Introduction

Nowadays a rising sector of the American political establishment is pushing for the decrease in Free trade and increases in tariffs on imported products, although this may seem as anachronic it was not so long ago when the government played a greater role in what and with whom the United States traded. In the aftermath of the Second World War a new economic order was established by the United States and its allies, this new regime envisaged free trade as a goal. So since 1945 American tariffs on imported products have gone down even if with some pushbacks along the way as can be seen in the graph below, at the same time the combined volumes of exports and imports as a percentage of Gross Domestic Product(GPD) also increased. Although this cannot be totally attributable to the decrease in tariffs, these two combined effects resulted in a higher number of products being consumed and produced from and for the world with lower tariffs being charged. (import duties data, USITC doc in bibliography, % of GDP world bank data internet)

The following paper started out of a question posed by Prof.Connelly under the History-Lab framework at Columbia University. The research proposed dealt with finding a connection between US trade and State Department official communications. In order to answer this query, there was a need to operationalize the variables. On the policy side, this was achieved by using the collection of State department communications known as Cables, publicly available for the period comprehended between the years 1973-1979. These documents have a series of TAGS an acronym that stands for “Traffic Analyses by Geography and Subject”, that are assigned by the cable drafters as a way to organize and manage communications disseminated across the State Department and its posts. As such Cables can have multiple TAGS that are related to different countries and topics, dependent on their content, therefore the independent variable will be the crossover between TAGS related to countries and TAGS related to trade. As the dependent variable, bilateral trade data will be used from a data base publicly available which was compiled by Robert Lipsey and Harry Bowen using UN World Trade Data. The paper will builds-up on previous research on Diplomacy and trade mainly Rose(2007) and Yakop (2009) that introduce the number of foreign missions(Embassies and consulates) into the Gravity Trade Model to test the impact of trade Diplomacy on commercial flows. The main added value of this paper will be to show that Diplomacy can work both ways depending contributing to increases and decreases in trade.

The paper will be organized with the following structure, first give an introductory background on the international events occurring during the period of 1973-1979 focusing on developments that were relate to international trade. Second, the paper will provide a brief summary of the specific literature reviewed that has dealt with this or similar topics in the past. Third, it will explain the adopted methodology specifically explaining the Gravity Trade Model nature and its structure. Fourth, results will be provided and briefly explained. Fifth, discussion of this paper’s finding and what is the effect introducing cables into the Gravity Trade model. Finally, the conclusion will summarize the paper’s findings and provide further research paths.

Historical backgound

As the collection of cables covers the year ranging from 1973 to 1979 the following section will provide a general bacground about the events developing during those years. As many of the episodes occurring during this period had roots in the 1960’s it requires to look slightly backwards. All of this period was marked by the Cold War, a period in which the Soviet Union and United States were competing for global dominance. As such, the relevant history that pertains this paper is a combination of Economic and political events that impacted US trade flows. The period was marked by a global tension between competing ideologies, Capitalism and Comunism, that were defended by the United States and the Soviet Union respectively.During this period the Super powers engaged in proxy wars in which each supported clients that were more aligned with their ideology being communist or capitalist.

The world was divided among major millitary alliances, mainly the North Atlantic Treaty Organization(NATO) integrated by the US and its western European Allies and the Warsaw Pact which gathered The Soviet Union(SU) and its Eastern European Socialist Allies. Besides these major European alliances there were multiple other regional defense pacts and alliances such as Cuba’s with the SU, or the Inter-American Treaty of Reciprocal Assitance between the US and Latin American nations. Apart from the countries allied to either the US and SU lines, there were other other countries that choose a third line such as the Non-alligned group with countries such India and Yugoslavia. Finally, China even if a communist country started to pivot towards more amicable relations with the US in the year 1972 with the start of the “ping-pong” diplomacy.

