Bruce E. Moon William J. Dixon
Dept. of Int'l Relations Dept. of Political Science
Lehigh University University of Arizona
Paper prepared for the Annual Meeting of the International
Studies Association,
Chicago, Illinois, February 21-25, 1995
The Politics of U.S. Export Flows
International trade occupies an interesting position at the
nexus of theory and policy in
international relations. The theory of international trade is
rooted in one analytic tradition, largely
prescriptive and illustrative, while the policy questions are
rooted in a completely different analytic
tradition, largely explanatory and concretely empirical. Our
ultimate goal is to bring these two facets
of international trade together to gain some insight into the
patterns evidenced in the global
distribution of U.S. exports. The present paper takes an initial,
if somewhat modest, step in that
direction by untangling the substantive implications underlying
what is probably the most widely used
approach for empirically modeling international trade flows.
Specifically, we examine the so-called
gravity model of trade in the context of American exports to 88
importing countries from 1970 to 1990. The
following analysis will show how the gravity model can mask
certain theoretically meaningful and policy-relevant features of
trade flows relating directly to import shares and import
openness. Our claim is not
to deny the utility of gravity models for some purposes but
rather to highlight the utility of decomposing
the gravity formulation. Throughout our principal question
concerns why American export success has
varied with potential trade partners and over time.
Theory and Policy in International Trade
On the one hand, trade policy questions are lodged in a
state-centric ontology of national
interactions in which governments are the major actors. The
issues involve the goals of states employment
and growth, payments balances, relative gains of power, wealth
and influence, and challenges to policy
autonomy and effectiveness posed by interdependence. The targets
of these issues are state policies
believed to distort the volume and composition of trade in ways
that differentially affect the interests
of different nations. Thus, one might reasonably expect that the
empirical theory of modern international
trade would consequently be rooted in the familiar perspectives
of foreign policy analysis.
Instead, international trade theory remains locked in the
Ricardian tradition, which focuses upon
the motivations of private economic actors facing an environment
devoid of political actors or extra-economic goals. Efforts to
introduce the political dimension have been sporadic and partial,
consisting
mostly of layering on top of traditional economic theory addenda
which marginally alter the fundamental
analysis or suggest exceptions which narrow the scope of the
theory.
The clearest example of this disjuncture is the way that the
basic data of international
trade individual transactions are aggregated for the purposes of
analysis. Sattinger (1978: 22) notes,
for example, that "economists [emphasis added] studying
international trade are typically more interested
in explaining ... which goods are exported ... than the total
magnitude of trade between two particular
countries." He further observes, "This emphasis is
natural [emphasis added], as more interesting
qualitative conclusions can be drawn about [commodity] patterns
of trade than about [partner]
magnitudes." Such an orientation retains the focus on the
sectoral or commodity composition of trade
which has informed orthodox trade theory since at least the time
of Adam Smith. The outstanding exemplar
of this tradition is the Ricardian theory of comparative
advantage that addresses the question of which
products should compose the imports and exports of a given
country. Until very recently, the literature
of international trade theory has experienced a continuing
sophistication of answers to this same basic
question, most notably with Heckscher-Ohlin's sharpening of the
concept of "national advantages" with the
advancement of a factor proportions - factor intensity model.
However, we disagree with Sattinger's contention that this way of
framing the problem of
explaining trade flows is somehow "natural". Instead,
its origins are to be found in the policy questions
which dominated the formative period of liberal trade theory,
particularly the role of the Corn Laws in
reducing British agricultural imports and manufactured exports in
the late eighteenth and early
nineteenth centuries. The identification of trade partners was a
distinctly secondary consideration,
largely because that identification was hardly problematic. The
demise of the Corn Laws (and the Imperial
preference system which pre-supposed them) predictably led to
increased British exports of manufactures
to nearly all other nations because very few foreign firms could
compete with Britain's technological
advantage. Britain's agricultural imports arrived from a variety
of sources as well since the opportunity
costs of agricultural production were higher in Britain than
anywhere else. By the middle of the
nineteenth century, it was of small concern whether continental
Europe or North America provided the bulk
of these imports.
