international trade: geography and inequality · introduction during the last decades international...
TRANSCRIPT
International Trade: Geography and Inequality
Jaime A. Meza-Cordero
University of Southern California
Abstract: International trade and factor mobility have significantly extended around the
whole world in the last decades. However, production factor returns show no sign of
convergence. Market sizes, cultural similarities and distance have historically been the
major determinants of trade between countries, but aspects such as closeness to big
markets don’t seem to affect income levels. This paper focuses on analyzing the role of
trade in the American Continent and its repercussions on income levels making use of the
standard gravity model. I find that market size, distance and cultural similarities have
very strong effects determining trade patterns, and I also find closeness to big markets
don’t seem to increase wages, justifying in part regional income inequality.
June 2010
Introduction
During the last decades international trade has grown constantly and thoroughly
throughout the world. Although trade barriers have decreased to an all-time low level and
factor mobility has significantly increased, international inequality shows no signs of
convergence. A possible cause of income differentials is distance to markets, for any
given productivity, the higher the transportation costs the lower the capacity to be a
competitive exporter and the more expensive it becomes to import.
Going further, the last decades have been characterized by the decrease of transportation
costs and a very quick diffusion of new long distance communication technologies. These
have led to a highly globalized economy, higher demand for foreign goods and a highly
specialized international supply. However, trade barriers such as: geographical distance
between major cities, no seashore access, no common borders, different legal traditions,
no historical partnerships, and different languages still remain.
Despite a significant increase in factor mobility and economic integration, international
wages don’t seem to show signs of convergence. My hypothesis is that even with
complete economic integration distance to big markets will always remain as a significant
cost that leads to fixing low wages in order to be internationally competitive.
I will focus on a set of 24 countries from the American Continent, specifically:
Argentina, Belize, Bolivia, Brazil, Canada, Chile, Colombia, Costa Rica, Cuba,
Dominican Republic, Ecuador, El Salvador, Guatemala, Honduras, Jamaica, Mexico,
Nicaragua, Panama, Paraguay, Peru, Trinidad & Tobago, USA, Uruguay and Venezuela.
An advantage of this list of countries is that most of them share cultural similarities (legal
traditions, institutions, language, among others), these similarities suggest that the main
factors that explain the differences in trade and incomes are due to the size of the markets
(GDP, GDP per-capita, population) and the distance between the countries, thus the
standard gravity equation would be very appropriate for explaining trading patterns.
For example, what could justify the difference in trade between Chile and Argentina or
Mexico? The three countries have basically the same GDP per-capita ($14529, $14408
and $14534 respectively), they share the same language and have very similar
institutions. Therefore I consider three key factors for explaining their trade patterns:
1. Distance: Santiago, Chile and Buenos Aires, Argentina are 1129 km away from
each other; while Santiago, Chile and Mexico D.F., Mexico are 6590 km away
from each other.
2. Common Border: Chile and Argentina share a long common border, while Chile
and Mexico are far apart from each other.
3. Political Agreements.
Another advantage of this group of countries is that they all have relatively easy access to
international trade. All but two of the countries (Bolivia and Paraguay) have seashores,
which enable them to have maritime transportation. Also, belonging to the same
continent facilitates diplomatic communication (for example: they are all part of the
Organization of American States), making it easier to trade with each other and to
elaborate trade agreements.
The final benefit of this sample of countries is that it contemplates heterogeneity as well;
it has big countries, small countries, landlocked countries, islands, countries with several
neighbors and countries with few neighbors. This heterogeneity allows us to analyze the
effects of the markets’ size, having common borders, being landlocked and most
importantly distance. A possible cause of factor remuneration differentials is distance, for
any given technologic level, the higher the transportation costs the lower the wages
offered in order to compensate and remain competitive.
Given that most of the trade gravity models are tested with data from OECD countries,
this research has the potential of being one of the first explaining current trade behaviors
and income differentials across the whole American Continent. Maybe the biggest
limitation that this study comes from the fact that variables such as historical partnerships
and political agreements are very hard to account for and measure, therefore I will use a
very geography-oriented view. Ahead I will introduce some of the existing literature on
trade and income inequality using gravity models, information on the dataset constructed,
the models and estimation methods used, and finally some concluding remarks.