During this period, there were three different administration in the US, Nixon a Republican was elected in 1968 and again reelected in 1972 for a second term that started in the year 1973. However due to a political scandal known as the “Watergate scandal” he was forced to resign and Ford took over the presidency. Ford governed till 1977 when he was succeded by Carter a Democrat. Carter took over power till the year 1981, after loosing in 1980 to Ronald Reagan.

In the aftermath of the Second World War the Coordinating Committee for Multilateral Export Controls(coCom) was created when it became clear that the relation with the East would not be as amicable as they were during the war. The organization created in 1948, gathered all NATO Countries, Japan, Turkey and other United States Allies. The objective of this group was to coordinate trade restrictions to be applied to Socialist States such as the SU in three main economic areas:1) Military equipment 2) Atomic material 3)industrial products. ALtough an informal group it had great effect in curtailing trade with Socialist countries and this is reflected in our data, and also in the increase in trade flows during the Detente as was explained before(which temporarily relaxed trade restrictions).\(^xpriceton paper\)

The Cold War, as the name reflects was not a direct confrontation between the SU and the US, as such tension levels varied across time. The period covered by this paper was a time of relatively low tension known as the “Détente”, a french term meaning “relaxation”. As such, the years ranging from 1969 to 1979 saw increasing cooperation among the superpowers that included arms controls such as the SALT agreement and the Helsinsky Accords which improved the relationship between the two blocks in Europe. As a result of this improvement in relations trade between the US and the SU increased, specially with the export of grain to the SU and import of non ferrous metals to the US.\(^xxarticleontheinternet\) The “Détente” came to its end with the Soviet invasion of Afganisthan and the election of Ronald Reagan, that engaged in a less conciliatory position to the SU.

The United States entered the 70’s engaged in a Proxy war in South East Asia which included Laos, Cambodia and Vietnam. The majority of the confrontation took place in Vietnam, with the US supporting South Vietnam against North Vietnam supported by both China and Russia. American involvement in the conflict started in the year 1959 with the deployment of advisors, but a major deployment of troops would only occur in the year 1965 when over 180.000 troops were sent to the Country. The war created a big strain in public finances as it meant a great increase in public expenditures, it had a cost of 2.3% of GDP in its peak year and a combined cost of 738 Billion 2011 dollars according to a Dagget(2010). American direction intervention ended in the year 1973 after the Paris accords, it continued supporting South Vietnam till the year 1975 when North Vietnamise forces finally overran the south. In relation to trade this means that the US ended its diplomatic relations with Cambodoia, Laos and Vietnam in 1975 and with it trade collapsed.

Another event that greatly affected, both politically and economically, the US during this period was the aftermath ot the 1973 Arab-Israeli Yom Kippur War. The Middle East served also as another scenario for the Cold War, with the US supporting Israel and the SU the Arab States. The October war, as it is also known, started with a surprise attack by the combined armies of the Egypt and Syria against the Jewish State. After initial success, the Arab armies were pushed back and finished behind their original combat lines, this happened with great US support to Israel by the way of an airlift that provided crucial and abundant war material during the war.

As a result of US support to Israel during the war, the Organization of Arab Petroleum Exporting Countries(OAPEC) impossed an oil embargo against America from October 1973 to March 1974. As prices of oil almost quadrupled in 1974 compared to 1973, this impossed great strains to the American economy which was forced to impose comsuption rationing as well as increases in fuel prices. Both these effects created a situation of inflation and economic recession that impacted US trade flows. Altough trade flows tend to change slowly across time as can be seen on the graph below for the top 10 trade partners in the years 1970 and 1980, there can be observed an increase in the flows relating to countries from which the US imported Oil such as Saudi Arabia(SAU) and Nigeria(NGA). Another effect of the oil embargo was the imposition of a tax exports on oil in Canada, this contributed to the reduction of oil exports to the US by the northern neighbour.\(^XXpage 46 oil canada\)

A final international event that greatly affected American politics and Trade was the Iranian Islamic revolution of 1979, which saw America lose one of its biggest allies in the region. The United States supported the Sha Mohammad Reza Pahlavi and had dynamic relations with Iran being one of the principal US allies in the region. Though, this situation dramatically changed in 1979 with the development of the Islamic Revolution. Ayatollah Khomeini was the leader of this movement, that deposed the Sha who flead Iran, the result was an Islamic Republic which broke relations with the West and as a result trade flows became non existent.