By contrast, political scientists and policy-makers, especially
in the modern era, are less
focused on the commodity composition of trade than on its partner
composition and magnitudes. While they
are cognizant of sectoral composition because domestic politics
require an awareness of the
distributional implications of trade, the prediction or
explanation of this pattern is hardly
theoretically problematic, except at the margins. The American
case, the focus of this paper, is
illustrative. Diminished trade barriers will surely shift
American imports toward labor-intensive
products (as predicted by both Heckscher-Ohlin and recent
experience) and shift American exports toward
products benefitting from technology, intellectual resources, and
advanced service capacities. However, the partner composition of
trade is much more central to contemporary policy
controversies: the differential openness of various countries to
U.S. exports (where American competitive
advantage is assumed), the discriminatory effect of regional
arrangements like the European Union, and
the predatory export promotion practices of a small group of
nations.
At the center of all these controversies are national policies in
an era of increased management
of trade. By 1980, it was estimated that 48% of global trade was
"managed" and by all accounts that figure
has probably risen since (Spero, 1985: 122). Avowedly
discriminatory arrangements, like the EU and NAFTA,
explicitly encourage trade from some partners and discourage it
with others. Yet, formal trade theory is
largely silent on the causes and consequences of this changing
policy environment. This limited ability
to incorporate the actions of governments not only diminishes the
role of trade theory in policy debates
(beyond issuing a largely-ignored call for free trade), but also
compromises the ability of orthodox trade
theory to explain existing trade patterns. As Paul Krugman (1983:
343) has put it, "Most students of
international trade have long had at least a sneaking suspicion
that conventional models of comparative
advantage do not give an adequate account of world trade."
Indeed, beginning with the famous Leontief
(1956) paradox and continuing through the more recent efforts of
Leamer (1984), systematic empirical
attempts to confirm the basic propositions of Heckscher-Ohlin
trade theory have not fared well.
Recent efforts to fill this lacunae include alternative theories
rooted in economic ideas like
technological innovation, rent-seeking, economies of scale,
intra-firm trade, demand biases and the like
(Linder, 1961; Vernon, 1966; Krugman, 1979; Dixit, 1983; Krueger,
1974; Helpman, 1984; Yarbrough and
Yarbrough, 1990). Strategic trade theory (Krugman and Smith,
1994), places the state at center stage, but
only for a quite restricted range of trade opportunities. Few
empirical studies have employed these ideas,
none designed to suggest a general model of trade determination.
Political considerations have been
incorporated into models designed to explain patterns in trade
barriers across sectors, but they have not
been expanded to cross-national analyses (Magee et. al, 1989).
Our central question is simple: "What political, social, and
economic factors explain the choice
of trade partners by nations?". Our case is a time series of
U.S. exports to 88 nations from 1970 to 1990.
Gravity Models of Export Flows
Our starting point is the one trade model which promises a
greater ability to treat policy
questions which turn on the partner composition of trade, the
so-called "gravity model" of bilateral trade
pattern determination. Anderson (1979:106) has called it
"probably the most successful empirical trade
device of the last twenty-five years." This
"gravity" model is so-named because it is similar to
the
equation describing the attraction (i.e. gravity) between any two
bodies of given sizes at a given
distance. Its basic form stipulates the gross volume of bilateral
trade flows to be a multiplicative
function of the size of each nation's economy along with one or
more resistance terms. For example, a
gravity representation of exports from nation i to importer j at
time t (Xijt) might be specified as
[1]
where Yit and Yjt denote the economic size of i and j
respectively; a (possibly multivariate) term, Rijt,
embodies forces impeding (or facilitating) the flow of exports; a
set of exponential terms serving as
weights; and a constant, à, and error term, .
It has been noted frequently that gravity models are not well
rooted in micro-economic theory,
despite Linnemann's (1966) assertion that it is "a reduced
form from a four-equation partial equilibrium
model of export supply and import demand" (Bergstrand, 1985:
474). A number of efforts to better link them
have appeared (Anderson, 1979; Bergstrand, 1985, 1989), but they
have succeeded only in showing that
several functional forms are compatible with the
"loose" theoretical ideas which lie beneath it.
Further,
the various additions to the model have carried it away from even
these disputed micro-economic roots.
Oguledo and Macphee (1994:110) put it simply: "Despite its
widespread empirical use, the gravity equation
has been a model in search of a theory."