Literature Revision
In international trade literature an analogy to the original gravity equation has proven to
have a strong fit with the data explaining bilateral trade flows between two countries (i to
j). Under this formulation, data on GDPs and population would indicate the size of each
market and data on distance between capital cities would be a good proxy for the costs of
transportation. If country i has a big market (high GDP) then we would expect it to
produce a big variety of goods, promoting its exports. If country j has a big market, then
we would expect it to have a high demand for foreign goods, promoting its imports. If
country i is far apart from country j we would expect transportation costs to be very high.
This analogy was first applied by Tinbergen (1962) and Pöyhöen (1963), it defines
bilateral trade as a function of the GDP of each country and the distance between them.
Anderson (1979) provided a theoretical explanation for the gravity equation applied to
the trade of commodities; he considered product differentiation by place of origin.
Bergstrand (1985) developed a world equilibrium model of world trade using the gravity
equation and based on micro foundations. Although this two micro-based formulations
gained o lot of attention in international trade literature, due to the specific purpose of this
research I will only use the standard gravity model based on macro level aggregate flows.
McCallum (1995) incorporated into the basic equation the existence of a “Border Effect“
which includes a dummy variable for adjacent countries. His argument is that trade
doesn’t depend primarily on distance. Having to go through a national border
significantly affects the quantity traded; therefore he considered very important to include
a “border” dummy variable. The dummy variable gets a value of 1 if there’s a common
border and 0 if there isn’t. A constant term θ will consider the time invariant effects like
traditional patterns of trade or world wealth, which may not be observable.
Another formulations, like in Frankel, Stein & Wei (1995) would include other variables
to define the size of the market considering GDP, population or dividing GDP by
population. Chaney (2008) altered the basic gravity equation to analyze the effects of the
elasticity of substitution (σ) on bilateral exports, such that a high elasticity of substitution
imply that goods among countries are highly homogeneous, and this would decrease the
exports:
Authors as Limao and Venables (2001) and Redding and Venables (2004), relate
transport costs to features of economic geography. They find that the median land-locked
country’s shipping costs are way higher than those of a country with seashore. Eaton and
Kortum (2002) develop a trade model linking comparative advantage and geographic
barriers using a cross section of 19 OECD countries. They show how price and factor
rewards dispersion are linked to distance between places. Their conclusion is that
geography plays a major role in trade and economic activity. Leamer (1997) finds that
Central and Eastern European countries’ differing access to Western European markets
creates differences in their potential to achieve higher standard of living, revealing the
welfare implications of having access to big markets.
Estevadeordal et al. (2008) use the gravity model to study the effects of Regional Trade
Agreements on global trade liberalization. Their goal is to determine if the pressures to
liberalize (after signing FTAs) due to trade diversions offset the effects of tariff
discrimination against non-partners. Waugh (2009) uses gravity equations to show the
asymmetry of transport costs among rich and poor countries. He rules out the standard
assumption of fixed trade costs for all the countries, then trade costs will be contingent
upon exporter. He tries to show the implications in income differences after trade cost
reductions. His conclusion is that cost asymmetries help reconcile price and quantity data
in standard trade models as well as the cross-country income differences.
Model
First I will try to identify the major geographical trading patters in the American
Continent, the model I will use is a standard trade gravity model. My biggest aim is to
assess the significance of distance as a given and permanent factor in bilateral trade. In
my gravity model I will define bilateral trade as a function of the GDP per-capita of each
country and the distance between them.
Xij = (GDPp/ci x GDPp/cj)/(Dij) (1)
Where:
Xij: Value of exports from country i to country j.
GDPp/ci: Gross Domestic Product per capita of country i.
GDPp/cj: Gross Domestic Product per capita of country j.
Dij: Distance from country i to country j.
Applying natural logarithms we obtain:
LnXij = LnGDPp/ci + LnGDPp/cj – LnDij (2)
Other important factor regards adjacent nations; hence I will include McCallum’s (1995)
Border Effect as an indicative variable. Including McCallum’s formulation I get:
Xij = θ(GDPi)α(GDPj)βe(γBorderij)/(Dij)δ (3)
The expected signs of the coefficients α and β are positive because if country i’s GDP
increases then we would expect it to have a greater amount and variety of goods to
export, if country j’s GDP increases then it’s demand increases and therefore it will
import more from country i. We should also expect δ to have a negative sign because it
becomes more costly to trade as distance among cities increase. Finally, γ should be
positive because having a common border facilitates trade between two countries.