US and trade

The post War System

In the aftermath of the Second World War a new Economic system was established under the tutelage of the victors and with a big imprint by the US. This system intended to revert the chaos seen in the interwar years and consisted of three main pillars: The International Monetary Fund(IMF), International Bank for Reconstruction and Development known as the World Bank and the International Trade Organization(ITO) that later was replaced by the General Agreement on Tariffs and Trade. The objective of this new system was to create a new world order that would liberalize trade and create a stable international monetary regime that respected national sovereignty. The IMF would work to establish a new currency framework which saw all other currency pegged to the US dollar which itself could be converted to Gold. The World Bank was intended to work as the basis for the reconstruction of the war affected countries although it was largely superseded by the Marshall Plan. Finally the ITO, was initially intended to rule over International Trade, however this institution failed to achieve the necessary Congress approval and was replaced by the GATT that was basically a loose set of rules inside an accord. The GATT, had three main objectives, first to set a new set of rules to govern international trade, second to establish a new forum for trade negotiations and finally to set a framework for the settlement of trade disputes. The agreement was based on two basic the first: principles no discrimination, meaning that benefits to any country given inside the agreement should be given to all members. Also, that foreign products once interned should be treated as if they were locally produced. The second, principle was reciprocity that intended to limit the possibility of free riding, under this principle benefits given by a country to another should trigger the same benefits to the initial benefactor by the beneficiary. This framework, was created with the idea of establishing a system that replaced the protectionist policies of the interwar period in a gradual way. For this reason, many safe guards were included that allowed countries to limit their trade in case it affected their national security, degraded the environment, products were the result of dumping policies among other opt out clauses. This gave ample room for countries to limit their imports while remaining inside the GATT framework.

Changes to the system

The post war regime worked well for over two decades, however changes in the underlying reality meant that in the mid 60’s it started to show strain and that by 1971 it would be cease to exist. It is difficult to pin point for a unique reason for this change, but a mixture of both domestic and international factors forced the changes which would also affect US approach to trade. Among the domestic factors, increasing Fiscal deficits, higher inflation and slower growth were crucial in influencing the US towards a less conciliatory stance in the liberalization of Trade.

Among the international factors, the US found itself in a privileged position in the war with its territory not being directly damaged by the war and a small share of its population affected. This meant that in 1950 the US represented over 35% of global GDP a proportion that was reduced by the recovery of the war affected countries which meant that in 1980 this share would shrink to 24%. Both of these effects meant that starting in 1969 the US would enter a new Current account dynamic that would reverse the post war surplus to constant deficits. This situation forced the US in the year 1971 under the Nixon administration to finish the US dollar convertibility to gold and impose temporary a 10% tariff on imports.

Both of these effects meant that starting in 1969 the US would enter a new Current account dynamic that would reverse the post war surplus to constant deficits. This situation forced the US in the year 1971 under the Nixon administration to finish the US dollar convertibility to gold and impose temporary a 10% tariff on imports. This would mark the beginning of a change in the US stance towards trade liberalization with the creation of diverse mechanism intended to limit imports among these we can identify the following: Multi fiber Agreement(MFA), Voluntary Export Restrictions(VER) and Section 301 of the 1974 Trade Act. All of these measures were meant to diminish imports by different means that will be discussed.

The rise of new industrialized countries such as Japan that had massive and cheap labor force meant that they quickly became highly competitive in the textile sector. As a result of this new phenomena the MFA was instituted in order to limit the quantities of textiles exported to industrialized countries such as Europe or the US.

