public savings. The much-hypothesized adverse effects of
government deficits in the neoclassical view come either
through current government consumption replacing private
investment or the assumed lower marginal productivity of
public capital (see footnote 11). In any event, no explicit
distinction between the conventional and neoclassical
views is made in the text because they are comparable in
their implications (though certainly not with the fullemployment assumptions).

SEPTEMBER/OCTOBER 1990

Other things the same, the degree of crowding
out will be larger, the more responsive interest
rates are to deficit spending and the more
responsive investment spending is to changes in
the interest rate. Thus, the larger the effect of
deficits on interest rates, the smaller should be
the effect on current output. Even with this
crowding out effect, however, deficits and out
put are still expected to be positively corre
lated—at least in the short run.1
1

Deficits and Inflation
To be confused about the effect of deficits on
inflation and the price level is easy; therefore, it
is important to understand and recognize the
distinction between these effects. The fun
damental distinction can best be seen by noting
that the price level is measured at a specific
point in time, while inflation is measured over
an interval. According to the conventional view,
an increase in the deficit increases aggregate de
mand and, hence, the price level. Because ad
justments in the price level take place over
time, such changes are reflected in the mea
sured rate of inflation during the period of price
adjustment. Other things the same, however,
prices eventually will adjust to their new higher
level after which no further price changes will
be forthcoming. Consequently, while deficit
spending may affect the price level permanent
ly, it has only a temporary affect on the rate of
inflation.

” Some people fear that deficits could have some longerterm adverse effects due to crowding out. They argue that,
if deficits lead to higher interest rates and lower invest
ment, society will be left with a smaller capital stock,
which, in turn, means that future output will be smaller
than it would have been in the absence of deficit-financed
expenditures. Consequently, in the short run, deficits and
output are expected to be positively correlated. In the long
run, however, they may be negatively correlated, or at
least negatively correlated with output growth. There is an
implicit assumption in this argument about the nature of
public expenditures or the relative productivity of “ public”
and “ private” capital. To see this, note that the supply of
output, Y „ is a function of labor, Lt, and the capital stock,
K„ i.e., Y, = f(L„ K,).
Output rises with both labor and the stock of capital.
Consequently, if a deficit-induced rise in interest rates
causes investment to fall in this period, the next period’s
capital stock will be smaller, as will the next period’s out
put. This assumes that all of the rise in the deficit was us
ed for current consumption. Alternatively, aggregate output
could be expressed as a function of labor and the public
and private capital stocks, K " and Kp'„ respectively, i.e.,
Y, = h(Lt, Kput, K "). If the rise in the deficit is used to ac
quire public capital, the immediate effect on output is in
determinant even in the case of complete crowding out,
i.e., the decline in private capital from what it would have
otherwise been is equal to the increase in public capital.
See Aschauer (1989a, 1989b).
FEDERAL RESERVE BANK OF ST. LOUIS

For deficits to really be considered infla
tionary, they must generate continuing in
creases in the price level. In terms of the ag
gregate demand and supply curves, deficits
would have to cause the aggregate demand
curve to keep shifting to the right at a faster
rate than the aggregate supply curve. This
could happen only if deficits become persistent
ly larger over time, or if they are being mone
tized (producing increases in the rate of money
growth). The dynamic aspect of inflation, how
ever, requires a permanent rise in the rate
of money growth in response to the larger
deficit.1
2

THE EFFECTS OF DEFICIT SPEND
ING: RICARDIAN EQUIVALENCE
An alternative view of the effects of deficit
spending is called Ricardian equivalence. Unlike
the Keynesian view, which sees public and
private saving as essentially unrelated, the
Ricardian view sees them as perfect substitutes.1
3
According to this view, changes in public saving
are matched by an equal but opposite change in
private saving. This fundamental difference
manifests itself in the answer to the question: Is
government debt part of society’s net wealth?
In the Keynesian view, the answer is "Yes,”
while the Ricardian alternative would answer
"No.” According to the Keynesian view, when

12Deficits can sustain inflation if the monetary authority at
tempts to peg the nominal interest rate [see Friedman
(1968)] or if deficits become explosive. In the latter case,
the ratio of interest-bearing government debt to output in
creases without limit forcing the monetary authority to
monetize the debt. See Sargent and Wallace (1981) and
Miller and Sargent (1984) for a discussion of these points
and McCallum (1984) for a qualification.
13This is merely a convenient, equivalent characterization.
Ricardian equivalence stems directly from an intertemporal
resource constraint. That is, for a given path of expen
ditures, a cut in present taxes necessarily implies an in
crease in future taxes of equal present value. If an in
dividual’s demand for goods and services depends on his
“ after tax” net worth—the present value of assets, the
present value of liabilities and the present value of
taxes—a fall in current taxes would necessarily be match
ed by an increase in the present value of future taxes.
Since a budget deficit merely rearranges the timing of the
tax liability, it cannot affect aggregate demand. The
decrease in public saving (the deficit) must be matched by
an increase in private saving.
Strictly speaking, however, Ricardian equivalence holds
only for a given path of government expenditures. There
can be real effects associated with changing the level or
timing of real government expenditures.

29
Figure 2
The Trade Deficit and the Government Deficit

I M

•

-1 5

-1 2

-9

-6

•
•

•

W
•• • • I
•

-3

G o v e rn m e n t Deficit

the government issues debt, the holder of the
debt views it as an asset; but taxpayers do not
view it as their liability. That is, they do not
believe that they will have to pay current or
future taxes to service or retire the debt. Conse
quently, their saving behavior is unaffected by
the debt issuance.
The Ricardian view, on the other hand, as
serts that individuals believe that they, or their
heirs, will have to pay taxes to service and re
tire the increased debt. Because they perceive
an increase in the present value of their taxes
that just offsets deficit-financed expenditures,
the stock of government debt is not part of so
ciety’s net wealth. Hence, a rise in the deficit (a
fall in public saving) will be matched by a rise
in private saving in anticipation of future taxes.
If public and private saving are perfect sub
stitutes, a decline in public saving—the increase
in the deficit—is offset by an increase in private
saving, with no effect on the gap between sav
ing and investment. According to this view,
then, deficit spending cannot be used to smooth
out cyclical variations in economic activity due
to exogenous swings in the gap between saving
and investment. Moreover, the increased de
mand for credit due to deficit spending will
be matched by an increase in the supply of
credit due to an increase in private saving. Also,
the increase in aggregate demand that results

14The countries are: Australia, Austria, Belgium, Canada,
Finland, France, Greece, Great Britain, West Germany,
Ireland, Japan, Norway, Netherlands, Sweden, Switzerland
and the United States. The interest rate was not available
for Austria, Canada, Finland or Greece.

from a rise in deficit spending will be offset by
a decrease in aggregate demand due to the fall
in private spending—that is, there is no net
change in aggregate demand.
Because the Ricardian view holds that changes
in deficit have no net effect on the excess de
mand for credit or aggregate demand, deficits
should be uncorrelated with interest rates, the
trade deficit, the price level, output or total sav
ing. Deficits and private saving, however, should
be perfectly, positively correlated.

EMPIRICAL EVIDENCE
The evidence presented here consists of sim
ple scatter plots between the government deficit
and the trade deficit, personal saving, real out
put growth, the price level, the inflation rate
and the nominal interest rate for 16 OECD
countries.1 (A more detailed statistical analysis
4
consistent with these scatter plots is presented
in the appendix). The currency-denominated
variables are expressed as a percent of the
country's gross domestic product, GDP, to put
them in common units, and all variables are ad
justed for their mean values.1
5
Figure 2 shows the relationship between the
so-called twin deficits: the trade and the govern
ment deficits. The scatter plots show that there

15These are pooled time-series cross-sectional data, where
the average level is allowed to vary each year. See the ap
pendix for details.

SEPTEMBER/OCTOBER 1990

is a weak (but statistically significant) positive
association between the two deficits. This
association is consistent with both the conven
tional view and U.S. time-series data.1 Barro
6
(1987), however, finds that the positive associa
tion exists for U.S. time-series data only when
data for the 1980s are included. This instability
is evident for these cross-sectional data as well.
When the sample includes only observations for
the earlier period, 1975-80, there is no signifi
cant positive relationship between the twin
deficits. These data are presented in figures 3,
while data for the later period, 1981-86, are
presented in figure 4. Moreover, further analy
sis suggests that the statistically significant posi
tive relationship does not hold up if other factors
are considered (see the appendix).
Figure 5 shows the relationship between per
sonal saving and the government deficit. The
strictest form of the conventional view argues
that these series should be unrelated, while the
Ricardian alternative argues that they should be
perfectly, positively correlated. Neither view is
supported by these cross-sectional data. Instead,
there is a statistically significant negative rela
tionship between these variables. This anomalous
result may stem from the response of both the
deficit and personal saving relative to GDP to
cyclical movements in output. For example, it is
well-known that deficits typically rise relative to
GDP as output falls and vice versa. If consumers
attempt to maintain their living standard in the
face of a temporary decline in output, personal
saving would fall relative to output. Consequent
ly, personal saving and government deficits rela
tive to GDP should be negatively correlated over
the business cycle. In any event, the statistically
significant negative relationship supports neither
hypothesis.1
7
Figures 6-9 show the relationship between the
government deficits and inflation, output growth,
the price level and interest rates. In all cases,

16See Barro (1987) and Dewald and Ulan (1990). Dewald
and Ulan argue that the observed positive association bet
ween the twin deficits in U.S. time-series data results from
failing to account for inflation or changes in the market
value of the U.S. federal debt or other major elements of
the U.S. net external wealth position. This argument,
however, requires a different definition of the deficit than
the one used in the conventional stories of the effects of
deficits on economic variables.
17Recently, Swamy, Kolluri and Singamsetti (1990) have sug
gested that finding any statistically significant relationship
between the deficit and interest rates—although the sign
of the coefficient is opposite that hypothesized by the con-

FEDERAL RESERVE BANK OF ST. LOUIS

there is no statistically significant relationship
between the deficit and these variables.1 Conse
8
quently, the cross-sectional data provides little
or no support for the conventional view. On the
other hand, support for the Ricardian alternative
is not overwhelming. While the lack of signifi
cant relationships between the deficit and in
terest rates, the trade deficit, output, prices,
inflation and economic growth is consistent with
the Ricardian view, the lack of a statistically
significant positive relationship between the
government deficit and private saving is not. In
deed, if deficits and private saving are perfectly,
positively correlated because public and private
saving are perfect substitutes, a positive rela
tionship between these variables should have
been apparent. Thus, at best, the evidence pre
sented here should be interpreted as cautiously
favoring the Ricardian view.

Implications o f the Results
Whether the evidence presented here is taken
to support, however cautiously, the Ricardian
alternative, the most interesting result is the
lack of evidence to support the conventional
view. In particular, the evidence suggests that
countries with large government deficits do not
have higher interest rates or larger trade defi
cits despite the widespread opinion to the con
trary. Of course, these puzzling results could be
an artifact of the simple measures of deficit
spending used here. If the relationships between
government deficits and either interest rates or
the trade deficit were as strong as many com
mentators suggest, however, it is odd that they
do not emerge in these annual data across coun
tries. These results, coupled with the fact that
more sophisticated empirical studies using U.S.
time-series data have also failed to uncover the
conventional relationships, should perhaps lead
advocates of the conventional view either to re
think their position or present some evidence to
support their claims.

ventional view—contradicts the Ricardian equivalence
paradigm. Such an extreme view is unwarranted.
18There are two exceptions when first-differences are used.
The one of some importance for the conventional wisdom
is the positive and statistically significant relationship be
tween the deficit and output growth, suggesting that larger
government deficits have at least some initial positive ef
fect on real output. Of course, if the rise in the deficit is
due to an increase in government spending, there is a
positive relationship between the deficit and GDP by
definition. See the appendix for details.

Figure 3
The Trade Deficit and the Government Deficit: 1975-1980
T ra d e Deficit

•

•M •

-1 0

-8

-6

-4

•
• •

• • «
• •

•

-2
G o v e rn m e nt Deficit

Figure 4
The Trade Deficit and the Government Deficit: 1981-1986
T ra d e Deficit

G o v e rn m e n t Deficit

Figure 5
Personal Saving and the Government Deficit
Personal S a vin g

P ersonal S a ving

G o v e rn m e nt Deficit

SEPTEMBER/OCTOBER 1990

Figure 6
The Rate of Inflation and the Government Deficit
Rate of Inflation

Rate of Inflation

Figure 7
The Growth Rate of Real GDP and the Government Deficit
G ro w th Rate of Real G D P

G ro w th Rate of Real G D P

•

5 I r..TU> ■.
M

-1 5

••

•

-1 2
G o v e rn m e n t Deficit

Figure 8
The Price Level and the Government Deficit

FEDERAL
RESERVE BANK OF ST. LOUIS

G o v e rn m e n t Deficit

•

Figure 9
The Interest Rate and the Government Deficit

REFERENCES
Aschauer, David. “ The Equilibrium Approach to Fiscal
Policy,” Journal of Money, Credit and Banking (February
1988), pp. 41-62.
________“ Does Public Capital Crowd Out Private Capital?”
Journal of Monetary Economics (September 1989a), pp.
171-88.
________“ Is Public Expenditure Productive?” Journal of
Monetary Economics (March 1989b), 177-200.
Barro, Robert J. “ The Ricardian Approach to Budget
Deficits,” Henry Thornton Lecture, City University Business
School, London, November 1987.
Bernheim, B. Douglas. “ Ricardian Equivalence: An Evalua
tion of Theory and Evidence,” NBER Working Paper No.
2330 (July 1987).
_______ . “A Neoclassical Perspective on Budget Deficits,”
Journal of Economic Perspectives (Spring 1989), pp. 55-72.
Chrystal, K. Alec, and Daniel L. Thornton. “ The
Macroeconomic Effects of Deficit Spending: A Review,” this
Review (November/December 1988), pp. 48-60.
Darrat, Ali F. “ Fiscal Deficits and Long-Term Interest Rates:
Further Evidence from Annual Data,” Southern Economic
Journal (October 1989), pp. 363-74.
Dewald, William G., and Michael Ulan. “ The Twin-Deficit Illu
sion,” Working Paper No. 90-17, U.S. Department of State,
Bureau of Economic and Business Affairs (February 1990).
Evans, Paul. “Are Consumers Ricardian? Evidence for the
United States,” Journal of Political Economy (October 1988),
pp. 983-1004.
_______ . “ Interest Rates and Expected Future Budget
Deficits in the United States,” Journal of Political Economy
(February 1987), pp. 34-58.
_______ . “ Is the Dollar High Because of Large Budget
Deficits?” Journal of Monetary Economics (November
1986), pp. 227-49.
Feldstein, Martin, and Douglas W. Elmendorf. “ Government
Debt, Government Spending, and Private Sector Behavior