Nonetheless, it has a long history of empirical verification
using both cross-sectional and pooled
time-series research designs across different samples of nations
and years (Tinbergen, 1962; Bergstrand,
1985). Its popularity in empirical studies derives from both its
predictive success and its relative ease
of use. Even simple formulations routinely fit the data extremely
well, commonly with an R2 above .60 and
often much higher. The gravity model's simple form also allows
easy extensions to encompass elaborations
relevant to researchers with a variety of theoretical interests.
Political economists have focused on the
composition of the "resistance" term, Rijt, which can
be used to represent the policy influences which
impede or facilitate trade (Pollins, 1989a,b; Dixon and Moon,
1993; Gowa and Mansfield, 1993).
Moreover, the equation has desirable properties for parameter
estimation. By taking logs of both
sides, equation [1] yields an additive, linear least-squares
estimating equation with a logarithmic
transformation that removes the high skewness associated with
cross-sectional variance in national GDP.
When other variables with similar skewness have been added to the
model, such as population size (Aitkin,
1973; Pelzman, 1977; Brada and Mendez, 1983, 1985), the
logarithmic specification has been retained, even
though on theoretical grounds it is not always clear why they
should enter multiplicatively.
While acknowledging the plausibility of its simple assumptions
that bilateral trade volumes can
be expressed as a function of the size of the respective
economies moderated by "resistance" terms we,
too, have doubts about the precise form of the relationship. Our
concerns flow from the observation of
Greenaway and Milner (1986: 109) that gravity models encompass
relevant empirical phenomena "even if they
are not able to tell us precisely how they are relevant". In
fact, they purchase an excellent fit with the
data at the price of employing parameters which do not admit of
easy theoretical interpretation. In
response, other researchers have reformulated the equation in
order to highlight questions of theoretical
interest. For reasons which will shortly become clear, we follow
that course as well.
Reformulating the Gravity Model
The great predictive power of the gravity model stems from a
simple idea which requires little
theoretical grounding: large nations trade more than small
nations. By way of illustration we will
estimate the parameters of equation [1] for the annual value of
U.S. exports to 88 nations from 1970 to
1990. Both size terms are operationalized by total GNP (World
Bank, 1993). To begin we incorporate three
resistance terms commonly used in gravity models, relative export
prices, geographic distance, and a
binary variable registering the geographic contiguity of Canada
and Mexico. We take account of the effect
of American export prices by including a relative price index.
Geographic distances are calculated in
miles between the principal city of each importing country and
New York or Los Angeles, whichever is
shorter. The contiguity indicator serves as an adjustment for the
very short distances registered for
Canada and Mexico.
[Table 1 about here]
GLS-ARMA estimation results for this initial gravity model
specification are presented in the
first column of Table 1. Because terms on both sides are
estimated in natural logarithms the coefficients
are directly interpretable as point elasticities. The strongly
positive effect of importer GNP and the
strongly negative effects of geographic distance and relative
prices are consistent with the general
expectations of gravity models. The the negative coefficient of
the contiguity indicator reveals a lower
level of exports to Canada and Mexico than we might otherwise
expect based solely on distance and economic
size. The one anomaly in these results is the apparent
irrelevance of U.S. GNP. While the GNP of the
exporting country is typically a significant predictor of trade
flows in cross-sectional designs, this
is not always so for time-series (e.g., Thursby and Thursby,
1987). Studies incorporating multiple
exporting countries must necessarily control for the huge
differences in productive capacity of
exporters, though such control is unnecessary and often omitted
in studies of only a single exporter
(Pollins, 1989; Hanink, 1990). Indeed it is probably desirable to
drop it from our estimation to avoid
mispecification. The second column of Table 1 presents results
for an estimation that excludes U.S. GNP.
Clearly, dropping this term has had little noticeable effect on
the parameter estimates.
Next we move beyond the traditional emphasis of gravity models by
introducing four political
variables expected to affect U.S. export flows. These include the
dollar value of U.S. development aid,
a measure of institutionalized democracy, and binary indicators
for Latin American nations and European
Community members. Expectations for these variables are fairly
straightforward. While the motivations
for foreign aid encompass a range of state imperatives, aid
relationships are generally assumed to promote
exports by opening markets, often with explicit ties to trade.
Democracy, which has been shown to be a
significant predictor of other aspects of foreign policy
behavior, here serves as a measure of political
similarity that facilitates trade by insuring open and
predictable political institutions (Dixon and
Moon, 1993). With respect to more general foreign policy
considerations, the historical relationship
between the U.S. and Latin American nations approximates the
colonial status that has been shown to affect
the trade relations of other former colonies. Finally we would
expect relatively lower exports to EC
members reflecting the special arrangements which, though
designed partially for political and security
purposes, systematically discriminates against the trade of
non-members.