To account for common cultural links I will use a dummy variable for common language.
I selected common language because sharing an official national language facilitate the
communication among countries and at the same time it reveals if the countries had the
same colonizer. Having the same colonizer and similar independence dates allows us to
assume that their institutions, legal system, religion, etc. must be somehow similar.
Finally, since most of international trade is done through maritime transportation, it
results very important to consider if the country has access to seashores. Therefore I will
include an indicative variable if the country is landlocked. My gravity model is:
Ln(Xij) = α0 + α1LnGDPp/ci + α2LnGDPp/cj + α3Ln(distij) + α4borderij + α5langi +
α6landlocki + uij (4)
I expect the GDP per-capita of both the exporting country as the importing country to
have a positive effect on total exports because the higher the GDP of the exporting
country, the variety and quality of tradable goods increase; and the higher the GDP of the
importing country, the bigger the demand for foreign goods. I expect distance to have a
considerable negative effect in trade, as it is a proxy for transport costs. Sharing a
common border or language should have a positive effect on bilateral trade as they
facilitate transportation and communication respectively. Finally, I expect the landlocked
coefficient to have a negative sign because it results very expensive to deliver the goods
if landlocked.
Once I have estimated the effects of distance on bilateral trade I will move on to evaluate
the opportunities given by the geographical location of each country. The methodology I
will use is the one used by Redding and Venables (2004). First, I will estimate another
gravity model for bilateral trade in order to get estimates of each country’s proximity to
export and import destinations (market access and supplier access). These geographical
measures will expose which countries have higher and lower trade potential. Once I have
found these potentials, the next step would be to establish a zero profit condition for
firms, this will implicitly tell us the maximum wage that a country can afford to pay
given its proximity to markets (we will call this the wage equation).
As in Redding and Venables (2004), I assume that individuals in country j possess a CES
utility function of the form:
Uj = [ΣiR nixij
(σ-1)/σ]σ/(σ-1) (5)
Where there are R countries, n is the variety of goods produced in country i, x are the
imports of country j from country i and σ is a constant elasticity of substitution which is
greater that 1. A price index, Gj, for manufactures in each country is defined over the
prices of individual varieties produced in country i and sold in country j, Pij, hence:
Gj = [ΣiR niPij
1-σ]1/(1-σ) (6)
Define country j’s total expenditure on manufactures as Ej. Then country j’s demand for
each product is: xij = pij-σEjGj
σ-1 (7)
Thus, the own price elasticity of demand is σ, and the term EjGjσ-1 is the demand curve
facing each firm in market j, which is called market capacity of country j. Again, from
Redding and Venables (2004), each representative firm has profits:
πi = ΣjRPijxij/Tij - Gi
αwiβvi
γci[F+xi] (8)
Where T is iceberg transport costs, w is wage, v is the price of an internationally mobile
factor, c is a productivity parameter and F are fixed costs. I assume that α+β+γ=1.
Firms set markup prices such that Pij = TijPi, where:
Pi = Giαwi
βviγciσ/(σ-1) (9)
Hence, profits of each country i firm become:
πi = (pi/σ)[xi-(σ-1)F] (10)
Firms break even (π=0) if their sales equal X=(σ-1)F, for this condition price must
satisfy:
PiσX = Σi
R EjGjσ-1Tij
1-σ (11)
Then after substituting with equation (9), firms break even if:
X (Giαwi
βviγciσ/(σ-1)) σ = Σi
R EjGjσ-1Tij
1-σ (12)
This is called the wage equation, so the wage will depend on the sum of distance
weighted market capacities. The demand equation (7) gives the volume of sales per firm
to each location, aggregating we get:
niPixij = niPi1-σTij
1-σEjGjσ-1 (13)
Then, exports depend on market capacity (EjGjσ-1) and supply capacity (niPi
1-σ).
Renaming we have that: niPixij = siTij1-σmj (14)
From here on I would consider E and n as exogenous and look for potential market and
supplier access, and the maximum labor remunerations that each country could offer.
Data Description
I created a 10 year unbalanced panel for the 24 countries under study. The first variable I
collected was the value of total exports from each country i to its 23 trading partners in
millions of US dollars. This data was obtained from the United Nations Commodity
Trade Statistics Database.