Revisited: Comment,” American Economic Review (June
1990), pp. 589-99.
Friedman, Milton. “The Role of Monetary Policy,” American
Economic Review (March 1968), pp. 1-17.
Koray, Faik, and R. Carter Hill. “ Money, Debt and Economic
Activity,” Journal of Macroeconomics (Summer 1988), pp.
351-70.
Kormendi, Roger, and Philip Meguire. “ Government Debt,
Government Spending, and Private Sector Behavior: Reply
and Update,” American Economic Review (June 1990), pp.
604-17.
Laumas, Gurcharan S. “Anticipated Federal Budget Deficits,
Monetary Policy and the Rate of Interest,” Southern
Economic Journal (October 1989), pp. 375-82.
Leiderman, Leonardo, and Assaf Razin. “Testing Ricardian
Neutrality with an Intertemporal Stochastic Model,” Journal
of Money, Credit and Banking (February 1988), pp. 1-21.
McCallum, Bennett T. “Are Bond-Financed Deficits Infla
tionary? A Ricardian Analysis,” Journal of Political Economy
(February 1984), pp. 123-35.
Miller, Presten J., and Thomas J. Sargent. “A Reply to Dar
by,” Federal Reserve Bank of Minneapolis Quarterly Review
(Spring 1984), pp. 21-26.
Modigliani, Franco, and Arlie G. Sterling. “ Government
Debt, Government Spending, and Private Sector Behavior:
A Further Comment,” American Economic Review (June
1990), pp. 600-03.
Sargent, Thomas J., and Neil Wallace. “ Some Unpleasant
Monetarist Arithmetic,” Federal Reserve Bank of Min
neapolis Quarterly Review (Fall 1981), pp. 1-17.
Swamy, Paravastur A. V. B., Bharat R. Kolluri and Rao N.
Singamsetti. “What Do Regressions of Interest Rates on
Deficits Imply?” Southern Economic Journal (April 1990),
pp. 1010-28.
Tatom, John A. “ U.S. Investment in the 1980s: The Real
Story,” this Review (March/April 1989), pp. 3-15.
Thornton, Daniel L. “ Monetizing the Debt,” this Review
(December 1984), pp. 30-43.

SEPTEMBER/OCTOBER 1990

Appendix
An Econometric Analysis of the Effect of
Government Deficits
The purpose of this appendix is to see if there
is a statistical association between government
deficits and some important macroeconomic
variables as hypothesized by both the conven
tional view and the Ricardian alternative.
The analysis begins with the following general
equation:
(A.l) DVit = ait + /itDEFit + £„, i = l,..., K, and
3
t-1,..., T,
where DVit and DEFit denote the tth observation
for the ith country of the dependent variable
and the deficit measure, respectively, a it, and p it
denote fixed parameters and £it denotes a ran
dom error.
Equation A .l cannot be estimated because the
number of parameters exceeds the number of
observations. This problem can be circumvented
by obtaining time-series and/or cross-section
representations of equation A.l. The time-series
representation is obtained by imposing the
restrictions a it = a i and p it = /, for all t. The
?
cross-sectional representation is obtained by im
posing the restrictions a it = a , and /it = /, for
3
?
all i. These specifications are:
(A .2) D V it = a, + /?,DEFit + £ t
|

and
(A.3) DVit = a, + /3,DEFlt + £it.
A pooled time-series/cross-section representa
tion can be obtained by imposing the restric

'The interest rate is the three-month money market rate.
2This was also done by measuring these variables relative
to their mean value for the period and indexing them to
1975. When the variables were normalized in this way, a
measure of the level of GDP was also used as a depen
dent variable. In nearly all cases, estimates of equations
A.2 and A.3 resulted in insignificant coefficients on the
deficit measure, so the results are not reported here.
Also, the OECD data are based on the system of na
tional accounts which differs in a number of respects from
our system of national income and product accounts. One
of these is that capital expenditures by government are
not treated as current expenditures, so that the deficit is
current expenditures plus consumption of fixed capital less
revenue.

FEDERAL RESERVE BANK OF ST. LOUIS

tions ait = a and / it = ft for all i and t, to
?
obtain
(A.4) DVit = a + /JDEFit + £it.
(This is equivalent to imposing the restrictions
a l = a and (lt = ji for all i on the time-series
model or a t = a and /, = / for all t on the
?
?
cross-sectional model).
Equations A.2-A.4 were estimated with annual
observations on the government deficits, nomi
nal interest rate, the trade deficit, the price
level (1980 = 100), the inflation rate, real output
growth and private saving for 16 OECD coun
tries (K = 16) for which the relevant annual data
are available.1 The period is from 1975 to 1986
(T = 12). Because the currency-denominated
variables are expressed in the respective coun
try's currency, it is necessary to put these data
in common units. This was done by measuring
these variables as a percent of gross domestic
product, GDP.2
The equations were estimated both in levels
and first-differences. The latter form was
estimated because some of the data showed a
tendency to trend with time. The sample was
too small, however, to perform the usual tests
for stationarity. Estimates of first-order autocor
relation from equation A.2 suggest that firstdifferencing may result in over-differencing in
many cases. Because some of these data ap
peared to have trends, the pooled time-series/
cross-section equation, A.4, was estimated by
allowing the intercept term to vary with time.3
Because the estimates of equations A.2 and A.3

3The reported intercepts in tables A.1 and A.2 are for 1986.
The equations were also estimated using a time trend. The
results with this variable were not qualitatively different
from those that allowed for a time-varying intercept. Con
sequently, they are not reported here. Also, the equations
were estimated using the alternative dependent variables
as regressors. Except for the interest rate, however, the
qualitative results were unaffected, so these results are
not presented here.

are qualitatively the same as those of equation
A.4, only estimates of equation A.4 are
presented.

The Results
Estimates of equation A.4 for both levels and
first-differences are reported in table A .I.4 The
only coefficients that are statistically significant
in both the level and first-difference specifica
tions occur when the trade deficit or private
saving is the dependent variable. In the latter
case, however, the coefficient was negative,
which suggests that increases in deficit spending
are associated with decreases in personal sav
ing: when public saving decreases, so does
private saving. This result is not consistent with
the Ricardian view that public and private sav
ing are substitutes; however, it might reflect an
endogenous response of both government
deficits and personal saving to cyclical variations
in output. It is well-known that deficits tend to
rise relative to GDP when output falls. Likewise
it might be that personal saving also rises when
output falls, as individuals try to maintain their
level of consumption in the wake of declining
income levels. In this case, personal saving
would decline relative to GDP. The simultaneous
response of both government deficits and per
sonal saving to cyclical variation in income
could account for the statistically significant
negative coefficient on the deficit when per
sonal saving is the dependent variable.
Barro (1987) also found a positive and statisti
cally significant relationship between the two
deficits using U.S. data, but stated that this re
lationship emerges only when data for the 1980s
were included. When the sample was partitioned
into 1975-80 and 1981-86 periods, a positive
relationship emerged between the twin deficits
during both periods; however, consistent with
the findings of Barro, the relationship was not
statistically significant during the first period.5
Dewald and Ulan (1990) argue that the posi
tive relationship between the twin deficits in
the U.S. results from failing to account for infla
tion or changes in the market value of the net
U.S. federal debt or other major elements of the

4lt was assumed that variance was constant over time, but
differed over cross-sections, that is, E(tuE.J= o\ for i= j
and t = s and 0 otherwise. Estimates of equation A.2 in
dicated considerable cross-sectional heteroskedasticity.
5The estimated coefficient on the deficit variable during the
first period for the level specification was .181, with a t-

U.S. external wealth position. When adjustments
for these factors are made, they find no statisti
cally significant relationship between the two
deficits. Because this argument might apply to
cross-sectional data as well, caution should be
used in interpreting these results as evidence of
the conventional wisdom on the twin deficits.
This is especially true because the estimated
coefficient on the deficit is not large: a 1
percentage-point rise in the government deficit
relative to GDP would result in only about a .20
percentage point rise in the trade deficit relative
to GDP. Moreover, as shown later, the result is
not robust.
The statistical significance of the coefficient
on the deficit when inflation or the price level
is the dependent variable depends on whether
the equation is estimated in levels or firstdifferences, being insignificant in the former
case and significant in the latter. In either case,
however, the coefficient has a sign opposite that
of the conventional story. Consequently, these
results are not evidence for or against either of
these views.
The same result occurs when the growth rate
of GDP is the dependent variable. In this case,
however, the conventional wisdom does not
state the direction of the expected change. From
a short-run perspective, output growth should
be positively related to deficits. From a long-run
perspective, they should be negatively related.
The results in table A .l suggest a positive rela
tionship if first differences are used, with a 1
percentage-point increase in the deficit being
associated with a quarter of a percent increase
in the rate of growth of GDP. Of course, if the
increase in the deficit is due to an increase in
government spending, there is a positive rela
tionship between deficits and GDP by definition,
if crowding out is not complete.
Estimates of the effect of the deficit on in
terest rates also are sensitive to the specification
of the equation. When levels are used, there is
a negative, but statistically insignificant, relation
ship between interest rates and the deficit.
When first-differences are used, however, there
is a positive and statistically significant (a one-

statistic of 1.42. The comparable statistics for the second
period are .307 and 3.52, respectively. The same statistics
for the two periods for the first-difference specification are
.094 and 1.37 and .226 and 3.01, respectively.

SEPTEMBER/OCTOBER 1990

Table A.1
Pooled Time-Series, Cross-Sectional Estimates of Equation A.4
Levels
Dependent
variable

First difference

Constant

Deficit
measure

Trade deficit

-1.396*
(2.43)

0.195*
(2.42)

- 0.453
(1.43)

0.127*
(2.26)

Personal saving

9.163*
(14.25)

-0.447*
(6.59)

0.197
(0.77)

-0.170*
(3.43)

Output growth

2.349*
(5.67)

-0.060
(1.34)

-1.206*
(2.87)

0.244*
(3.73)

Inflation

4.444*
(5.76)

-0.061
(0.77)

0.327
(1.70)

-0.154*
(4.40)

Price level

1.517*
(75.44)

- 0.002
(1.17)

0.07*
(12.11)

-0.004*
(3.58)

Interest rate

8.120*
(10.89)

-0.020
(0.28)

-0.886
(1.96)

0.248*
(1.95)

Constant

Deficit
measure

’ Indicates statistical significance at the 5 percent level.

tailed test at the 5 percent significance level)
relationship between the two variables. Even if
one takes the results that are most supportive
of the conventional wisdom, however, the effect
of deficits on interest rates is fairly weak: a 1
percentage-point rise in the deficit relative to
GDP is estimated to produce about a 25 basispoint rise in nominal interest rates.

The First-Difference Results
While the results for the first-difference specifi
cation appear to indicate statistically significant
relationships between the government deficit and
all of the dependent variables, these relationships
are not as strong as the results in table A.l would
suggest. Scatter plots of the first-differenced data,
mean-adjusted, are presented in figures A.1-A.6.
These data suggest that much of the reported
statistically significant relationship could be due to
a relatively few outliers. In most cases, these are

6AII other variables save the interest rate, since the interest
rate was unavailable for Austria, Canada, Finland and
Greece.

FEDERAL
RESERVE BANK OF ST. LOUIS

attributable to two countries, Greece and Nor
way. Also, the reported results could be due to
the failure to account for other factors that af
fect the relevant dependent variable. Both issues
are investigated in table A.2, which reports the
results of estimates of the first-difference speci
fication when Greece and Norway are deleted
and, alternatively, when all other variables are
included as regressors.6 As the table shows, the
results are sensitive to both of these changes.
Deficits are not significantly related to the price
level or interest rates when the two countries
are deleted and no longer have a statistically
significant effect on the trade deficit, the price
level or the interest rate when the other vari
ables are included. The only significant effects
that are robust to these changes are those for
the effect of the deficit on private saving, out
put growth and inflation. In the first and last
case, however, the sign of the coefficient is op
posite that suggested by the conventional view.

Figure A1
Change in The Trade Deficit and the Government Deficit

Figure A2
Change in Personal Saving and the Government Deficit
Personal S a vin g

P ersonal S a vin g

G o v e rn m e nt Deficit

Figure A3
Change in the Inflation Rate and the Government Deficit
Rate of Inflation

Rate o f Inflation

G o v e rn m e n t Deficit

SEPTEMBER/OCTOBER 1990

38
Figure A4
Change in the Growth Rate of Real GDP and the
Government Deficit
G ro w th Rate of Real G D P

G ro w th Rate o f Real G D P

G o v e rn m e n t Deficit

Figure A5
Change in the Price Level and the Government Deficit

G o v e rn m e nt Deficit

Figure A6
Change in the Interest Rate and the Government Deficit
Interest Rate

FEDERAL
RESERVE BANK OF ST. LOUIS

Interest Rates

G o v e rn m e n t Deficit

Table A. 2
Estimates of Two Alternative First-Difference Specifications
Greece and
Norway deleted
Dependent
variable

All alternative dependent
variables included

Constant

Deficit
variable

Constant

-0.611*
(2.03)

0.128*
(2.22)

0.072
(0.21)

-0.013
(0.23)

Personal saving

0.312
(1.28)

-0 .143’
(2.97)

0.400
(1.63)

-0.213*
(4.55)

Output growth

-1.056*
(2.31)

0.342*
(4.01)

-0.930*
(2.26)

0.330*
(4.66)

Inflation

0.378
(1.94)

-0.126*
(3.65)

0.112
(0.54)

-0.093*
(2.63)

Price level

0.051 *
(7.91)

-0.002
(1.73)

0.062*
(8.81)

-0.001
(0.95)

Interest rate

-1.093*
(2.29)

0.128
(0.78)

Trade deficit

-0.934
(1.95)

Deficit
variable

0.090
(0.61)

'Indicates statistical significance at the 5 percent level.

SEPTEMBER/OCTOBER 1990

Cletus C. Coughlin

Cletus C. Coughlin is a research officer at the Federal Reserve
Bank of St. Louis. Thomas A. Pollman provided research
assistance.

What Do Economic Models
Tell Us About the Effects of
the U.S.-Canada Free Trade
Agreement?