The estimates in the third column of Table 1 generally bear out
these expectations with only the
EC parameter estimate failing to exceed twice the value of its
standard error. With the exception of
distance and contiguity, which are understandably adjusted by the
Latin American indicator, the other
terms in the model are not much affected by the introduction of
these new variables.
One particularly interesting feature of all three columns of
Table 1 is that the importer GNP
estimate closely approximates unity, a result commonly found in
other gravity studies. Indeed, formal t-tests would reveal none
of these estimates to be significantly different from 1. This
result allows U.S.
to take a step that illuminates an important new research
direction. If we assume the value of parameter
2 to be exactly 1.0, we can divide both sides of equation [1] by
importer GNP (i.e., Yij). This operation
leaves U.S. with the quantity Xijt/Yjt on the left hand side and
eliminates importer GNP entirely from the
right hand side. Note that the term on the left now refers to
U.S. imports relative to importer GNP. Our
assumption that 2 equals 1 insures the mathematical equivalence
of these two forms of the specification.
Are they empirically equivalent, as well? Empirical estimates for
this alternative form are presented in
the first column of Table 2. Note that these estimates closely
approximate those in the third column of
Table 1 in every instance (with the difference in constant term
due to the use of a percentage scale). We
are thus satisfied that assuming 2 equal to 1 has little effect
on the empirical estimates.
[Table 2 about here]
The point of this exercise is revealed by observing that our
modified dependent variable measuring
U.S. imports relative to importer GNP can be decomposed into two
theoretically interesting components with
clearly defined meanings,
[2]
The first term on the right side of this expression, Xijt/Xjt,
represents the market share of j's imports
captured by I, while the second term, Xjt/Yjt, represents the
openness of j's economy to imports. Three
excellent reasons immediately appear for treating these two
components separately rather than as a
composite.
First, from the stand-point of U.S. trade policy designed to
increase American exports, this
decomposition highlights two very different potential problems.
U.S. exports may not reach expectations
because a given nation is not very open to trade, the charge
leveled at Japan and many Third World nations,
for example. The usual explanation for this, high levels of
protectionism or other distorting policies,
suggests ameliorative measures such as macro-economic reform that
might be encouraged by systemic forces
(e.g. the IMF or GATT). Alternatively, the short-fall may occur
because of discrimination against U.S.
products, a charge leveled against the EU (and other regional
arrangements) because it imposes higher
barriers against American products than those of its regional
partners. Not only is this a much more
charged political issue, but it also requires a different set of
policy instruments in response. From a
policy stand-point, therefore, it is important to distinguish the
"openness" from the "market share"
interpretation of short-falls in bilateral trade volumes.
Second, from a theoretical stand-point, these two attributes of a
national political economy are
determined by very different forces. If we wish to explain rather
than merely predict trade volumes, we
must be able to distinguish these two paths. Indeed, the
literatures which would be used to build an
explanation of them have quite different roots. Openness is known
to be a function of the size of the
economy and various political forces which effect levels of
protection. Trade share is more likely to
result from the compatibility between two political economies,
including, we have hypothesized,
institutional and foreign policy considerations.
Third, from the stand-point of estimation, some independent
variables would seem best suited to
predicting share while others are more appropriate to predicting
openness. Moreover, some factors might
well predict both openness and share but with different signs
whose countervailing impact could wash out
entirely for estimations of total export flows. That loss of
information may not harm our ability to
predict trade volumes, but it undermines our theoretical
understanding of the forces that affect trade.
Indeed, as the following analyses will show, the literature may
well contain mistaken inferences resulting
from treating trade volumes in the aggregate rather than
examining the components.
These latter two points are vividly illustrated by estimations on
share and openness reported in
columns two and three of Table 2. The right side of both
specifications is identical to that in column one
while the left side is now decomposed into logged percentage
measures of U.S. share and import openness.
Several observations can be made about these results. Note first
that U.S. development aid carries a
strong positive effect on share while having little appreciable
effect on openness. This is a satisfying
result because it is generally consistent with how aid is most
often thought to influence trade flows.