Other specific industries that were affected by the rise of these new industrial powers were the automotive and steel industries. In this case bilateral agreements were reached between the US and the other countries in which the counterpart agreed to limit the amount exported to the US which in practice amounted to the establishment of a quota. However, because this was agreed and not imposed it could be implemented under the GATT framework. Finally, Section 301 of the 1974 Trade Act, under this section the president was given the power to retaliate against discriminatory practices to US exports by other countries. This Section, that defined discriminatory practices in the broadest way created a mechanism for the US to implement measures against imports.

OPIC

The overseas Private Investment Corporation, was an organization created by President Nixon in the year 1969 by the Foreign Assistance Act to help American investments in development nations by mobilizing private capital. Its two main objectives were provide insurance against political risk and provide the necesary funding for american projects in less developed nations. I invested in allies of the United States being the main beneficiaries South Korea, Indonesia, Brazil, Ghana, Taiwan, the Philippines and other nations. These nations reflect the fact that OPIC has wide geographical impact and was used in a wide array of sectors industrial, agribusiness, services and techonology.

Methodology

Gravity Model

The origins of Economics can be closely related to the International Trade, as such David Ricardo a British Economist in the XVIII century devoted a big part of its academic production focusing on the isssue of trade, and specifically comparative advantange. However,even if commercial flows paid such a central role in economics it would be only in 1962 that Jan Tinbergen would come up with an empirical model that would predict the volume of flows betweeen two countries. The Dutch economist came up with a model borrowed from physics originally developed by Isaac Newton. The application known as the Gravitational Trade model can be seen below:

\[T_{US,Other} = G_{US,Other} \frac{(GDP_{US})^\beta(GDP_{other})^\gamma}{(Dist_{US,Other})^\zeta}\] Where:

\(T_{US,Other}\) :trade volumes between the United States and other countries. \(G_{US,Other}\) :A constant refered by the letter G that affects the size of trade between the pair of countries.\((GDP_{US})^\beta(GDP_{other})^\gamma\): the product of the Gross Domestic Product of both Countries.\((Dist_{US,Other})^\zeta\):The distance between the pair of countries.

The simplicity of the model and its explanatory power made it really atractive but one of the major critics it attracted were due to the fact that in its original form it lacked any type of formal economic foundation. It would take till 1979 when Anderson developed a Gravity model based on formal economic assumptions. In order to do so the author assumed that every country produced a unique imperfectly substitutable product and identical preferences across countries. At the moment this asumptions were seen as arbitrarely, but it still managed to provide the first theoretical foundations that connected GDP size to GDP per capita to the share of tradable goods in an economy a crucial factor for trade.\(^1\)

In 1985 Bergstrand further expanded the microeconomic foundations by allowing for different preferences among countries and different prices between products that were produced and consumed locally and the ones that were exported. However, this new model had limited empirical applications as price indexes need to have a base year for comparisson, which is suitable for time series but has limited applications for panel data analysis. \(^2\)

In 2001 Anderson and Van Wincoop developed a new model that could better incorporate both relative prices and different preferences thus solving part of the critiques to the Gravitational Trade model. Published in 2003, “Gravity with Gravitas: A Solution to the Border Puzzle”, provided a sound framework that would be explained using Baldwin & Taglioni (2006) paper.

The first step required to understand the Gravity equation is to determine the expenditure share identity:

\(p_{US,Other}X_{US,Other} \equiv share_{US,Other}E_{other}\);

\(P_{US,Other}\) refers to the price of American exports to other country for a single product, \(X_{US,Other}\) refers to the quantity being exported. The multiplication of price and quantity need to be equal to \(share_{US,Other}\) of \(E_{other}\), meaning the share of the other country expenditure on tradable goods that is spent of the specifict product of interest exported by the US. These identity would work both ways with the US as being the importer country as well.