S t u d i e s o f t h e historic U.S.-Canada Free
Trade Agreement have produced conflicting
estimates of its economic effects.1 Not surpris
ingly, the numerous changes resulting from the
Free Trade Agreement will benefit some people
and harm others. To summarize the agreement’s
effects, m an y studies have estimated the mini
mum amount that individuals who gain would
be willing to accept to forego the changes and
the maximum that those who lose would be
willing to pay to prevent them.2 Subtracting the
value of the losses from the gains produces a
measure of net national welfare change. Using
such a measure, estimates for Canada, expressed
as a percentage of total economic activity, range
from large gains to small losses, while estimates
for the United States range from small gains to
small losses.
This paper closely examines five recent studies
to better understand their estimates as well as

1See Coughlin et al. (1988) for an introduction to the
theoretical arguments underlying protectionist trade
policies and the empirical evidence indicating that the
costs borne by consumers of such policies generally far
exceed the benefits captured by domestic producers and
government.
FEDERAL
RESERVE BANK OF ST. LOUIS

identify why they contradict each other. While
the five studies focus on the U.S.-Canada Free
Trade Agreement, they represent the typical
modeling approaches used to quantify the impact
of changes in trade laws. Thus, the following
discussion helps explain why these analyses
o fte n reach w id e ly d iffe re n t conclusions about

the same trade policy change. In addition, the
discussion points out some limitations of these
kinds of analyses.

AN OVERVIEW OF THE U.S.CANADA FREE TRADE
AGREEMENT
To assess the results and limitations of the dif
ferent studies, the key aspects of the U.S.-Canada
Free Trade Agreement first must be identified.
The agreement, summarized in table 1, estab-

2This measure of welfare change is called an equivalent income variation. For a brief discussion of this measure, see
Henderson and Quandt (1980).

Table 1
The Major Provisions of the U.S.-Canada Free Trade Agreement
Tariffs
Eliminates all tariffs on U.S. and Canadian goods by January 1, 1998.
Rule of origin To prevent third-country goods from receiving preferential tariff treatment, goods
must pass a “ rule of origin” test. Goods produced entirely in the United States
or Canada qualify immediately, while goods containing imports from countries
outside the agreement qualify if they are processed enough to result in one of
several specified changes in tariff classification.
Ends customs user fees for goods and duty drawback programs, which returns
Customs
previously paid duties on imports when they are incorporated in goods subse
quently exported, by 1994 for bilateral trade and duty waivers linked to perfor
mance requirements by 1998 (except for the Auto Pact).
Eliminates import and export quotas unless allowed by the General Agreement
Quotas
on Tariffs and Trade or grandfathered by the Free Trade Agreement.
Reaffirms General Agreement on Tariffs and Trade principle preventing
National
discrimination against imported goods.
treatment
Prohibits use of product standards as a trade barrier and provides for national
Standards
treatment of testing labs and certification bodies.
Eliminates all bilateral tariffs and export subsidies and limits or eliminates quan
Agriculture
titative restrictions on some products, including meat. Eliminates Canadian im
port licenses for wheat, oats and barley when U.S. crop price supports are
equal or less than those in Canada.
Wine and dis Removes most discriminatory practices against wine or spirits imported
from the other country.
tilled spirits
Prohibits most import and export restrictions on energy goods, including
Energy
minimum export prices. Requires any export quotas used to enforce short supp
ly or conservation measures to share resources proportionately. Provides for
Alaskan oil exports of up to 50,000 barrels per day to Canada.
Replaces the Canadian content rule for duty-free Auto Pact imports into the
Autos
U.S. with tougher Free Trade Agreement content rule. (Most auto trade already
is duty-free under the U.S.-Canada Auto Pact.) Does not change rules for Auto
Pact-qualified companies importing duty-free into Canada, but does not allow
new companies to qualify. Permits U.S. auto and parts exports that meet the
Agreement rule to enter Canada at Agreement tariff rates, which phase out over
10 years. Ends all Canadian duty remission programs for autos by 1998.
Allows temporary import restrictions to protect domestic industries harmed
Emergency
by imports from the other country in limited circumstances.
action
Expands the size of government procurement markets that will be open to
Government
procurement suppliers from the other country.
Commits governments not to discriminate against covered service providers of
Services
the other country when making future laws or regulations. (Exempts transporta
tion services.)
Temporary
Facilitates travel for business visitors, investors, traders, professionals and
executives transferred intra-company.
visas
Provides national treatment for establishment, acquisition, sale and conduct and
Investment
operation of businesses. (Exempts transportation.) Commits Canada to end
review of indirect acquisitions and to raise to C$150 million (in constant 1992
Canadian dollars) the threshold for review of direct acquisitions. Bans imposition
of most investment performance requirements.

SEPTEMBER/OCTOBER 1990

Table 1 (continued)
The Major Provisions of the U.S.-Canada Free Trade Agreement
Financial
Exempts U.S. bank subsidiaries in Canada from Canada’s 16 percent ceiling
services
on assets of foreign banks. Ends Canada’s foreign ownership restriction on U.S.
purchases of shares in federally regulated insurance and trust companies.
Reviews U.S. firms’ applications for entry into Canadian financial markets on
the same basis as Canadian firms’ applications. Permits banks in the U.S. to
underwrite and deal in debt securities fully backed by the Government of
Canada or political subdivisions. Guarantees continuation of multi-state bran
ches of Canadian banks.
General dispute Establishes a binational commission to resolve disagreements (except for finansettlement
cial services and countervailing duty and anti-dumping duty cases).
Countervailing Allows countries to continue to apply existing national laws. Replaces
duty and anti- court review with a binational panel (when requested), which must apply
dumping dispute national law in rendering decisions under international law.
settlement
Softwood
Preserves the 1986 agreement with Canada on provincial pricing practices,
lumber
Culture
Exempts cultural industries from the Free Trade Agreement, but authorizes
measures of equivalent commercial effect in response to actions otherwise in
consistent with the Agreement. Cultural activities exempted include the publica
tion, sale, distribution or exhibition of books, magazines and newspapers; recor
ding of all kinds; and radio, television and cable dissemination.
SOURCE; U.S. Department of Commerce, International Trade Administration, Summary of the U.S.Canada Free Trade Agreement (February 1988).

lishes a free trade area to be phased in between
January 1, 1989, and January 1, 1998.3 By that
time, the United States and Canada will have
eliminated nearly all of the barriers restricting
trade in goods and services between each other,
while retaining their individual trade policies
with all other countries.4

which virtually all tariffs on bilateral trade were
removed previously. Some of the tariffs listed in
table 2 have already been eliminated, some will
be eliminated in five equal reductions (20 per
cent per year) on each January 1 beginning in
1989 and the remainder will be eliminated in 10
equal reductions (10 percent per year).5

Trade Barriers

The agreement is not limited to tariff barriers.
Virtually all import and export restrictions, such
as import quotas and embargoes, have been
eliminated. Less visible trade barriers have been
eliminated as well. For example, in government
procurement, discrimination between U.S. and
Canadian suppliers is prohibited on qualifying
nonmilitary purchases exceeding $25,000, as
defined in the General Agreement on Tariffs

By January 1, 1998, all tariffs on merchandise
trade between the United States and Canada
will be eliminated. As table 2 reveals, prior to
the agreement, U.S. tariff rates on imports from
Canada were lower than Canadian tariffs on im
ports from the United States in every industry
except transportation equipment, an industry in

3See Little (1988), Copeland (1989) and the U.S. Depart
ment of Commerce (1988) for summaries of the agree
ment.
4The retention of individual trade policies relative to non
partner countries distinguishes a free trade agreement
from a customs union. The countries in a customs union,
such as the “ pre-1992” European Community, have
FEDERAL
RESERVE BANK OF ST. LOUIS

abolished trade barriers among themselves and have the
same trade barriers on imports from non-member coun
tries.
5According to the U.S. International Trade Commission
(1990), requests from traders on both sides have prompted
tariffs on more than 400 products to be eliminated more
quickly than was originally agreed upon.

Table 2
Average Post-Tokyo Round Tariff Rates of the United States
and Canada1
Industry

U.S. tariff rates
on imports from

Canadian
tariff rates
on imports from

Canada

Agriculture
Food
Textiles
Clothing
Leather products
Footwear
Wood products
Furniture and fixtures
Paper products
Printing and publishing
Chemicals
Petroleum products
Rubber products
Nonmetal mineral products
Glass products
Iron and steel
Nonferrous metals
Metal products
Nonelectric machinery
Electric machinery
Transportation equipment
Miscellaneous manufactures
Average

Other

United States

Other

1.6%
3.8
7.2
18.4
2.5
9.0
0.2
4.6
0.0
0.3
0.6
0.0
3.2
0.3
5.7
2.7
0.5
4.0
2.2
4.5
0.0
0.9

1.8%
4.8
9.1
21.4
3.8
8.9
3.8
2.9
1.3
0.7
3.5
0.1
2.0
7.2
5.8
3.9
0.8
4.4
3.2
4.1
2.5
2.0

2.2%
5.4
16.9
23.7
4.0
21.5
2.5
14.3
6.6
1.1
7.9
0.4
7.3
4.4
6.9
5.1
3.3
8.6
4.6
7.5
0.0
5.0

1.8%
6.1
16.4
22.1
8.7
21.9
4.9
14.1
6.5
1.0
7.0
0.1
6.0
8.5
7.9
5.5
2.7
8.9
4.8
7.1
2.5
5.3

0.7

4.3

3.8

7.4

'Weighted by bilateral trade.
SOURCE: Brown, Drusilla K. and Robert M. Stern. “ A Modeling Perspective,” in Robert M.
Stern, Philip H. Trezise and John Whalley, eds. Perspectives on a U.S.-Canadian Free
Trade Agreement (The Brookings Institution, 1987).
and Trade (GATT). Currently, GATT prohibits
discrimination on procurements exceeding
$171,000.6

Industry Issues
The agreement also deals with trade issues
peculiar to specific industries such as agricul
ture, automotive products, energy and alcoholic
beverages. Tariffs affecting agricultural trade
will be eliminated by 1998. Export subsidies and

6While the agreement virtually eliminates national
discrimination between U.S. and Canadian producers,
there is still discrimination between producers within the
free trade area and those outside. When imported parts
and materials are used in the production of a good that is
shipped from one partner country to another, the origin of

quantitative import restrictions on some pro
ducts, like meat, were eliminated immediately.
The countries also agreed to cooperate with
each other in negotiations with other countries
to eliminate all subsidies that distort agricultural
trade.
Due to the 1965 U.S.-Canada Auto Pact,
95 percent of bilateral auto trade is already duty
free. A major issue, however, had arisen be-

the good must be determined. In the agreement, goods
with imported inputs qualify as North American if they
have sufficient value-added to permit them to be exported
under a different tariff classification than the one under
which the inputs were imported.

SEPTEMBER/OCTOBER 1990

cause Canada had been enticing Asian auto pro
ducers to locate production facilities in Canada
by rebating duties (taxes) paid on parts im
ported to Canada when these parts, after some
production activity, were then exported. All
Canadian duty remission programs will be ter
minated by 1998. Until then, the agreement
does not change the rules for companies already
qualifying for duty-free imports into Canada
under the existing auto agreement, although it
does not allow any new firms to qualify.
Most trade restrictions on energy resources
are prohibited. The exceptions are limited to
cases of shortages, conservation or national
security; however, even in the case of short
ages, the reduced supplies must be shared be
tween both countries.
Trade barriers on alcoholic beverages have
been only partially eliminated. Although the
agreement eliminates some barriers limiting the
trade of wines and distilled spirits, Canadian
barriers limiting the importation of U.S. beer
will remain unchanged.7

Investment
The agreement provides national treatment
for all aspects of the establishment and opera
tion of businesses. This means that U.S.-owned
firms in Canada and Canadian-owned firms in
the United States will be treated as domestic
firms. The agreement addresses U.S. concerns
about Canadian policies designed to influence
foreign investment. Specifically, Canada agreed
to stop imposing performance requirements,
such as requiring an investor to export a cer
tain amount of goods, and, beginning in 1992,
to stop screening U.S. direct acquisitions of
Canadian assets of less than C$150 million (in
constant 1992 Canadian dollars).

Financial Services
The concept of national treatment has been
extended to financial services, making this the
first time that the United States has reached a
bilateral agreement covering all financial ser
vices. Virtually all discriminatory Canadian prac
tices are eliminated. For example, the Canadian
assets of foreign bank subsidiaries operating in

7For additional information on U.S.-Canada beer trade, see
Carter et al. (1989).
eThe agreement’s resolution of countervailing duty and anti
dumping cases is temporary. The United States and
FEDERAL RESERVE BANK OF ST. LOUIS

Canada previously were limited to no more than
16 percent of all domestic assets of the Cana
dian banking system. Under the agreement, U.S.
bank subsidiaries are no longer subject to this
limitation on their market share.

Other Services
The agreement is also noteworthy because it
is the first international agreement dealing with
trade and investment barriers in the service in
dustries. Many service industries such as trans
portation, basic telecommunications, health,
education and social services, however, are not
covered. Nonetheless, the principle of national
treatment has been extended to most commer
cial services such as construction, tourism, com
puter services, wholesale and retail trade, man
agement services and other business services.
Since many of these require the movement of
personnel for limited periods, the agreement
changes immigration regulations to facilitate
business-related travel.

Implementation and Dispute
Settlement
The Canada-United States Trade Commission
has been established to implement the agree
ment. This group will resolve disputes on all
matters except financial services and those in
volving charges of either foreign government
export subsidies, called countervailing duty
cases, or sales of a good abroad at a price lower
than is charged in the domestic market, called
anti-dumping cases. Disputes involving financial
services will be handled by a formal consulta
tive mechanism between the U.S. Department of
the Treasury and the Canadian Department of
Finance. Countervailing duty and anti-dumping
cases are subject to review, upon request, by a
special binational panel.8 This panel, whose deci
sion is final, reviews the case in light of the
domestic laws of the importing country. Thus,
each country retains the right to enforce its
own laws.

MODELING THE AGREEMENT:
UNDERLYING ISSUES
The preceding overview identifies the many
legislative changes. Researchers attempting to

Canada have five to seven years to develop a permanent
solution; otherwise, either country may terminate the
agreement.

estimate their likely consequences face numerous
issues involving economic theory, modeling ap
proaches and measurement. While economic
theory provides much assistance in modeling
the effects of the agreement, it provides no
definitive conclusion about the welfare conse
quences for a specific country. One aspect of
the modeling process in which economic theory
plays an important role is in the selection of the
modeling approach. No matter which approach
is chosen, the far-reaching nature of the agree
ment prevents some aspects from being incor
porated into quantitative models. The wide
range of trade barriers affected by the agree
ment poses further problems. To understand
fully the usefulness of the studies discussed later,
these underlying issues are examined below.