The second variable recollected was the Real GDP per-capita at a current price level for
each of the 10 years. This data was recollected from the 6.3 Penn World Table. After
collected, the data was matched to each country for each year, and then I also matched
the GDP per-capita of each of its partners. With these two variables we can see in Graph
1 (in the Appendix) how USA, Canada and Mexico are by far the biggest exporters in
2007.
The third variable recollected was the distance between countries. For this I used the
most standard method, recording the distance in a straight line from the capital of each
country to the capital of each of its 23 trade partners. This information was obtained from
the website: timeanddate.com. Finally, the information for the dummy variables was
included (common border, common language and landlocked).
Empirical Estimation
First I estimated the standard gravity model as defined in the last section. For this panel
data estimation I used a Fixed Effects Model and a Random Effects Model, the results are
listed on Table 1. To check for which model specification is better I used a Hausman
Test, the results from this test are listed in Table 2.
Since the Fixed Effects and Random Effects estimates differ by a large margin, then I can
say that the assumptions for using the Random Effects model are not valid, so I will use
the results from the Fixed Effects Model. I get that the elasticity of income from both the
exporter and importer country are positive and significant. Also, the elasticity of distance
has a negative and significant effect, showing how the higher the distance the higher the
costs of trading. The value of the Common Border dummy is positive but insignificant.
Common language is positive but insignificant also. We can see how there’s a negative
and significant constant term. This term is negative due to the fact that this is a log-log
regression; we need to “delog” the results. For the random effects model, if the
independent variables are zero then we’ll get that LnXij = -4.92, thus Xij = e-4.92 which
approaches zero, thus, there would be almost no exports.
So far I have found a strong effect from distance between countries, to stay with this idea
I move on to estimate the trade equation (13) using bilateral trade data. Here I will get
predictions of siTij1-σ and Tij
1-σmj for each exporting country i and importing partner j. As
in Redding and Venables (2004), I will capture supply and market capacity with country
and partner dummies. This is justified because of the existence of unobserved economic
variables as well as to control for components of transport costs or trade policy that is
common across partners. Transportation costs is modeled using data on the distance
between capital cities. Then I will estimate:
Ln(Xij) = θ + αictyi + αjptnj + δ1Ln(distij) + uij (15)
The coefficients in the country and partner dummies provide estimates for market and
supply capacities of each country. The distance coefficient provides estimates of the
bilateral transport cost. In my results the country dummy is denoted with the full name of
the country and the partner dummy with a capital letter abbreviation. Since there’s trade
among every country included the data is not censored. Therefore, OLS should give good
estimates. The results using a pooled OLS regression are listed in Table 3.
The results are as expected, the coefficients for all the country dummies are positive and
significant, showing how all these countries have exporting behaviors. Regarding the
partner dummies most of the coefficients are not significant, and the ones that are
significant are positive and refer to the most open countries (Brazil, Canada, Chile,
Mexico and USA). This tells us that having these countries as partners would induce
higher trading levels. It important to mention that the richest countries are the ones that
have the higher values for the country dummies.
The distance coefficient is significant and negative as expected (-2.16). It is also bigger
than the one fount by Redding and Venables (-1.53), showing that distance in this
continent makes a bigger difference than the one among a bigger group of countries. This
formulation has an explanatory power of 80% in the trade variation; a simple F test
rejects the null hypothesis of coefficients of country of partner dummies equal to zero.
The next step would be to evaluate how good it is for each of these countries to have
these trading partners. I will construct estimates of total market access for each exporting
country, MAi, and total supplier access for each importing country, SAj. I will sum the
coefficient of each partner and country respectively. To include the transport cost I will
divide each of these terms by the respective distance. Thus, the total distance weighted
market and supplier access would be obtained in the following way:
MAi = Σ(exp(ptnj))αj/distij
SAj = Σ(exp(ctyi))αi/distij
One problem arises with this estimation; since the capitals of Canada and the USA are
less than 1000 km away from each other then the Market Access value for Canada would
be huge. The correction I made was using the distance between Ottawa and Los Angeles
(instead of Washington DC). The results are shown in Table 4.