Moreover, the absence of impact on openness suggests that
aid-induced increases in U.S. shares are more
likely the result of shifting markets than of any general effort
to expand total imports. The
institutionalized democracy indicator reveals a rather similar
pattern of effects when decomposed into
share and openness elements. The small but discernable effect of
democracy seen in the earlier results on
aggregate trade flows is now revealed to be entirely due to
effects on U.S. shares; indeed, it would appear
that openness of political system has little to do with openness
of the economy. This finding is consistent
with arguments that trade is enhanced by stable and predictable
institutional arrangements (Pollins,
1989a, b; Dixon and Moon, 1993).
Equally interesting is the fact that the two indicator variables
designating EC membership and
Latin American nations actually reveal countervailing effects on
share and openness. Recall that none of
our gravity model estimations on aggregate U.S. imports managed
to detect any systematic impact of EC
membership. We now see that EC nations do have a tendency to be
more open than most importing nations,
though they favor U.S. imports neither more nor less than those
of other exporters. It would appear from
these initial results that the effects of trade creation dominate
those of trade diversion (Viner, 1950).
The Latin American region is quite a different story. Here we see
a very strong positive share estimate
indicating a substantial bias toward U.S. imports. On the other
hand, Latin America also reveals a clear
tendency to be less open on average than other U.S. markets.
Finally it is important to underscore the tentativeness of the
preceding observations regarding
the factors affecting U.S. share and import openness. Ultimately
we must construct models of share and
openness that reflect more than just the mechanical application
of a modified gravity specification.
Despite the fact that share and openness have been shown here to
be decomposable elements of the trade
flows traditionally fit to gravity models, there is no reason to
believe that the traditional gravity
specification is theoretically appropriate for the components
when taken individually. Geographic
distance and the index of relative prices aptly illustrate this
point. While we might well expect U.S.
prices relative to those of other developed country exporters to
help determine U.S. import shares, we are
at a loss as to why U.S. relative prices could have any bearing
whatsoever on openness. Note however that
the estimates in Table 2 reveal no effect of U.S. relative prices
on share and, even more curiously, a
substantial negative effect on openness. Because the relative
price index varies only over time, not over
import nation cross-sections, it seems likely that the observed
price effect on openness is confounded
with some excluded time-dependent factor, such as exchange rate
fluctuations, global business cycles, or
evolving international regimes governing trade and finance. Our
preliminary analysis, too tentative to
include in this report, has already uncovered indications of
these effects. Similarly, the observed effect
of geographic distance on openness is almost certainly confounded
with some excluded factor, though in
this case one that varies only by cross-section, not by time. In
addition, the size of the importing state,
excluded in our transformation of the gravity model, must now be
re-introduced as a known correlate of
openness.
It is the share model, however, that will benefit most from our
re-conceptualization. Our
preliminary analysis suggests that political factors associated
with foreign policy considerations
strongly affect the distribution of American exports, a pattern
more visible with the confounding effects
of openness removed.
Conclusion
These preliminary results confirm that the disaggregation of the
gravity model offers promise for
improvements in our theoretical understanding and statistical
modeling of the bilateral trade flows
between nations. Indeed, we believe that we can transform the
gravity formulation from "a model in search
of a theory" to a prediction equation that suggests two
separate analyses, each far better rooted in a
theory appropriate to its empirical determinants than the
existing literature has provided. In
particular, we are convinced that the market share of exporters
is far more dependent upon political
considerations than previously appreciated, but that to expose
these effects will require establishing
theoretical roots in the traditional concerns of foreign policy
analysis.
We close with a brief observation which hints at the policy
implications that also flow from this
treatment. It appears that we can usefully view U.S. export
levels to three groups of nations the European
Union, Latin America (particularly Mexico), and fast-growing East
Asian nations (especially
Japan) through the lens we are beginning to fashion. Further
integration in Europe is likely to accentuate
the pattern we have already seen, namely increased openness
accompanied by further bias in national
sources. This would not only operate to the detriment of U.S.