The Uncertain Welfare Effects o f
Changes in Trade Laws

and services from other member countries, then
the greater specialization in production based
on comparative advantage will enhance the eco
nomic welfare of the member countries. The
term for this welfare-increasing reallocation of
production is "trade creation.”
"Trade diversion,” however, occurs when
lower-cost imports from outside the free trade
area are replaced by imports produced at higher
cost from a member country. This can occur
because goods imported from a member are not
subject to tariffs or other restrictions, while
goods potentially imported from non-member
countries continue to face the same barriers as
before. This trade diversion shifts production
away from the pattern consistent with com
parative advantage.1
0
The relative magnitudes of trade creation and
trade diversion determine whether the welfare
of members rises or falls. Thus, it is natural to
use quantitative models to assist in assessing
whether the agreement is likely to be beneficial
or harmful to the two countries.

International trade theory generally concludes
that free trade leads to the most efficient utiliz
ation of the world’s resources and, consequent
ly, maximizes the value of world output. Every
move toward freer trade, such as the elimina
tion of trade barriers among a group of coun
tries, however, does not necessarily increase na
tional welfare.9 The reason for this apparent
contradiction is that the formation of a free
trade area, while eliminating one trade distor
tion, creates another. The tariff reduction
resulting from a free trade agreement will elimi
nate the distortion between domestic goods and
imports from the partner country, a change that
increases national welfare. A new distortion,
however, is created between imports from the
partner country and those from non-partner
countries that reduces welfare.

The standard way to model the effects of in
ternational trade policy changes is to construct
and solve a theoretical model using assumed
values of critical parameters to derive the solu
tion. General equilibrium models usually are
chosen to capture the numerous market interac
tions that take place both within and among
countries. Thus, the standard model used for
this purpose is called an "applied general equi
librium model.”1
1

These opposing welfare effects, illustrated in
the shaded insert on pages 47-50, can be de
scribed very simply. If the formation of a free
trade area results in the domestic production of
some goods and services in one member coun
try being replaced by imports of these goods

Equilibrium in this type of model is character
ized by a set of prices such that the market de
mand for each output and input equals the mar
ket supply. The market supply for each output
reflects the production decisions of firms moti
vated by profit maximization. Input demand

9Viner (1950) showed that the formation of a customs union
could have different welfare effects for partner as well as
non-partner countries. Viner’s finding is an example of the
“ theory of the second best.” This theory shows that, if all
conditions required to maximize welfare are not satisfied,
then satisfying one or more additional conditions will not
necessarily produce a higher level of welfare.
10The empirical importance of trade diversion for Canada
can be doubted because, prior to the agreement, the ma
jority of Canadian imports—more than 71 percent in 1986,
according to the Council of Economic Advisers (1988)—
were already provided by the United States. Thus, the

M odeling the Effects o f Trade
Policy Changes: T w o Approaches

chances that the share of Canadian imports provided by
the United States would rise substantially are small. Note
also that information contained in table 2 shows that
average Canadian tariff rates on imports from the United
States already were below those on imports from other
countries.
11See Shoven and Whalley (1984) for an introduction to ap
plied general equilibrium models. This introduction
highlights how these models work using a numerical ex
ample. In addition, they provide a review of research using
these models to examine international trade issues.

SEPTEMBER/OCTOBER 1990

functions are derived from production functions
that use only capital and labor. The market de
mand for each product in this model reflects
utility-maximizing consumption decisions of in
dividuals. Demand functions are derived from
maximizing utility functions subject to a budget
constraint. The budget constraint contains a
measure of income generated through the sup
ply of inputs that are used in the production
process.
Values for parameters in the production and
utility functions must be specified to solve such
a model. For example, the elasticity of substitu
tion between capital and labor in the produc
tion function, as well as that between goods in
the utility function, must be specified.1 The
2
solution to the model, characterized by the
market prices of the inputs and outputs and the
corresponding quantities, ensures that market
demand equals market supply for all inputs and
outputs and that profits are zero in each in
dustry. There are major differences, however,
among models that fall into this category. The
estimates discussed below rely on one of two
fundamental theoretical approaches.
The traditional approach focuses on the gains
from comparative advantage. A key assumption
in the comparative advantage approach is that
markets are competitive. Using the HeckscherOhlin approach to international trade, produc
tion costs vary across countries prior to interna
tional trade because of differences in the en
dowments of productive resources. Countries
relatively well-endowed with certain resources
are able to p rod u ce those goods w h o se p ro d u c
tion requires relatively large amounts of these
resources at lower cost than other countries.
With free trade, countries gain by exchanging
export goods that they produce at relatively
lower cost for imports produced at relatively
lower cost from other countries. In essence,

12The elasticity of input substitution is a measure of the
responsiveness of the optimal labor/capital combination to
a change in the relative prices of these inputs. The
elasticity of substitution between goods is defined
analogously.
13The welfare consequences of economies of scale can be
negative. In models of trade between a large and a small
country, Markusen and Melvin (1981) and Ethier (1982)
show that the output of goods with increasing returns to
scale in the small country might decrease rather than in
crease with a change from autarchy, a situation in which
the country engages in no international trade, to free
trade. The increase in average production costs in the
small country may more than offset the benefits of
specialization due to comparative advantage.
FEDERAL RESERVE BANK OF ST. LOUIS

trade allows each country to export its abun
dant productive resources in exchange for the
relatively abundant productive resources of its
trading partners. The existence of trade bar
riers prevents some of the gains from produc
ing, trading and consuming on the basis of com
parative advantage from being realized.
The alternative approach used in estimating
the effects of the reduction of trade barriers ap
plies standard models used in industrial organi
zation to international trade. In this approach,
most output markets are assumed to be imper
fectly competitive rather than perfectly com
petitive. Frequently, the imperfectly competitive
markets result from the existence of economies
of scale in production. These economies of scale
cause declining average production costs as the
level of output expands. In this case, the in
crease in the size of the market for individual
producers allows for gains from trade.1
3

Non-Quantified Features o f the
Agreem ent
Irrespective of the modeling approach chosen,
several features of the agreement are not quan
tified.1 Many of these features are potentially
4
important and, thus, could significantly alter the
agreement's net impact.
One of the agreement’s goals is the creation of
a more stable business environment for all forms
of international business activity. Many Cana
dians believe secure access to the U.S. market is
essential for Canadian economic prosperity, and
that such access is being threatened by the U.S.’
increasing use of trade laws for protectionist
reasons.1 As an example justifying this concern,
5
Copeland (1989) notes that the United States
recently placed a 35 percent import duty on
Canadian shakes and shingles to protect U.S.
producers.

14Shea (1988) highlights similar aspects of the agreement
that are not quantified.
15See Lea (1987) and Trezise (1987) for assessments of the
Canadian perspective on this issue. The use of trade laws
for protectionist reasons is termed “ contingent protection.”
Contingent protection encompasses a range of import
restrictions, such as anti-dumping and countervailing
duties, escape clause petitions and legislation dealing with
“ unfair” international trade. Ethier (1988), p. 234, has
argued that the “ use of the anti-dumping law has greatly
increased and the statute is likely to become a principal
protectionist tool,” while a recent article in The Economist
(October 22, 1988), p. 16, referred to the “ capricious inter
pretation and enforcement” of U.S. anti-dumping laws as
a potentially important trade barrier.

A Supply and Demand Analysis of the Welfare
Consequences of a Free Trade Agreement
The models discussed in the text are gener
al equilibrium models in that the prices and
quantities of nearly all the goods and services
are determined simultaneously. This approach
is appropriate in assessing the consequences
of the U.S.-Canada Free Trade Agreement be
cause the reduction of tariffs on numerous
goods sets in motion simultaneous adjustments
in a large number of markets. The following
discussion, however, is couched in terms of
one good. Such an approach, termed partial
equilibrium, is used because the primary
theoretical welfare consequences of the Free
Trade Agreement can be illustrated using
supply and demand diagrams that are rela
tively easy to understand.1

Figure 1
Welfare Analysis Using Supply and
Demand Curves

Consum er’s Surplus and P r o
ducer’s Surplus
Underlying the welfare consequences of a
tariff reduction are two standard welfare
measures, consumer’s and producer’s sur
plus.2 These measures can be illustrated sim
ply using a supply and demand diagram. Fig
ure 1 shows the Canadian supply, Sc, and de
mand, Dc, for a specific good. Equilibrium in
this market is characterized by a price per
unit of IJ and a quantity of Q
,,.
The benefits to Canadians from consuming
this good are given by the area under the de
mand curve from O to Q0 or area OZBQ,,.
Since Canadian consumers pay a price per
unit of IJ, their expenditures on the good are
IJ times Qo or area OP0
B(^0. Consumer’s sur
plus is the difference between the total bene
fits and the total expenditures, which is the
triangular area IJZB. The supply curve indi
cates the price that producers must receive
to induce them to produce each additional
unit of the good. The area OABQ0 represents
the value of goods foregone (the costs) to

1The discussion ignores the welfare consequences of
terms-of-trade effects.
2These welfare measures are widely used as approx
imate measures of the equivalent variation measure of
welfare change that is actually used in the empirical
studies examined in the text.
3ln the text, a modeling approach based on comparative
advantage is discussed. Such models assume produc

produce this good. Producer’s surplus is the
area above the supply curve and below the
horizontal line reflecting the market price or
triangular area AIJB.3 These welfare measures
can be added to generate an estimate of the
increase in Canadian welfare from producing
and consuming this good. Graphically, this is
simply area AZB.

Trade Creation and Trade
Diversion
The effects of a trade-creating free trade
agreement are illustrated in figure 2. Iden
tical to figure 1, Sc and Dc remain the Cana
dian supply and demand for a specific good.
In figure 2, however, the Canadians can im
port the good from the United States at a
fixed price of PF From a national perspec.4

tion functions with constant returns to scale. When all
supply curves are horizontal, producer’s surplus is
zero.
4The models discussed in the text generally assume that
products in a given industry are slightly differentiated
across countries. This assumption is ignored here to
simplify the analysis.

SEPTEMBER/OCTOBER 1990

Figure 2
A Trade-Creating Free Trade Agreement

subject to a tariff equal to the distance PF ,
PX
so the amount of tariff revenue (PF times
PT
Q2 ) is the rectangular area GFIH.
Q3
The Free Trade Agreement eliminates the
tariff on imports, so the price of this good in
Canada becomes PF This lower price causes
.
Canadian consumption to increase from Q3 to
Q4 and production to decrease from Q2 to Q,.
Thus, imports increase from Q2 to Q,Q„.
Q3
This is an example of trade creation because
some production in Canada is replaced by im
ports from the United States.

tive, the supply of this good from the United
States, S ,s is represented by a horizontal line.
i= ,
Even though Canadian production and con
sumption decisions are assumed to have no
effect on the price that Canada pays for im
ports from the United States, Canadian trade
policies can affect the price of this good
within Canada. With free trade, the Canadian
price of the good is PF the same price that
,
Canadians pay to import the good from the
United States. Thus, S/ represents the supply
JS
of imports of this good from the United
States under free trade. If, however, a tariff
equal to the distance PF is imposed on
PT
imports, then S?s represents the supply of im
ports relevant for Canadian production and
consumption decisions.
Before the Free Trade Agreement, imports
from the United States were subject to a
tariff, so S“s is the relevant supply of imports.
In this case, the price of this good in Canada
is PT, which means that Canadian production
is Q2 and consumption is Q3 The difference
between Canadian consumption and produc
tion, represented by the distance Q2 ,
Q3
reflects the quantity of Canadian imports
from the United States. These imports are

5The area EFG is referred to as the production effect,
while the area HIJ is the consumption effect. Viner
(1950), who pioneered this analysis, concentrated on
the production effect and ignored the consumption ef
fect. Thus, the definition of trade creation focuses on

FEDERAL RESERVE BANK OF ST. LOUIS

The Canadian welfare gain from this trade
creation can be identified graphically. The
elimination of the tariff, which allows in
creased consumption at a lower per unit
price, causes consumer’s surplus to increase
by the area PF
PTIJ. Part of this gain is a
transfer from Canadian producers whose
surplus drops by the area PfPtFE because of
the lower prices and the resulting lower out
put they produce. The elimination of the tar
iff on imports from the United States means
that the tariff revenue on imports from the
United States, area GFIH, originally paid by
Canadian consumers is returned to them as
part of the increase in their consumer’s sur
plus. After considering the transfers to con
sumers from producer’s surplus and tariff
revenue, the net Canadian welfare gain is
represented by the sum of the triangular
areas EFG and HIJ.5
The preceding analysis assumes that the
United States is the lowest-cost foreign sup
plier of the imported good. Trade diversion
arises, however, if the United States is not
the lowest-cost foreign supplier. Figure 3 il
lustrates a trade-diverting free trade agree
ment. Figure 3 contains the same supply and
demand curves as figure 2 plus two addi
tional curves. S“ is the free trade supply
curve of imports from a country M that is
not a party to the U.S.-Canada Free Trade
Agreement. S^is the free trade supply curve
adjusted by the tariff that Canada imposes on
imports from this country.
Before the Free Trade Agreement, Canada
imposes an identical tariff on imports of this

the replacement of domestic production by imports. It
is now standard practice, however, to view trade crea
tion as encompassing both the production and con
sumption effects.

Figure 3

A Trade-Diverting Free Trade Agreement

price. The reduction in producer’s surplus is
the area PF M
PT KE. Finally, since imports from
the United States are not subject to a tariff,
tariff revenue declines from WKBX to zero.
Netting out these welfare changes leaves two
triangular areas of gain, EKL and NBJ, and
one rectangular area of loss, WLNX.
The two triangular areas are the same
welfare gains that were highlighted in the
previous example. The area of loss, WLNX,
reflects the higher price that Canada pays
per unit for QS , the amount of Q imported
Q6
before the Free Trade Agreement. Obviously,
if the welfare effect associated with trade
diversion exceeds the sum of the two triangu
lar areas associated with trade creation, then
the net welfare change from the Free Trade
Agreement is negative.