In order to have a good idea of how good geographical location each country has, graphs
2 and 3 show how much access each country has in order to export their products, and
how much supplier access each country has to import their supplies. In a way what these
measures show us is which countries have the best geographical location, not which
countries are the richest or the more open to trade. From here we can see how the
countries in Central America and the Caribbean have the best location, since they are
close to the big markets in North and South America. We can also observe how the
biggest markets such as the USA and Brazil have small values for both measures, this
tells us that their partners gain o lot from having them close, but they don’t gain much
from having these partners due to the size of their markets.
Having these market and supplier access measures, now I can check which countries have
higher prospective benefits based on their geographical location. I will rewrite the wage
equation (12) as:
(wiβvi
γci) σ = [(1/X)(σ-1)/σ] Gi-σαΣj
R EjGjσ-1Tij
1-σ (16)
Renaming [(1/X)(σ-1)/σ] as A and using equation (14) we get the wage equation as a log-
linear function of its supplier access and market access:
(wiβvi
γci)σ = A(SAi) ασ/(σ-1)(MAi) (17)
From this equation we can see how countries with high market and supplier access should
be able to pay higher wages. Currently this doesn’t empirically happen, the biggest
economies in the continent are far from each other and the countries with better access to
these markets are relatively poor. The puzzle that arises is: Why does this group of
countries with such a good location is unable to improve their economies using
specialization and trade?
Conclusion
The very basic gravity model successfully explained important trading patterns. Almost
all of the predictions were confirmed, the size of markets and the distance between them
are arguably the main factors determining international trade. Nevertheless, this project
showed how other aspects such as being landlocked and having common cultural
characteristics are not very relevant in determining trade flows between countries in the
American Continent.
When we obtain the distance between capital cities we are assuming that all the trade
among countries take place in the areas around the capitals. However this is not true for
the case of big countries like Brazil, Canada, Mexico or the USA where big cities are far
apart from each other. For example, automobile-parts trade is very big between Mexico
and USA, so to get a better estimation on the effect of distance it would be better to
measure the distance between Mexico D.F. and Michigan, rather than between Mexico
D.F. and Washington DC. To make this variation in the distance calculation, research on
specific commodities traded is needed as well as information on the location of industrial
settlements inside countries.
So far integration hasn’t led to equality convergence. I thought that remoteness from
markets and lack of openness was the main cause for this inequality; however, my
finding is that poor countries have very favorable conditions for trade. I can say two
things from this result. First, there are several factors not considered in this study (such as
trade with the rest of the world, human capital differentials, technological gaps and trade
agreements) that are very significant and would help explain wage divergence even with
fully opened economies. And finally, if these external factors become common to all the
countries in this continent, then these geographical measures presented expose that there
are high potential gains from promoting technological specialization and international
trade as an elemental policy for poor countries in Latin America.
References
Alan Heston, Robert Summers and Bettina Aten, Penn World Table Version 6.3, Center
for International Comparisons of Production, Income and Prices at the University of
Pennsylvania, August 2009. http://pwt.econ.upenn.edu/php_site/pwt63/pwt63_form.php
Anderson, J. E.; Wincoop, E. Gravity with Gravitas: A Solution to the Border Puzzle.
National Bureau for Economic Research Working Paper, 8079, 2001.
Bergstrand (1985). The Gravity Equation in International Trade: Some Microeconomic
Foundations and Empirical Evidence. The Review of Economics and Statistics, Vol. 67,
No. 3 (Aug., 1985), pp. 474-481.
Chaney (2008). Distorted Gravity: the Intensive and Extensive Margins of International
Trade. American Economic Review, September 2008, Vol. 98, No. 4.
Anderson, J. E. A Theoretical Foundation for the Gravity Equation, American Economic
Review, 69, 1979, pp.106-16.
Distance between cities taken 11/15/09 from:
http://www.timeanddate.com/worldclock/distance.html
Eaton and Kortum (2008). Technology, Geography and Trade. Econometnca, Vol. 70, No. 5 (September, 2002), 1741-1779. Estevadeordal, Freund and Ornelas (2008). Does Regionalism Affect Trade
Liberalization Towards Non-Members? CEP Discussion Paper No 868.
Lewer, Joshua J. & Van den Berg, Hendrik, 2008. "A gravity model of immigration," Economics Letters, Elsevier, vol. 99(1), pages 164-167, April. McCALLUM, John (1995), “National Borders Matter: Canada-US Regional Trade
Patterns”, American Economic Review, vol. 85, n.º 3, 615-623.