exports, but would also constitute a step
away from the multilateral and non-discriminatory vision of
Bretton Woods. A parallel movement is already
underway in the Western hemisphere with the contemplated
expansion of NAFTA into Latin America. From the
stand-point of one set of U.S. interests, this initiative rests
on the pattern of strong import bias toward
U.S. products within economies that remain relatively closed. It
seems likely that the expansion of
American exports to Latin America would be advanced more by
greater openness than by further enhancements
of U.S. market share. Our first back-of-the-envelope estimates
suggest that U.S. exports toward Latin
America could increase by about 2% of the region's total product
if the openness of these economies could
be increased to global averages without sacrificing U.S. market
share. Finally, the pattern of Japanese
imports the American market share is unusually high but the
import sector as a whole is unusually
low suggests some caution in American policy which, though
designed to open the Japanese market, may be
self-defeating if the result is damage to the political
relationship which sustains the U.S. market share. Notes
References
Aitkin, Norman D. 1973. "The Effect of EEC and EFTA on
European Trade: A Temporal Cross-section Analysis," American
Economic Review 63:881-92.
Anderson, James E. 1979. "A Theoretical Foundation for the
Gravity Equation," American Economic Review 69:106-116.
Bergstrand, Jeffrey H. 1985. "The Gravity Equation in
International Trade:Some Micro-Economic Foundations and
Empirical Evidence," The Review of Economics and Statistics
67:474-481.
Bergstrand, Jeffrey H. 1989. "The Generalized Gravity
Equation, Monopolistic Competition, and the Factor-Proportions
Theory in International Trade," Review of Economics and
Statistics 71:143-153.
Brada, Josef and Jose Mendez. 1983. "Regional Economic
Integration and the Volume of Intra-regional Trade: A
Comparison of Developed and Developing Country Experience,"
Kyklos 36:589-603.
Brada, Josef and Jose Mendez. 1985. "Economic Integration
Among Developed, Developing and Centrally Planned
Economies: A Comparative Analysis," Review of Economics and
Statistics 67:549-56.
Dixit, Avinash. 1983. "International Trade Policy for
Oligopolistic Industries," The Economic Journal 94,
supplement:1-16.
Dixon, William J. and Bruce E. Moon. 1993. "Political
Similarity and American Foreign Trade Patterns," Political
Research Quarterly 46:5-25.
Gowa, Joanne and Edward Mansfield. 1993. "Power Politics and
International Trade," American Political Science Review
87(2):408-420.
Greenaway, D. and C. Milner (1986) The Economics of
Intra-Industry Trade Oxford: Basil Blackwell.
Gurr, Ted Robert, Keith Jaggers, and Will H. Moore. 1991.
"The Transformation of the Western State: The Growth of
Democracy, Autocracy, and State Power since 1800." In On
Measuring Democracy, ed. Alex Inkeles. New
Brunswick, NJ: Transaction.
Hanink, Dean M. 1990. "Linder, Again,"
Weltwirtschaftliches Archiv 126:257-267.
Helpman, Elhanan. 1984. "A Simple Theory of International
Trade with Multinational Corporations," Journal of
Political Economy 92:451-471.
International Monetary Fund. 1994. International Financial
Statistics Yearbook 1994. Washington, 122-3.
Kmenta, Jan. 1986. Elements of Econometrics, 2nd Ed. New York:
Macmillan.
Krueger, Anne. 1974. "The Political Economy of the
Rent-Seeking Society," American Economic Review 64:291-303.
Krugman, Paul. 1979. "Increasing Returns, Monopolistic
Competition, and International Trade," Journal of
International Economics 9:469-79.
Krugman, Paul. 1983. "New Theories of Trade Among Industrial
Countries," American Economic Review 73:343-47.
Krugman, Paul and Alasdair Smith. 1994. ed. Empirical Studies of
Strategic Trade Policy. Chicago: University of
Chicago Press.
Leamer, Edward E. 1984. Sources of International Comparative
Advantage: Theory and Evidence. Cambridge: MIT Press.
Leontief, Wassily. 1956. "Factor proportions and the
Structure of American Trade," Review of Economics and
Statistics.
November: 386-407..
Linder, Staffan. 1961. An Essay on Trade and Transformation. New
York: Wiley
Linnemann, Hans (1966) An Econometric Study of International
Trade Flows Amsterdam: North-Holland Publishing Co.
Magee, Stephen, William Brock, and Leslie Young. 1989. Black Hole
Tariffs and Endogenous Policy Theory. Cambridge:
Cambridge University Press.
Oguledo, Victor and Craig Macphee. 1994. "Gravity Models: A
Reformulation and an Application to Discriminatory Trade
Arrangements," Applied Economics 26:107-120.