The Effects o f a Free Trade
Agreem ent Assuming Econom ies
o f Scale

good from all countries. Thus, the distance
PfmPtm is equal to PF Because Canada im
PT.
posed the same tariff on imports from both
countries, PT is less than Px and Canadian
M
imports of this good will be solely from M.
Given the price of PT for this good, Canadian
M
production is Q5 and consumption is Q6 Thus,
.
imports from M are represented by the dis
tance Q5 . The amount of Canadian tariff
Q6
revenue is PpmPjm times Q5 or, more simply,
Q6
the rectangular area WKBX.
The Free Trade Agreement eliminates the
tariff on imports from the United States, but
not on those from M. Since the price of the
U.S.-produced good without the tariff, PF is
,
lower than the price of the same good pro
duced in M with the tariff, PX , Canadian im
M
ports will be diverted from M to the United
States. In addition to the switch in Canada's
imports from one country to the other, Cana
dian production declines from Q5 to Q,, Cana
dian consumption increases from Q6 to Q4
and Canadian imports rise from Q5 to Q,Q4.
Q6

The preceding analysis ignores the possibili
ty that Canadian manufacturing may be char
acterized by economies of scale. Figure 4
shows a case in which Canada exports a good
for which there are increasing returns to
scale; this is illustrated by the negatively
sloped average cost curve, ACC for Canadian
,
producers. The demand for this product by
Canadian consumers is represented by Dc;

Figure 4
Welfare Consequences of a Free Trade
Agreement Assuming Economies of Scale

The net welfare consequences are unclear.
Consumers are better off as a result of a
lower price and increased consumption. The
increase in consumer’s surplus is represented
by the area PF M On the other hand, Can
PT BJ.
adian producers are harmed by the lower

SEPTEMBER/OCTOBER 1990

the demand by U.S. consumers for this pro
duct is represented by Du D?s is the U.S. de
s.
mand before the Free Trade Agreement, while
Dps represents the larger U.S. demand that
results when the U.S. tariff on the Canadianproduced good is eliminated. Thus, Dc + D£s
represents the market demand faced by
Canadian producers before the agreement
and Dc+D£s represents the market demand
after the agreement.

curve, ACC with the market demand curve,
,
Dc + D?s. Thus, the equilibrium price and
quantity are PT and Qx. A reduction in U.S.
trade barriers shifts the U.S. demand for
Canadian exports of this good from D?sto
D This demand shift induces a production
Jr\
response as Canadian production increases
from Q t to QF In addition, price declines
.
from PT to PF
.

Following Hill and Whalley (1985), Canadian
producers are assumed to price their good at
its average cost to highlight graphically the
welfare consequences stemming from tariff
reductions in the case of economies of scale.6
With average cost pricing, the price and out
put levels before the agreement are deter
mined by the intersection of the average cost

Even though Canadian producers will not
generate increased profits, there are net
benefits for Canada. Canadian consumers
benefit from increased consumption stem
ming from the price (cost) savings associated
with economies of scale. This increase in con
sumer’s surplus is represented by the area
PfPtVY.

6Theory suggests, however, that if there really are in
creasing returns to scale at all levels of production, the
result would be a monopolist who produces at an outTo provide a more stable business environ
ment, the agreement set up a binational panel
to settle disputes in anti-dumping and counter
vailing duty cases. Precisely how this panel will
function, however, is unknown; moreover, it is
virtually impossible to quantify the value of
trade currently foregone because of these legal
and political uncertainties that the agreement
will reduce.1
6
A second feature of the agreement that can
not be quantified easily involves the consequen
ces of the liberalized trade in services. Since
models typically view the service sector as pro
ducing a non-traded service, they do not ana
lyze explicidy the trade in services. Even if
models allowed for trade in services, however,
translating policies that discriminate against
trade in services into tariff measures would be
extremely difficult. Finally, it is hoped, especial
ly in the United States, that the U.S.-Canada
agreement of national treatment for service pro
viders will encourage the current GATT negotia
tions to reach a similar agreement in a multilat
eral context. Though such an agreement would
have far-reaching consequences, these conse

16According to the U.S. International Trade Commission
(1990), the dispute settlement process reached decisions
on two noncontroversial cases involving red raspberries
and paving equipment during 1989. Upcoming cases in

FEDERAL RESERVE BANK OF ST. LOUIS

put level for which marginal revenue equals marginal
cost and sets his price accordingly.
quences cannot be estimated in the context of
the U.S.-Canada Free Trade Agreement.
Because most economic models incorporate
the service sector, and all other sectors as well,
at aggregation levels that lump many industries
together, another problem is created. The costs
and benefits of the agreement tend to be under
stated to an unknown degree; therefore, short
of disaggregating the model, there is no way to
tell precisely how much the costs or the benefits
have been understated.
For example, using a two-digit Standard In
dustrial Classification scheme, transportation
equipment is treated as one industry. The agree
ment, however, could cause one sector of tran
sportation equipment to contract and another
sector to expand. The movement of workers
from the contracting to the expanding sector,
which entails temporary unemployment and
other costs for the affected workers, is not
captured by a model that treats transportation
equipment as a single industry.
On the other hand, this aggregation also un
derestimates the benefits of the agreement.

volving steel rails and pork are expected to be controver
sial. The amount of trade under dispute in all current
cases, which involve primarily agricultural commodities, is
less than 0.5 percent of the value of bilateral trade.

Highly aggregated models, by averaging tariff
rates across sectors within an industry, under
represent the distortions caused by tariffs. In
other words, tariffs appear to distort relative
import prices by less than they actually do.1
7
Eliminating these distortions is one source of
the gains from the agreement because produc
tion and consumption decisions no longer will
be artificially distorted. Since the elimination
of larger distortions generates larger benefits,
highly aggregated models understate the bene
fits associated with tariff reductions.
The models are also not well-suited to identify
the gains resulting from other possible effects
of the agreement. Positive effects can result
from the increased competition stemming from
reductions in trade barriers. When firms are in
sulated from competition, they may not mini
mize their production costs.1 When owners are
8
separated from managers, the absence of com
petitive pressures may allow managers to incur
costs to achieve their own interests at the ex
pense of reduced profits. The increase in com
petitive pressures from increased international
trade increases the probability that production
costs will be minimized.
In addition, these firms might be pressured in
to additional research and development expen
ditures that generate either new products or
cost-saving production processes. The enlarge
ment of the market might also attract new in
vestment from non-partner countries. These
possibilities, which are potentially significant,
tend to be ignored by quantitative trade models.
Finally, all models share one other quantifica
tion problem: how to incorporate the elimina
tion of non-tariff barriers into the analysis. This
poses a problem because non-tariff barriers
must first be identified and, then, converted in
to their tariff-equivalents.1 Even if a non-tariff
9

17A simple example can illustrate this argument. Assume
two import-competing industries, each protected by an
average 5 percent tariff rate; thus, the relative price distor
tions caused by the tariffs in each industry appear similar.
For one industry, however, the average 5 percent tariff
rate results from a 5 percent tariff rate for each sector. For
the other industry, the average results from averaging a 10
percent rate and a zero percent rate. The latter industry
has more distortions that the former because the differen
tial tariff rates distort the relative prices across sectors.
The effect of aggregation is to treat every sector within an
industry as if it had the same level of protection from im
port competition. Consequently, the gains from eliminating
the distortions within an industry are ignored. The higher
the level of aggregation, the more these distortions within
industries are ignored.

barrier can be identified, it is often difficult to
calculate its tariff-equivalent accurately. For ex
ample, government procurement policies, a wellknown non-tariff barrier, are not easily con
verted into an equivalent tariffs. As a result,
estimates of the effects of eliminating trade bar
riers typically omit some non-tariff barriers. In
addition, estimates based on non-tariff barriers
must be viewed cautiously. Indeed, some re
searchers ignore non-tariff barriers entirely and
simply report the consequences of eliminating
tariffs alone.

EMPIRICAL ESTIMATES
Models based on either the perfect competi
tion/comparative advantage or the imperfect
competition approaches have been estimated to
identify the welfare consequences of eliminating
trade barriers in the agreement. To make the
discussion of the agreement’s overall effect on
the United States and Canada manageable, the
results of five recent studies are examined:
Hamilton and Whalley (1985), Brown and Stern
(1987 and 1989), Cox and Harris (1986) and
Wigle (1988). As table 3 shows, these studies ex
emplify the different approaches and yield con
flicting results. The studies by Hamilton and
Whalley, and Brown and Stern (1987) are based
on comparative advantage, while those by Cox
and Harris, Wigle, Brown and Stern (1989) are
based on imperfect competition.
The results conflict in terms of the gainers
and losers as well as the magnitudes of these
gains and losses. Depending on which study is
used, the results show both the United States
and Canada gaining, the United States losing
and Canada gaining or the United States gaining
and Canada losing. Relative to each country's
gross domestic product, the welfare conse
quences for the United States range from -0.03

18This possibility, termed X-efficiency, has been stressed in
a more general context by Leibenstein (1980).
19The tariff-equivalent of a specific non-tariff barrier is the
tariff rate that would generate the same effect on the price
of the imported good as the non-tariff barrier. To illustrate,
assume first that world prices of some imported good are
fixed. Like a tariff, a non-tariff barrier causes the price of
the imported good to rise. A non-tariff barrier, such as a
quota, causes this price increase by reducing the supply
of imports. The percentage increase in the price of the im
ported good is the tariff-equivalent. For many non-tariff
barriers, it is difficult to quantify the supply-reducing con
sequences of the barriers.

SEPTEMBER/OCTOBER 1990

Table 3
Welfare Consequences of the U.S.-Canada Free Trade
Agreement, Selected Studies
Model type

Trade
barriers removed

United States1

Perfect competition/
comparative advantage

All
Tariffs

0.03%
-0 .0 3

Perfect competition/
comparative advantage

All

0.04

Imperfect competition/
economies of scale

All

—

Imperfect competition/
economies of scale

Tariffs

0.06

-0 .0 5

Imperfect competition

Tarrifs

0.09

1.00

Authors

Hamilton/Whalley
(1985)

Brown/Stern
Cox/Harris
Wigle

(1987)

(1986)

(1988)

Brown/Stern

(1989)

Canada1
0.63%
0.54
-0 .3 5

8.74

’ The value of the welfare effect in each country, calculated by an equivalent income variation, is ex
pressed as a percentage of gross domestic product.
percent to 0.09 percent and for Canada, from
-0.35 percent to 8.74 percent.2
0
Whether these changes seem large or small
depends on your point of view. Hamilton and
Whalley (1985), for example, found that the
elimination of tariff and non-tariff barriers in
creased U.S. welfare by only 0.03 percent of
gross domestic product in 1977, which may
seem trivial. In dollar terms, however, the effect
could be called substantial. Using 1977 prices,
this rise in welfare is $0.6 billion. Using 1989
prices, the rise is $1.1 billion.
The conflicting and substantially divergent
results from these studies are due to a variety
of reasons. One key reason is that the values
chosen for the elasticities of substitution be
tween capital and labor and between different
consumer goods vary across these studies. Table
4 lists several other characteristics that could
explain the conflicting findings. These include
the numbers of countries and goods, production
functions, market structure, how prices are set
and the mobility of resources. Some insights in
to the importance of these differences are pro
vided below.

20Gross domestic product is gross national product less net
property income from abroad. In 1987, Canadian gross
domestic product was approximately 3 percent larger than
gross national product, while U.S. gross domestic product
was less than 1 percent smaller than gross national
product.
RESERVE BANK OF ST. LOUIS
FEDERAL

Perfect Competition/Comparative
Advantage M odels
Hamilton and Whalley’s (1985) model departs
from the textbook comparative-advantage model
using the Heckscher-Ohlin approach in two
ways. First, demand and production function
parameters differ across countries and, second,
products are heterogeneous rather than homo
geneous across countries.2
1
The differences in demand and production
function parameters across countries require
specific assumptions about the structure of
substitution possibilities for both demand and
production. The assumed values for the elastic
ities of substitution in these functions
determine the price elasticities associated with
goods and factors of production. In turn, since
products are differentiated by the country in
which they are produced and their physical
characteristics, these elasticities of substitution
determine import and export demand elasticities
for each country as well. Not surprisingly, the
values chosen for these substitution elasticities
determine the results generated by the model,

21The disaggregation of imports of a product according to
their places of origin is called the Armington assumption
For details, see Armington (1969).

Table 4
Summary of Model Characteristics
Characteristic

Number of countries
Number of goods
Traded
Non-traded
Returns to scale
United States
Canada

Market structure
United States
Canada

Price setting by
noncompetitive firms
Labor mobility
Across industries
Internationally
Capital mobility
Across industries
Internationally

Hamilton/Whalley Brown/Stern
(1985)
(1987)

Wigle
(1988)

Brown/Stern
(1989)

5
1

22
7

29
0

22
7

Constant
Constant

Not modeled
Increasingmanufacturing
Constantnonmanufacturing

Constant
Increasingmanufacturing
Constantnonmanufacturing

Perfect
competition
Perfect
competition

Not modeled

None

Yes
No

Yes
Yes

Yes
No

and different assumptions yield quite different
results.
The assumption of national product differen
tiation enables the model to generate intra
industry, bilateral trade flows since each coun
try exports (and imports) differentiated pro
ducts. Welfare analysis is complicated beyond
the calculation of the efficiency gains and losses
stemming from trade creation and trade diver
sion. Terms-of-trade changes occur because all
countries in the model can have some control
over their export prices by changing their tar
iffs on imports.
For example, a Canadian tariff reduction on
imports from the United States, which lowers
the price of these goods for Canadian consum
ers (but not for Canada as a whole), will cause

Cox/Harris
(1986)

Imperfect
competitionmanufacturing
Perfect
competitionnonmanufacturing
Monopolistic
competition
and collusive

Constant
Constant

Perfect
Imperfect and
competition
perfect competition
Imperfect
Imperfect and
competitionperfect competition
manufacturing
Perfect
competitionnonmanufacturing
Monopolistic
Monopolistic
competition
competition
and collusive

Canadian consumers to substitute imports from
the United States for some goods produced
domestically. The reduction in demand for
Canadian-produced goods causes a fall in the
price of these goods both in Canada and the
world market. C eteris p aribu s, Canada is harm
ed by this decline in the price of Canadian ex
ports relative to the price of Canadian imports;
this decline produces an adverse terms-of-trade
effect.
The welfare consequences, however, are not
limited to the preceding terms-of-trade effect,
because other things are changing as well.
There are efficiency gains that benefit Canada
associated with trade creation between the
United States and Canada. In addition, the cor
responding tariff reduction on U.S. imports
from Canada allows for increased Canadian

SEPTEMBER/OCTOBER 1990

sales in the U.S. market. The resulting increase
in U.S. demand for some Canadian-produced
goods will increase the price of these goods in
Canada and the world market; this is another
change that benefits Canada. Thus, there are
beneficial and adverse terms-of-trade effects oc
curring simultaneously.
The net welfare consequences for Canada and
the United States depend on the relative impor
tance of these effects. From the Canadian per
spective, the adverse terms-of-trade effects are
larger if Canada’s tariffs on U.S. exports are
higher on average than U.S. tariffs on Canadian
exports. This is, in fact, what is shown in table
2. Canadian terms of trade will also decline the
more (less) similar U.S. and Canadian goods are
to Canadian (U.S.) consumers.
Although Canada's tariffs on U.S. exports be
fore the agreement were higher on average
than U.S. tariffs on Canadian exports, Hamilton
and Whalley still find that the gains from the
agreement primarily accrue to Canada (see table
3). They attribute this finding to the fact that
Canada is the smaller partner. The smaller part
ner’s production and consumption behavior are
less likely to affect world prices, enabling it to
take greater advantage of the trade diversion ef
fects in the larger region.
Yet, using a model with features similar to
that used by Hamilton and Whalley, Brown and
Stern (1987) found that the United States gain
ed, but Canada lost. As shown in table 3, the
bilateral removal of trade barriers by Canada
and the United States leads to an increase in
U.S. welfare, 0.04 percent of gross domestic
product, but a decrease in Canadian welfare,
0.35 percent of gross domestic product. Brown
and Stern argue that the reduction in Canadian
welfare stems from the relatively higher Cana
dian tariff rate prior to the agreement. The
removal of this protection, which causes Cana
dian consumers to substitute imported goods
from the United States for Canadian-produced
goods, leads to a reduction in the relative price
of Canadian goods.
Why do these two studies differ so much with
respect to the outcome for Canada? Brown and