Pöyhöen, P. 1963. A tentative model for the volume of trade between countries. Weltwirtschaftliches Archiv, 90(1): 93-99.
Tinberg, J. 1962. Shaping the World Economy, Appendix VI. An analysis of World Trade Flows. The Twenty Century Fund, New York.
United Nations Commodity Trade Statistics Database: http://comtrade.un.org/db/
Waugh, M. (2009). International Trade and Income Differences
Wei, S-J. Intra-National Versus International Trade: How Stubborn are Nations in Global
Integration?, National Bureau for Economic Research Working Paper 5531, 1996.
World Development Indicators Database from the World Bank:
http://web.worldbank.org/WBSITE/EXTERNAL/DATASTATISTICS/0,,contentMDK:2
0535285~menuPK:1192694~pagePK:64133150~piPK:64133175~theSitePK:239419,00.
html
Appendix
Graph 1
Table 1
Fixed Effects Random Effects Variable Coefficient Variable Coefficient lngdpr .26062159* lngdpr .80482181*** lngdpp 1.5303695*** lngdpp 1.1952241*** lndist -1.1227056* lndist -.35983757*
commonborder -0.47943467 commonborder 2.6799607*** commonlang 0.99159861 commonlang .57041573* landlocked 0 landlocked -1.7385414***
_cons -4.9205183 _cons -12.898423***
Observations 5241 Observations 5241 legend: * p<.05; ** p<.01; *** p<.001
Table 2
Coefficient Difference GDPi -0.5442 GDPj 0.3351 Dist ij -0.7628 Border -3.1593 Language 0.4211
Table 3
F( 48, 5192) 404.46
Prob > F 0 Coef. P>t
R-squared 0.789 lndist -2.160381 0 Adj R-squared 0.787 _cons 13.06874 0
Country Coef. P>t Partner Coef. P>t
Argentina 8.867505 0 ARG 1.330224 0.04 Bolivia 4.488257 0 BOL -0.4922543 0.447
Brazil 9.952323 0 BRA 2.757234 0
Canada 8.935779 0 CAN 3.840161 0
Chile 8.570811 0 CHI 2.251464 0 Colombia 7.09082 0 COL 1.234492 0.056
CostaRica 5.277139 0 COS 0.246408 0.703
Cuba 3.896194 0 CUB -0.3796199 0.557
DominicanR 4.161979 0 DOM 1.13121 0.08 Ecuador 5.945257 0 ECU 0.3161484 0.625
ElSalvador 3.605642 0 ELS -0.148992 0.818
Guatemala 4.813438 0 GUA 0.765453 0.236
Honduras 2.886416 0 HON -0.2097581 0.746 Jamaica 2.037201 0 JAM -0.2644629 0.682
Mexico 9.175837 0 MEX 3.250204 0
Nicaragua 2.024862 0 NIC -0.9108259 0.159
Panama 3.388408 0 PAN 0.75509 0.243 Paraguay 4.08285 0 PAR -1.137679 0.079
Peru 6.711671 0 PER 1.2877 0.046
TrinidadT 5.696866 0 TRI 0.2072317 0.748
Uruguay 6.024391 0 URU -0.1484027 0.819 USA 11.7312 0 UNI 6.488961 0
Venezuela 7.284739 0 VEN 1.640054 0.011
Number of obs 5241
Table 4
Country Market Access Supplier Access Argentina 29.7980085 0.121912698 Bolivia 36.26341592 0.154951968 Brazil 24.57413389 0.130297363 Canada 35.88624988 0.20141518 Chile 28.6946162 0.1158091 Colombia 44.1060109 0.237205226 CostaRica 49.3701641 0.277626534 Cuba 76.58737395 0.450840354 DominicanR 61.87551614 0.353100384 Ecuador 40.88261743 0.212302772 ElSalvador 54.11626523 0.311764238 Guatemala 55.06075196 0.307387935 Honduras 54.82033534 0.313955232 Jamaica 62.55894495 0.364337568 Mexico 44.80045168 0.267225924 Nicaragua 52.15980162 0.29744262 Panama 49.41867111 0.270115383 Paraguay 39.77912651 0.137008774 Peru 34.35146349 0.162991976 TrinidadT 47.45076325 0.253625132 Uruguay 54.56329672 0.132216555 USA 17.44155661 0.091789002 Venezuela 47.93597347 0.263250186
Graph 2
Graph 3