Pelzman, Joseph. 1977. "Trade Creation and Trade Diversion
in the Council of Mutual Economic Assistance: 1954-1970,"
American Economic Review 67:713-722.
Pollins, Brian. 1989a. "Does Trade Still Follow the
Flag" American Political Science Review 83:465-80.
Pollins, Brian. 1989b. "Conflict, Cooperation, and Commerce:
The Effect of International Political Interactions on
Bilateral Trade Flows," American Journal of Political
Science 33:737-61.
Robinson, Arthur H., Randall D. Sale, Joel L Morrison, and
Phillip C. Muehrcke. 1984. Elements of Cartography, 5th ed.
New York: Wiley.
Sapir, Andre. 1981. "Trade Benefits Under the EEC
Generalized System of Preferences," European Economic Review
15:339-355.
Sattinger, Michael. 1978. "Trade Flows and Differences
Between Countries," Atlantic Economic Journal 6:22-30.
Spero, Joan. 1985. The Politics of International Economic
Relations, 3rd ed. New York: St. Martin's Press.
Srivastava, Rajendra K. and Robert T. Green. 1986.
"Determinants of Bilateral Trade Flows," Journal of
Business
59:623-640.
Stimson, James A. . 1985. "Regression in Space and Time: A
Statistical Essay." American Journal of Political Science.
29:914-47.
Summary, Rebecca M. 1989. "A Political-Economic Model of
U.S. Bilateral Trade," Review of Economics and Statistics
71:179-182.
Thursby, Jerry G. and Marie C. Thursby. 1987. "Bilateral
Trade Flows, the Linder Hypothesis, and Exchange Risk,"
Review of Economics and Statistics 69:488-495.
Tinbergen, Jan. 1962. Shaping the World Economy: Suggestions for
an International Economic Policy. New York: Twentieth
Century Fund.
Vernon, Raymond. 1966. "International Investment and
International Trade in the Product Cycle," Quarterly Journal
of Economics 80:190-207.
Viner, Jacob. 1950. The Customs Union Issue. New York: Carnegie
Endowment for International Peace.
World Bank. 1993. World Tables of Economic and Social Indicators,
1950-1992 (Computer file). Ann Arbor, MI: Inter-University
Consortium for Political and Social Research.
Yarbrough, Beth and Robert Yarbrough. 1990. "International
Institutions and the New Economics of Organization,"
International Organization 44:235-259.
Table 1. Pooled Time-Series GLS-ARMA Estimates for
U.S. Export Flows, 1970-1990
Alternative Gravity Models
Variable Model 1 Model 2 Model 3
ln U.S. GNP -.019
(.032)
ln Importer GNP .992 .987 .987
(.014) (.013) (.011)
ln Relative price index -.671 -.671 -.697
(.074) (.074) (.075)
ln Geographic distance -1.277 -1.288 -.833
(.050) (.051) (.061)
Contiguous importer -.751 -.758 -.320
(.284) (.282) (.127)
ln U.S. aid .134
(.016)
Institutionalized democracy .012
(.004)
EC member -.027
(.048)
Latin America .666
(.058)
Constant 7.337 7.196 2.725
(.572) (.459) (.533)
GLS R2 .838 .834 .878
Standard error of estimate .843 .870 .747
Note: Entries in parenthesis are standard errors. The sample
consists of 88 importing nations over
21 years (N=1848).
Table 2. Pooled Time-Series GLS-ARMA Estimates for
U.S. Export Shares and Partner Openness, 1970-1990
U.S. Imports % U.S. Share of Partner
Variable of Partner GDP Partner Imports Openness
ln Relative price index -.691 -.026 -.537
(.075) (.057) (.056)
ln Geographic distance -.828 -.434 -.380
(.061) (.053) (.047)
Contiguous importer -.347 .805 -1.168
(.125) (.177) (.129)
ln U.S. aid .136 .110 .019
(.016) (.013) (.010)
Institutionalized democracy .010 .009 .003
(.003) (.003) (.003)
EC member -.040 -.055 .083
(.046) (.039) (.034)
Latin America .673 1.130 -.402
(.058) (.044) (.047)
Constant 2.553 5.249 6.372
(.516) (.450) (.403)
GLS R2 .374 .483 .121
Standard error of estimate .779 .770 .786
Note: Entries in parenthesis are standard errors. The sample
consists of 88 importing nations over
21 years (N=1848).