22The model is not a complete general equilibrium model.
The two sectors “ foreign” to Canada, the United States
and all other countries in the rest of the world, are sum
marized by exogenous import prices and a set of export
demand functions.
FEDERAL
RESERVE BANK OF ST. LOUIS

Stern found their results were sensitive to the
assumptions about the elasticity of substitution
among imports from various sources. In other
words, the results were sensitive to the degree
of substitutability between U.S. imports from
Canada and the rest of the world and between
Canadian imports from the United States and
the rest of the world.
The greater the degree of substitutability, the
larger the U.S. gain and the smaller the Cana
dian loss. If imports from various sources are
close substitutes, the preferential tariff reduc
tion induces a substitution from third-country
suppliers to the partner. Relatively speaking, lit
tle substitution out of the domestically produced
good occurs. As the demand for output from
the third-party countries declines, the terms of
trade for both Canada and the United States im
proves. Even for very high assumed values for
the elasticity of substitution among imports,
however, Brown and Stern found a decline in
Canadian welfare.
Brown and Stern also examined whether the
decline in Canadian welfare was associated with
a movement of capital from Canada to the United
States; this possibility could not occur in Hamil
ton and Whalley’s model. While Brown and Stern
did find a capital movement from Canada to the
United States, the Canadian welfare loss is near
ly invariant to different assumptions about the
sensitivity of capital flows to U.S. and Canadian
rate-of-return differences.

M odels with Im perfect Competition
and Econom ies o f Scale
Models based on comparative advantage as
sume that all markets are perfectly competitive.
This assumption is inaccurate for many markets
in Canada. To address this issue and others, Cox
and Harris (1985) developed a general equili
brium model of the Canadian economy that in
corporates both economies of scale and im
perfect competition.2 In a later paper, Cox and
2
Harris (1986) present estimates of economic ef
fects of the Free Trade Agreement.
Wonnacott (1987) notes that the difference be
tween many U.S. and Canadian manufacturing

operations has been a research topic for Cana
dian economists since the mid-1960s. Canadian
manufacturers, especially those producing
consumer durables, have tended to produce a
wide range of products, each in relatively small
quantity. The standard explanation is that diver
sified, small-scale production is caused by the
trade barriers of both countries.
Canadian trade barriers, by protecting domes
tic producers from foreign competition, enable
Canadian firms to produce a variety of products
profitably, even though these products are ex
pensive by international standards. Meanwhile,
U.S. trade barriers restrict Canadian access to
the U.S. market and, in turn, provide an incen
tive for Canadian producers to focus on the
Canadian market.2 Thus, the reduction of tariff
3
barriers in the agreement should lead to ex
panded production with lower per-unit costs.
The gains stemming from these changes are
called rationalization gains.
Cox and Harris’ modeling innovation was to
incorporate economies of scale into the analysis.
Production in each manufacturing industry is
assumed to be characterized by increasing re
turns to scale, which results in lower per unit
average production costs as output increases.
Production in each non-manufacturing industry
is assumed to be characterized by constant re
turns to scale.
Since non-competitive firms are price-searchers
and, hence, set their prices to maximize their
profits, assumptions about price-setting are re
quired. Two price-setting hypotheses are used.
One is a monopolistic competitive pricing hy
pothesis in which profit-maximizing firms set
the price of their products as a given mark-up
over their marginal cost of production. The size
of the mark-up depends on the price elasticity
of demand. T h e secon d hypothesis relies on a
collusive model in which all firms set their prices
equal to the world price plus the tariff.2
4
Cox and Harris combine these hypotheses by
assuming that the actual prices are a weighted
average of the monopolistically competitive and
collusive prices. The set of these weighted prices
that clears both goods and factor markets is the
equilibrium set of prices for Canadian firms.

23An additional incentive noted by Wonnacott (1987) for
diversified, small scale production is caused by Canadian
exposure to U.S. advertising that reinforces Canadian de
mand for a wide range of products.
24A fundamental problem with this assumption is that,
because firms do not make profits in this model, there is
no incentive to collude.

Irrespective of the pricing assumption, tariff
reductions increase import competition and,
thus, the prices of imported goods for Canadian
consumers tend to decline. For monopolistically
competitive firms, the increased competition
raises the elasticity of demand and, thereby,
reduces the mark-up over marginal cost. Simi
larly, the collusive price declines because it is
set equal to the world price plus the shrinking
tariff. The resulting stimulation of Canadian
consumption is accompanied by an increase in
output by Canadian firms to satisfy the zeroprofit condition.2
5
Five sets of parameters, whose specific values
are based primarily on previous estimates in
other studies, are especially important in deter
mining the equilibria.2 A set of export price
6
elasticities for Canadian firms is one set of para
meters. The removal of U.S. tariffs on imports
from Canada eliminates the difference between
prices paid by U.S. consumers and prices re
ceived by Canadian exporters. The removal of
U.S. tariffs tends not only to lower the prices
for U.S. consumers of imported goods from
Canada, but also to increase the prices received
by Canadian exporters. The extent of Canadian
penetration of the U.S. market depends on the
responsiveness of Canadian exports to these
price increases. Conversely, a set of Canadian
import price elasticities is needed to assess the
consequences of the reduction of Canadian
tariffs.
As suggested above, an assumption about the
weighting parameter is necessary to determine
whether prices tend to be set more according
to monopolistic or collusive behavior. The pricesetting behavior influences the degree of the
reallocation of productive resources that the
agreement causes.
Estimates for the elasticity of the average cost
curves for the Canadian manufacturing indus
tries are used. These parameters play a key role
in determining numerous results such as the
potential gains from the reallocation of produc
tive resources and the degree of increased sales
in U.S. markets.

25This general description is only suggestive of the general
tendency for firm output to expand. With Canadian
resources fixed and numerous relative price changes, the
output of each and every firm will not have risen when the
new equilibrium is attained.
26Additional details on the calibration of the model can be
found in Cox and Harris (1985).

SEPTEMBER/OCTOBER 1990

A fifth set of parameters is the trade policy
parameters. Foreign and Canadian tariffs, as
well as tariff equivalents of some non-tariff bar
riers, are used.
Cox and Harris (1986) estimate that the elimi
nation of all barriers to bilateral trade results in
Canadian welfare gains of 8.74 percent relative
to its gross domestic product (see table 3). They
argue that this large gain is due to the preferen
tial access to the U.S. market that Canadian pro
ducers will receive. As a small country, Canada
benefits because the source of U.S. imports is
diverted from other countries to Canada.2 The
7
Canadian benefits of this diversion are magni
fied because of the assumed economies of scale
in Canadian manufacturing.
Another study of the agreement that incor
porated scale economies was done by Wigle
(1988). Despite incorporating similar features to
those used by Cox and Harris, Wigle did not
find large Canadian gains. As shown in table 3,
Wigle estimates that the bilateral abolition of
tariffs produces a slight reduction in Canadian
welfare, 0.05 percent relative to its gross
domestic product. Meanwhile, U.S. welfare in
creased by 0.06 percent relative to its gross
domestic product. The sharp contrast between
his results and those of Cox and Harris caused
Wigle to explore the specific features of the two
models that were responsible for the very dif
ferent conclusions.
Cox and Harris combined the two assumptions
about price-setting behavior—monopolistic com
petitive pricing and collusive pricing—by assum
ing that prices are set as a weighted average of
the prices set by these methods. Wigle, on the
other hand, assumed that all firms in the non
mechanical manufacturing sector used monopo
listic competitive pricing, while firms in the
equipment and vehicles sector used collusive
pricing. According to Wigle, the differences in
the pricing assumptions are negligible, and
changes in them did not eliminate the difference
in his and Cox and Harris’ results.
Differences in the assumed values of trade
elasticities, on the other hand, can produce sub
stantively different results. Since Cox and Harris

27With bilateral free trade, the proportion of total Canadian
trade accounted for by the United States rises from 71
percent to 76 percent. The volume of Canadian trade with
the United States increases by more than 97 percent.
28Wonnacott (1987) notes that Cox and Harris did not make
independent estimates of the economies of scale, but
RESERVE BANK OF ST. LOUIS
FEDERAL

used much higher price elasticities of export
supply and import demand for both Canada and
the United States, Wigle reestimated his model
using trade elasticities that were approximately
twice as high as he used originally. In addition,
he introduced capital mobility between Canada
and the United States; this feature was used by
Cox and Harris. These features did make Wigle’s
amended results closer to those of Cox and
Harris; however, the latter's results were still
twice as large as Wigle’s new ones.
Wigle speculates that the rest of the differ
ence between the results in these two studies
are produced by two other factors. First, Cox
and Harris assumed higher values for econo
mies of scale; the greater economies of scale in
crease the welfare gains because the economy
will become more specialized.2 Second, Wigle us
8
ed two manufacturing industries in his model,
while Cox and Harris used 20. Disaggregated
models may generate larger efficiency gains than
more aggregated ones, because there is more
scope for reallocating resources across industries.
While both the Cox and Harris and the Wigle
models stress the role of economies of scale,
Brown and Stern (1989) estimated a model with
imperfectly competitive industries, but without
economies of scale. Their model addresses criti
cisms that can be raised about previous models.
Brown and Stern ruled out economies of scale
in their model because they doubted its signifi
cance for Canada. Canadian firms, because of
already low U.S. tariffs, had access to the U.S.
market. Thus, they argue, gains from the inter
industry reallocation of resources are likely to
be more important than intra-industry changes.
The Cox and Harris and Wigle models also in
corporated collusive pricing. Given this assump
tion, trade liberalization causes increased output
per firm. Brown and Stern suggest, however,
that a collusive market structure is not likely to
persist in the face of free entry. They also point
out that market structures in Canada, as well as
the United States, show much more variety than
has been assumed in previous models. As a re-

rather relied on previous estimates based on econometric
and engineering studies. He notes that many econometric
issues remain unsettled and that the engineering
estimates are likely biased upward.

suit, Brown and Stern incorporate a variety of
imperfectly competitive market structures into
their analysis that do not rely on what they
view as questionable assumptions about firm
behavior.
Like the previously discussed Brown and Stern
(1987) study, the model uses four countries and
29 industries, 22 tradeable and seven non-tradeable. Each industry is characterized by one of
five market structures: perfect competition; mo
nopolistic competition with free entry; monopo
listic competition without entry; market segmen
tation with free entry; and market segmentation
without entry.
Both perfectly competitive and monopolistically competitive industries are characterized by
product differentiation. Product differentiation
by country applies to perfectly competitive in
dustries, while products are differentiated by
firms in monopolistically competitive industries.
Perfectly competitive firms determine their
profit-maximizing output levels by setting price
equal to marginal cost, while monopolistically
competitive firms maximize profits by setting
price as a mark-up over marginal cost.
Homogeneous (that is, identical) products are
assumed for the remaining imperfectly com
petitive industries that are characterized by
market segmentation. With segmented markets
and each firm producing the same product with
in an industry, all firms selling to consumers in
a specific country must charge the same price,
though this price may vary across countries.
Each firm, assuming that output by other firms
is fixed, establishes a profit-maximizing price for
each national market.
Equilibrium in each industry is characterized
by zero profits. With free entry, the number of
firms in equilibrium assures that price equals
average total cost. For market structures with
out entry, the number of firms remains cons
tant. Equilibrium prices are determined in world
markets. Tariffs and exchange rates connect the
equilibrium prices to the prices paid by con
sumers and received by sellers in the individual
regions.
As shown in table 3, the welfare conse
quences of the bilateral elimination of tariffs are
small. Canadian welfare rises by $2 billion, which
is 1 percent of its gross domestic product in

1976. U.S. welfare rises as well, but only by
$1.6 billion, 0.09 percent of U.S. gross domestic
product in 1976.

CONCLUSION
Quantitative models produce conflicting re
sults about the economic effects of the U.S.Canada Free Trade Agreement. Results for the
United States range from small negative to small
positive effects on welfare, while results for
Canada range from small negative to large posi
tive effects. The conflicting results emerge both
from different assumptions about market struc
tures and the values of certain parameters as
sociated with supply and demand and from dif
ferences in the level of detail as to commodities
and countries. Since there is no consensus
about the "best” assumptions, and because inter
national trade theory provides no definitive con
clusion about a nation’s welfare following the
formation of a free trade area, it is important
that users of these models understand the rea
sons for their conflicting results.
Since the assumption of perfectly competitive
markets throughout all sectors of all countries
is unlikely to hold, the incorporation of im
perfectly competitive markets is a promising
development. Models that incorporate such mar
kets raise a number of problems, however, be
cause of the various pricing assumptions that
have been used and the need for characterizing
the extent of the economies of scale. These
newer models are also more sensitive to the
values chosen for the parameters than those
based on perfect competition.
Perhaps just as important, several key aspects
of the agreement are not included in these
models because they are extremely difficult to
quantify. These unmeasured aspects may be
more important than the measured ones in
terms of the final outcomes. Changes in the
rules and procedures governing international
trade and investment can yield large benefits
that are not included in these models. For ex
ample, many Canadians believe that secure ac
cess to the U.S. market is essential for Canadian
economic prosperity. Consequently, the Cana
dian assessment of the desirability of the agree
ment might hinge on whether or not the agree
ment provides this security. Analogously, the
precedent-setting aspects of the agreement con
cerning services are likely to influence the U.S.
assessment of the benefits of this agreement.

SEPTEMBER/OCTOBER 1990

Thus, quantitative estimates derived from models
are simply some of the many pieces of informa
tion that are useful in the decision process and,
in some cases, may not represent the most im
portant pieces.
Quantitative trade models have improved sub
stantially in recent years. Nevertheless, as this
review points out, let the user beware.

REFERENCES
Armington, Paul S. “A Theory of Demand for Products Dis
tinguished by Place of Production,” International Monetary
Fund Staff Papers (March 1969), pp. 159-76.
Brown, Drusilla K., and Robert M. Stern. “Computable Gen
eral Equilibrium Estimates of the Gains from U.S.-Canadian
Trade Liberalization,” in David Greenaway, Thomas Hyclak
and Robert J. Thornton, eds., Economic Aspects of
Regional Trading Arrangements (New York University Press,
1989), pp. 69-108.
________“A Modeling Perspective,” in Robert M. Stern,
Philip H. Trezise and John Whalley, eds., Perspectives on a
U.S.-Canadian Free Trade Agreement (Brookings Institution,
1987), pp. 155-90.
Carter, Colin, Jeffrey Karrenbrock, and William Wilson.
“ Freer Trade in the North American Beer and Flour
Markets,” in Andrew Schmitz, ed., Free Trade and
Agricultural Diversification: Canada and the United States
(Westview Press, 1989), pp. 139-86.
Copeland, Brian R. “ Of Mice and Elephants: The CanadaU.S. Free Trade Agreement,” Contemporary Policy Issues
(July 1989), pp. 42-60.
Council of Economic Advisers. Economic Report of the Presi
dent (U.S. Government Printing Office, February 1988).
Coughlin, Cletus C., K. Alec Chrystal, and Geoffrey E. Wood.
“ Protectionist Trade Policies: A Survey of Theory, Evidence
and Rationale,” this Review (January/February 1988), pp.
12-29.
Cox, David, and Richard G. Harris. “ Trade Liberalization and
Industrial Organization: Some Estimates for Canada,” Jour
nal of Political Economy (February 1985), pp. 115-45.
________“A Quantitative Assessment of the Economic Im
pact on Canada of Sectoral Free Trade with the United
States,” Canadian Journal of Economics (August 1986),
pp. 377-94.
“ Erasing the 49th Parallel,” The Economist, October 22-28,
1988.

FEDERAL
RESERVE BANK OF ST. LOUIS

Ethier, Wilfred J. “ Decreasing Costs in International Trade
and Frank Graham’s Argument for Protection,”
Econometrica (September 1982), pp. 1243-68.
________Modem International Economics (W.W. Norton,
1988).
Hamilton, Bob, and John Whalley. “ Geographically Discrimi
natory Trade Arrangements,” Review of Economics and
Statistics (August 1985), pp. 446-55.
Henderson, James M., and Richard E. Quandt. Microeco
nomic Theory: A Mathematical Approach (McGraw-Hill,
1980).
Hill, Roderick, and John Whalley. “ Canada-U.S. Free Trade:
An Introduction,” in John Whalley and Roderick Hill, eds.,
Canada-United States Free Trade (University of Toronto
Press, 1985), pp. 1-42.
Lea, Sperry. “A Historical Perspective,” in Robert M. Stern,
Philip H. Trezise and John Whalley, eds., Perspectives on a
U.S.-Canadian Free Trade Agreement (Brookings Institution,
1987), pp. 11-30.
Leibenstein, Harvey. Beyond Economic Man (Harvard Univer
sity Press, 1980).
Little, Jane Sneddon. “At Stake in the U.S.-Canada Free
Trade Agreement: Modest Gains or a Significant Setback,”
New England Economic Review (May/June 1988), pp. 3-20.
Markusen, James R., and James R. Melvin. “ Trade, Factor
Prices, and the Gains from Trade with Increasing Returns
to Scale,” Canadian Journal of Economics (August 1981),
pp. 450-69.
Shea, Brian F. “The Canada-United States Free Trade Area
Agreement,” Economic Discussion Paper 28, U.S. Depart
ment of Labor, March 1988.
Shoven, John B., and John Whalley. “Applied GeneralEquilibrium Models of Taxation and International Trade: An
Introduction and Survey,” Journal of Economic Literature
(September 1984), pp. 1007-51.
Trezise, Philip H. “ U.S.-Canadian Free Trade: An Idea
Whose Time Has Come?” in Robert M. Stern, Philip H.
Trezise and John Whalley, eds., Perspectives on a U.S.Canadian Free Trade Agreement (Brookings Institution,
1987), pp. 1-10.
U.S. International Trade Commission. “ U.S.-Canada Free
Trade Agreement: After One Year, How Does It Shape
Up?” International Economic Review (March 1990), p. 6.
U.S. Department of Commerce, International Trade Ad
ministration. Summary of the U.S.-Canada Free Trade Agree
ment (February 1988).
Viner, Jacob. The Customs Union Issue (Carnegie Endow
ment for International Peace, 1950).
Wigle, Randall. “ General Equilibrium Evaluation of CanadaU.S. Trade Liberalization in a Global Context,” Canadian
Journal of Economics (August 1988), pp. 539-64.
Wonnacott, Paul. The United States and Canada: The Quest for
Free Trade (Institute for International Economics, March 1987).

Michael T. Belongia

Michael T. Belongia is an assistant vice president at the
Federal Reserve Bank of St. Louis. Lynn Dietrich provided
research assistance.

Monetary Policy on the 75th
Anniversary of the Federal
Reserve System: Summary of
a Conference Proceedings

T
h e f e d e r a l reserve b a n k o f st . l o u is
held its fourteenth annual Economic Policy Con
ference on October 19-20, 1989. Nearly 75 years
since November 16, 1914, when it and the other
regional banks of the Federal Reserve System
opened for business, the Bank chose this occa
sion to review the recent monetary policy ex
perience in the United States. The papers and
discussion below offer an overview of the con
ference proceedings that will be published later
this year.1

FEDERAL RESERVE
PERFORMANCE SINCE 1964:
A CRITIQUE
Allan H. Meltzer, University Professor and
John M. Olin Professor of Political Economy at
Carnegie Mellon University, presented the con
ference's main paper, a critique of Federal Re
serve performance since 1964. Because extensive
surveys by Friedman and Schwartz (1963) and

’ Copies of these proceedings may be obtained by writing to
the publisher. Please direct inquiries to: Kluwer Academic
Publishers, 101 Philip Drive, Norwell, MA 02061, Attn:

Brunner and Meltzer (1964) thoroughly examined
the policy record of the System’s first 50 years,
Meltzer chose to analyze recent Federal Reserve
performance. Overall, he argued that the Fed’s
performance has been poor and that whether
the public has benefitted from the existence of
the Federal Reserve System is questionable.
His reasons for this assessment were several.
Perhaps foremost among these has been the
Federal Open Market Committee’s (FOMC’s)
focus on money market conditions and free re
serves (member bank excess reserves minus
member bank discount window borrowings) as
indicators of monetary policy’s restraint or
stimulus. This focus, in the view of Meltzer and
many others, inevitably translates into an objec
tive for the federal funds rate that creates two
systematic policy problems. The first problem is
that short-term interest rates often give an in
correct view of monetary policy’s thrust. For ex
ample, by interpreting rising nominal interest

Customer Service, or contact the Kluwer order department
at: (617) 871-6300.

SEPTEMBER/OCTOBER 1990

rates as an indication of “tight” monetary policy
when, in fact, they merely reflect the infla
tionary consequences of excessive past money
growth, the Fed gets a mistaken signal to ease;
falling nominal interest rates provide the op
posite (and still incorrect) signal to tighten mon
etary policy.
By reading market conditions incorrectly,
Meltzer argued, the Fed’s approach creates a
second problem: procyclical patterns in money
growth. To support his argument Meltzer cited
several episodes when monetary policy was
restrictive when viewed from a fed funds rate
perspective, but expansionary when viewed
from a money growth perspective. In Meltzer's
view, these procyclical policy actions caused or
contributed importantly both to the rising and
permanent inflation of the last 25 years and to
every recession of the postwar period.
Meltzer also argued that the FOMC's reliance
on forecasts of future economic activity to imple
ment current monetary policy has led to serious
policy mistakes. The reason is that the forecast
errors made by the Board of Governors' staff
are so large that its analysis cannot distinguish
whether in the current quarter the economy is
in a boom or a recession. Moreover, the fore
casts appear to be biased, consistently under
estimating future inflation. This is not to say, as
Meltzer was careful to note, that the Board
staff’s forecasts demonstrate inferior perfor
mance on its part; to the contrary, many studies
have shown these forecasts to be at least as ac
curate as any available alternative. Instead, even
if the Board’s forecasts reflect the highest stan
dards of the economics profession, the wide
range of forecast errors can mislead policy
makers into making policy changes in the wrong
direction.
Among other criticisms of Federal Reserve
performance discussed by Meltzer are vague
and changing policy statements that confuse
market participants about the course of policy,
mistakes associated with implementation of the
FOMC’s "monetarist experiment” of 1979-82 and
the return to an interest rate objective for policy
since 1982 disguised as an objective for borrow
ings from the discount window.
To improve future Federal Reserve perfor
mance, Meltzer advocates the adoption of a rule
for monetary policy with explicit targets for
growth of the monetary base. Should these tar
gets be missed, the Board would be required to
explain the reasons for its failure to achieve

FEDERAL
RESERVE BANK OF ST. LOUIS

them. In another change from the current struc
ture, the President would be given the option of
accepting the Board’s explanation or requiring
the resignation of the members.

Comments on M eltzer's O verview
Because Meltzer’s paper covered a broad
range of topics, three discussants were asked to
comment on specific pieces of his analysis. Jef
frey A. Miron, University of Michigan, in
vestigated whether the Federal Reserve has con
tributed to economic stability by reviewing a
consistent data set since 1870. Over intervals of
10 years or longer, Miron’s analysis concluded
that real economic growth has been nearly a
full percentage point lower in the post-World
War II period compared with 1870-1913; even
more striking was the contrast of price stability
(actually, a slight average annual decline in the
price level) in the period before 1913 with the
4.4 percent average annual inflation rate since
1947. Finally, comparisons of standard devia
tions showed little change in the variability of
output over time. Overall, Miron’s analysis sup
ported Meltzer’s arguments about monetary
policy actions causing the extended peacetime
inflation, but contradicted the argument that
the Federal Reserve had reduced output vari
ability significantly. Miron argued, however,
that finding the Fed to be associated with only
small reductions in output variability should not
be surprising since its emphasis has been on
smoothing financial market conditions (interest
rates) in ways and at times that often exacer
bate output variations.
K. Alec Chrystal, City University of London,
then made comparisons between monetary
policy in the United States and the policies im
plemented by the world’s other major central
banks. He argued that Switzerland, West Ger
many and Japan, by adhering to monetary
targets in most years, have been able to achieve
reasonable price stability. Moreover, when ex
ternal pressures on exchange rates induced
these central banks to abandon their money
growth targets, the cost was higher inflation in
subsequent years. Chrystal also was critical of
efforts by the Federal Reserve and foreign cen
tral banks to influence exchange rates through
intervention activities because, in his view, it
added considerable uncertainty about the thrust
of monetary policy and economic performance
in the future. Finally, he argued that the finan
cial system in the United States undoubtedly
would be more stable if geographic restrictions

Monetary Policy on the 75th Anniversary of the
Federal Reserve System
A Conference Sponsored by the
Federal Reserve Bank of St. Louis
October 19-20, 1989
SESSION I.

An O verview o f F ed era l
R eserv e P erfo rm a n ce

"The Federal Reserve at
Seventy-Five”
Allan H. Meltzer, Carnegie
Mellon University
"Has the Fed Made a Difference?
A Comparison of Pre- and
Post-1914 Conditions”
Jeffrey A. Miron, University of
Michigan and NBER
"The Behavior of Foreign Central
Banks: Comparisons and Con
trasts with Fed Performance”
K. Alec Chrystal, The City
University of London
"The Federal Reserve Policy
Process”
Donald L. Kohn, Board of Gover
nors of the Federal Reserve
System
SESSION II. M on etary Policy O bjectives

Comment by Michelle R. Garfinkel, Federal Reserve Bank of
St. Louis
“Precommitment to Rules in
Monetary Policy”
Edmund S. Phelps, Columbia
University
Comment by Manfred J.M.
Neumann, University of Bonn
SESSION III. M oney S tock M easu rem en t A nd
T he E ffe c t O f M oney On Real
Activity

"Monitoring Monetary Ag
gregates under Risk Aversion”
William A. Barnett, University of
Texas at Austin
Comment by Julio J. Rotemberg,
Massachusetts Institute of
Technology
"Money and Business Cycles: A
Real Business Cycle
Interpretation”

"Why Does the Fed Smooth In
terest Rates?”

Charles I. Plosser, University of
Rochester and NBER

Alex Cukierman, Tel Aviv
University and Princeton
University

Comment by N. Gregory
Mankiw, Harvard University and
NBER

on banking were removed. The primary argu
ment against this sort of change, however, was
that a different microeconomic banking struc
ture could have important influences on the
macroeconomic relationships of monetary policy
(e.g. velocity growth). Although he did not offer
an explicit answer to this question, Chrystal em
phasized the importance of recognizing trade
offs between micro-efficiency gains in banking
and (perhaps) greater ambiguity about the macro
effects of central bank policies.

Donald L. Kohn, director of the Division of
Monetary Affairs at the Board of Governors of
the Federal Reserve System, was asked to pre
sent the Board's view of its performance. Kohn
acknowledged that difficulties in distinguishing
between real and nominal interest rates (among
other factors) created operational problems for
a policy procedure based on nominal interest
rates. Moreover, Kohn recognized that these
difficulties—and the accelerating rate of infla
tion they produced in the late 1970s—led to the

SEPTEMBER/OCTOBER 1990

adoption of a money stock target for policy
that, in his mind, was quite similar to the pro
posal made at the end of Meltzer's paper. But,
Kohn recalled, this procedure failed as well,
perhaps suggesting that the implementation of
monetary policy in practice is different from
that in theory.
Kohn explained why the money stock targets
were abandoned as a guide to policy in 1982 by
listing well-known arguments: the impact of
financial innovations, especially nationwide in
troduction of NOW accounts, on M l; the in
creasing interest elasticity of M l; the abrupt
shift in the trend growth rate of velocity. Al
though Kohn did not make the case directly, he
implied that many of these problems might have
been expected because the monetarist argu
ments for money stock targets were based on
empirical regularities rather than an underlying
structural theory of how monetary policy af
fects economic activity. Meltzer’s discussion of
targeting the monetary base also was rejected
for the instability in its velocity and the belief
(based on a simulation experiment by the Board
staff) that it would produce unacceptably large
variations in interest rates.
Kohn ended his discussion by summarizing
the current policy process adopted by the FOMC.
He noted that the Committee’s members monitor
the economy’s performance by using a variety
of indicators and that both interest rates and
exchange rates receive considerable attention as
variables that transmit policy actions to the
economy; the monetary aggregates continue to
receive some attention as well but more in
terms of long-run economic performance. Cur
rent procedures lead to frequent, small adjust
ments in the thrust of policy to achieve the
Committee’s stated goal of promoting economic
expansion through price stability.

A POSITIVE THEORY OF
FEDERAL RESERVE BEHAVIOR
Having debated the successes and failures of
actual Federal Reserve behavior in the first ses
sion, Alex Cukierman, Tel Aviv University, at
tempted to explain w hy the Fed had chosen to
behave in a particular manner. Specifically, he
tried to answer the question: "Why Does the
Fed Smooth Interest Rates?”
Cukierman constructed a choice-theoretic
model of Fed behavior in which the Fed was
concerned both with price stability and financial

FEDERAL
RESERVE BANK OF ST. LOUIS

market stability as policy objectives. In his mod
el, changes in bank profits affect financial stabili
ty by altering the risks of bank failures; more
over, bank profits are assumed to be negatively
correlated with interest rates. These relation
ships provide the intuitive result that the Fed
will focus relatively more on price stability
when bank profits are high (and the risk of
bank failures is low) and more on financial sta
bility when bank profits are low (and the risk
of failures is high). In practice, this systematic
switching between the two goals of monetary
policy leads to interest rate smoothing and that
such a policy of interest rate smoothing has an
inflationary bias. Finally, he noted that his theo
retical model is consistent with changes in inter
est rate behavior after the founding of the Fed
reported by Mankiw, Miron and Weil (1987).
In commenting on Cukierman’s analysis,
Michelle R. Garfinkel, Federal Reserve Bank of
St. Louis, chose to focus on ambiguities in the
definition of "financial stability" and the Federal
Reserve’s objective function that could affect
Cukierman’s qualitative results. For example,
she questioned whether maximizing bank pro
fits was a socially desirable objective for mone
tary policy in the context of the model. More
over, she argued that the Fed might have tw o
policy instruments (e.g., reserve requirements
and the growth of reserves) to achieve two of
its objectives; in this event, the tradeoff between
price stability and financial stability implied by
Cukierman need not exist. Garfinkel also argued
that Cukierman’s particular choice of a one-shot
Nash equilibrium was responsible for the infla
tionary bias in his model and that choosing
another equilibrium could eliminate this feature
of his analysis. She then presented an alternative
model in which the Fed could achieve its goal of
financial stability while simultaneously reducing
the inflationary bias.

LIMITING DISCRETION AND
ACHIEVING PRICE STABILITY
Following this positive discussion of why the
Fed has chosen to adopt various policies and
practices, Edmund S. Phelps, McVickar Pro
fessor of Political Economy at Columbia Univer
sity, addressed two normative aspects of central
bank behavior: Should its discretion be limited
by a policy rule and should price stability be its
primary (if not sole) objective? The focus of this
broad topic was narrowed and made more rele
vant to the actual policy process by limiting the

discussion to a group of concrete proposals that
have been offered to limit the Fed’s discretion
and achieve price stability. The proposals re
viewed included those by Hall (1984), McCallum
(1988), Meltzer (1984), and Melzer (1987).
Phelps raised several criticisms with the gener
al notion of policy rules to achieve price stabili
ty. Perhaps most relevant to the experience of
the 1980s is that the growth rate of velocity
must be trend-stationary. If not, shocks to veloci
ty may carry it far from its trend path without
eventually returning to it. Under these circum
stances, the targeted ranges for the monetary
base or some monetary aggregate would have
to be revised occasionally to account for shocks
to its behavior. But, in Phelps’ view, this reason
able response to dealing with actual shocks
would, in reality, give a central bank convenient
excuses for missing its announced targets; even
tually, the potential to offer e* p o s t explanations
for target misses would remove credibility from
the rule and render it ineffective as a constraint
on policy discretion.
Another criticism of a rule such as that pro
posed by McCallum is that greater stability in
the growth of, say, the monetary base, can be
expected to be associated with greater volatility
in other variables; of particular concern to
Phelps in this regard were the CPI and employ
ment. Sluggish adjustment of wages, for exam
ple, could exacerbate the negative employment
effects of a supply shock if the mechanics of a
policy rule did not permit some monetary ac
commodation of the shock. Phelps also raised
the important issue of defining “price stability”
precisely as (1) a constant inflation rate, (2) a
constant aggregate price level or (3) limiting the
variability of the price level within some narrow
band. Finally, he was concerned about the
unknown, but potentially large, start-up costs as
the transition was made from discretionary poli
cies to the implementation of the rule.
Manfred J.M. Neumann, University of Bonn,
criticized Phelps for not providing a clear
framework to evaluate each rule’s operating
characteristics relative to its final goals. Neumann
analyzed whether several proposed rules could
respond to current shocks, changes in trend
velocity or output shocks and whether any of
them would guarantee an expected inflation

rate equal to zero. Against these criteria,
Neumann found that only Meltzer’s (1984) pro
posal was consistent with each standard.
Neumann, however, questioned the need for a
coercive rule when, instead, better monetary
policy performance might be achieved from a
monetary constitution that creates "the ultraconservative central banker.” This constitution
would free the central bank from responsibility
for supporting the government’s output, em
ployment or exchange rate policies and leave it
free to concentrate only on price stability.
Moreover, the people making monetary policy
would serve for long terms and be paid salaries
comparable to those in the private sector. These
features (and several others) would create a
central bank truly independent from political
pressures, a condition that in Neumann’s view is
crucial for the achievement of price stability.

MEASUREMENT OF THE U.S.
MONEY STOCK
Like most activities, conducting monetary
policy generally will produce results that are
above or below the desired objective if the
wrong phenomenon is being analyzed. Thus,
while many observers have commented on the
“anomalous” behavior of M l and its velocity in
the 1980s, another viewpoint argues that their
“unusual” behavior stems from the fact that the
monetary aggregates, as currently constructed
and reported, do not measure the theoretical
concept that we call "money.” William A.
Barnett, University of Texas at Austin, reported
the results of his continuing attempt to con
struct a monetary aggregate consistent with
microtheoretic foundations.2
Barnett’s previous investigations of the
characteristics of a desirable monetary index
number assumed risk-neutrality on the part of
individuals. In this conference paper, he con
sidered the consequences of incorporating risk
aversion, an issue that had been raised in work
by Poterba and Rotemberg (1987). In his ex
tended framework, Barnett argued that an ag
gregate monetary variable has to be derived
from a four-step process consisting of checks
for admissibility, approximation, monitoring and
application. These steps determine which assets
can be grouped to form an aggregate, how the

2Barnett’s paper was co-authored with Melvin Hinich,
University of Texas at Austin and Piyu Yue, visiting
scholar at the Federal Reserve Bank of St. Louis.

SEPTEMBER/OCTOBER 1990

individual assets will be weighted within the ag
gregate, how closely the resulting index tracks
the theoretically ideal measure and how well
the aggregate performs in its final use.

that Botemberg’s suggested revisions to his
currency-equivalent index do indeed produce a
measure of the economic stock of money.

After deriving and applying the microeconomic
and index number principles consistent with
these guidelines, Barnett advocated the use of a
monetary measure that he calls "Theoretic M2.”
He then demonstrated that large empirical dif
ferences exist between the theoretic aggregates
and the official simple-sum aggregates reported
by the Federal Beserve. He also found differ
ences, although much smaller, between Divisia
aggregates constructed under the assumption of
risk neutrality and the theoretic aggregates that
allowed for risk aversion. Barnett concluded
that the Federal Beserve should abandon its
simple-sum measures of the money stock and
construct new measures following the statistical
practices of the Bureau of Labor Statistics and
other government agencies.

DOES MONETARY POLICY
REALLY MATTER?

Although generally in agreement with Barnett’s
argument, Julio J. Botemberg, Massachusetts In
stitute of Technology, raised several issues
regarding Barnett’s strategy. For example, he
suggested a revision in weighting the dollar
values of individual asset categories when con
structing an aggregate monetary variable. Bather
than using r^-rj,, the difference between the in
terest rate on a benchmark asset (one that
yields no liquidity services) and the i,h asset’s
own rate of return at some base period,
rb r*
Botemberg suggested using ( ' „ 1).
This revised measure has the advantage that
sensitivity to choice of a base period is elimi
nated and the addition of new assets to the ag
gregate is a straightforward operation, should
that become necessary. Perhaps more important,
however, is that a model with risk aversion will
include some measure of ex p ected returns, which
will introduce errors into the measurement of
any aggregate. But while an error in a Divisia
aggregate will persist forever, the error will ap
pear only in the period it occurs in Botemberg’s
“currency equivalent” (CE) measure. Botemberg
also suggested that a different benchmark rate
of return should be used (stock market returns
rather than the yield on a long bond) and that
further testing is required to determine which
assets legitimately can be aggregated as a group
(rather than merely re-weighting the Board’s of
ficial groupings called Ml, M2, etc.). In his reply
to Botemberg’s comments, Barnett demonstrated

FEDERAL RESERVE BANK OF ST. LOUIS

The real business cycle approach to economic
fluctuations raises the question of whether mon
etary policy has any significant effect on real
economic activity. While most economists ac
knowledge the primary, if not exclusive, rela
tionship between changes in the nominal quanti
ty of money and changes in the aggregate price
level, establishing a link between nominal mon
ey and the business cycle has long been debated
by macroeconomists. Charles I. Plosser, Univer
sity of Bochester, examined this issue; his con
clusion was that explanations of variations in
real economic activity should focus on shocks to
"tastes and technology” rather than on changes
in the money stock.
Plosser first surveyed prominent studies that
attribute an important role to the money stock
for explaining fluctuations in real output. Prima
ry among these references are those by Fried
man and Schwartz (1963) and Bomer and
Bomer (1989) who identify explicit business
cycles and discuss the monetary policy changes
that coincide with them. Plosser noted that,
while several of the important episodes clearly
reflected changes in monetary policy, these poli
cy changes were "real” as opposed to "nominal”
ones. That is, rather than a change in the nomi
nal money stock (as one often thinks of a mone
tary policy change), these episodes often involved
adjustments in reserve requirements, which are
better classified as real changes that affect rela
tive prices and, in turn, induce reallocations of
resources. Thus, Plosser argues, while previous
authors may have been correct to attribute a
particular business cycle movement to a mone
tary policy change, they were wrong to con
clude that the causal factor was a change in the
nominal stock of money. Instead, in his view, it
was a real change (such as a change in reserve
requirements) that was the cause of the eco
nomic fluctuation.
Plosser supported his argument by examining
correlations of the components of M l and M2
(the source base, the reserve adjustment magni
tude (BAM) and the appropriate multiplier) and
various deposit to currency ratios with the

growth of real output to see whether real or
nominal aspects of monetary policy changes
were more closely related to output growth. In
these simple correlations, as well as more elabo
rate VAR results, Plosser found the evidence
generally to be consistent with his view that
changes in the nominal quantity of money alone
have had relatively minor effects on real eco
nomic activity.
In his comments on Plosser’s paper, N.
Gregory Mankiw, Harvard University, generally
agreed with Plosser's specific conclusion that it
is difficult to find a significant correlation be
tween changes in nominal money and real out
put. However, Mankiw disagreed with Plosser’s
overall conclusion that money is neutral even in
the short run. Mankiw argued that, even if the
Fed were not the cause of fluctuations in out
put, monetary policy still could be very impor
tant to how the economy reacts to an exogenous
real shock. For example, if shocks to technology
caused output fluctuations, monetary policy
may be able to help the economy adjust to this
new output path and avert even larger declines
in output. Mankiw also disagreed with Plosser's
criticisms of theories that attempt to explain
why changes in money may affect output.
Mankiw argued that, while perhaps correct in
his comments on specific theories, Plosser had
overlooked a more fundamental issue—because
no economic theory is “complete,” a combina
tion of different theories may be necessary to
explain real world phenomena. Because the real
world is "messy,” then perhaps, in Mankiw’s
view, economic reasoning will inevitably be
messy as well.

REFERENCES
Brunner, Karl, and Allan H. Meltzer. "The Federal Reserve's
Attachment to the Free Reserve Concept” reprinted in Karl
Brunner and Allan H. Meltzer, Monetary Economics (Ox
ford: Basil Blackwell, 1989).
Friedman, Milton, and Anna J. Schwartz. A Monetary History
of the United States (Princeton University Press, 1963a).
________“ Money and Business Cycles,” Review of
Economics and Statistics (February 1963b), pp. 32-64.
Hall, Robert E. “ Monetary Strategy with an Elastic Price
Standard,” in Price Stability and Public Policy (Federal
Reserve Bank of Kansas City, 1984), pp. 137-59.
Mankiw, N. Gregory, Jeffrey A. Miron and David N. Weil.
“ The Adjustment of Expectations to a Change in Regime:
A Study of the Founding of the Federal Reserve,” American
Economic Review (June 1987), pp. 358-74.
McCallum, Bennett T. “ Robustness Properties of a Rule for
Monetary Policy,” Carnegie-Rochester Conference Series on
Public Policy (1988), pp. 173-203.
Meltzer, Allan H. “ Overview” in Price Stability and Public
Policy (Federal Reserve Bank of Kansas City, 1984).
Melzer, Thomas C. “A Proposal for the Adoption of Monetary
Base Growth Constraints,” mimeo, Federal Reserve Bank
of St. Louis (1987).
Poterba, James M., and Julio J. Rotemberg. “ Money in the
Utility Function: An Empirical Implementation,” in William
A. Barnett and Kenneth J. Singleton, eds., New Ap
proaches to Monetary Economics, Proceedings of the
Second International Symposium in Economic Theory and
Econometrics (Cambridge University Press, 1987),
pp. 219-40.
Romer, Christina, and David Romer. “ Does Monetary Policy
Matter: A New Test in the Spirit of Friedman and
Schwartz,” NBER Macroeconomics Annual 1989 (The MIT
Press, 1989), pp. 121-70.

SEPTEMBER/OCTOBER 1990

Federal Reserve Bank of St. Louis

Post Office Box 442
St. Louis, Missouri 63166

The Review is published six
times per year b y the Research
and Public Information
Department o f the Federal
R eserve Rank o f St. Louis.
Single-copy subscriptions are
available to the public f r e e o f
charge. Mail requests f o r
subscriptions, back issues, or
address changes to: Research
and Public Information
Department, Federal R eserve
Rank o f St. Louis, P.O. Rox 442,
St. Louis, Missouri 63166.
The views expressed are those
o f the individual authors and do
not necessarily reflect official
positions o f the Federal R eserve
Rank o f St. Louis o r the Federal
R eserve System. Articles herein
may b e reprinted provided the
source is credited. Please provide
the Rank’s Research and Public
Information Department with a
cop y o f reprinted material.

Full text of Review (Federal Reserve Bank of St. Louis) : September/October 1990

FRASER