The second step is to descrive how is the expenditure share is determined:

\(share_{US,Other}\equiv (\frac{p_{US,Other}}{P_{other}})^{(1-\sigma)}\)

, where \(P_{other}\equiv (\sum_{k=1}^{R}n_{k}(p_{kother})^{(1-\sigma)})^\frac{1}{(1-\sigma)}\) , \(\sigma\) > 1

From the above equations \(p_{US,other}/P_{other}\) is the real price in the other nation of the exported product by the US. \(P_{other}\) is the ideal price index at the other nation if all goods were traded. ‘R’ is the total number of of tradding partners including itself, \(\sigma\) is the elasticity of substitution among all varities, that are assumed to be symetric. \(n_{k}\) is the total number of variaties being traded with other from country k and \(p_{k-other}\) is the price of the k nation landed at the nation of interest.

In our third step we would describe how the price at the destination, which is compossed of of production costs at the US, a bilateral markup and bilateral costs.

\(p_{US,other} = \mu p_{US} \tau_{US,other}\)

\(p_{US}\) is the production cost at the US, \(\mu\) is the bilateral markup which is determined by the producer in order to maximize its profits. Finally \(\tau_{US,other}\) represents the bilateral trade cost factor that contains all costs that are man mande or natural barriers that affect price and will always be bigger than one. This last variable of \(\tau\) will be our variable of interest as we will try to estimate the government influence in it.

The fourth step would require to aggregate the values of all products being traded in order to reach the total value of bilateral trade.

\(V_{US,other}\equiv n_{US}S_{US,other}E_{other}\) Where the total volume traded from the US to other country(\(V_{US,other}\)), would be the product of the different types of products(\(n_{US}\)) by their respective(\(S_{US,other}\)) share and the other country expenditure in tradable goods(\(E_{other}\)).

by rearranging the above formula we reach:

\(V_{US,other}\equiv n_{US}(p_{US}\tau_{US,other})^{(1-\sigma)}\frac{E_{other}}{P_{other}^{1-\sigma}}\)

However as we lack information about prices and different products we will use the general equilibrium condition, where we assume that all markets clear and all products are either consumed in the US or being exported to other countries.

\(Y_{US} = \sum_{d=1}^{R}V_{US,other}\), where the total US output would vbe the summation of all its billateral flows including with itself.

Relating the markets clearing conditions with equation 4 we reach that:

\(Y_{US} = n_{US}p_{US}^{(1-\sigma)}\sum_{d=1}^{R}(\tau_{US,d}^{(1-\sigma)}\frac{E_{d}}{P_{d}^{(1-\sigma)}})\)

Again the summation is done over all markets including US own market. solving for \(n_{US}p_{US}^{(1-\sigma)}\):

(6)\(n_{US}p_{US}^{(1-\sigma)} = \frac{Y_{US}}{\Omega_{US}}\), where \(\Omega_{US}\equiv\sum_{i=1}^{R}(\tau_{US,i}^{(1-\sigma)}\frac{E_{i}}{P_{i}^{(1-\sigma)}})\)

\(\Omega_{US}\) is the “market potential”, meaming the sum of all possible tradding partners weigthed by bilateral distance.

By exchanging \(n_{US}p_{US}^{(1-\sigma)} = \frac{Y_{US}}{\Omega_{US}}\) in \(V_{US,other}\equiv n_{US}(p_{US}\tau_{US,other})^{(1-\sigma)}\frac{E_{other}}{P_{other}^{1-\sigma}}\) we reach:

\(V_{US,other} = \tau_{US,other}^{(1-\sigma)}(\frac{Y_{US}E_{other}}{\Omega_{US}P_{other}^{(1-\sigma)}})\)

These equation greatly resembles our initial Gravity one, and going one step forward while assuming that \(E_{other} \approx Y_{other}\) we reach:

\(T_{US,Other} = G_{US,Other} \frac{(GDP_{US})^\beta(GDP_{other})^\gamma}{(Dist_{US,Other})^{(1-\sigma)}};\) \(G\equiv\frac{1}{\Omega_{US}}\frac{1}{P_{other}^{(1-\sigma)}}\)

\(^1\) : Taglioni & Anderson \(^2\) : Bergstrand, Gravity sruvey pag.18

Regression

ln(\(V_{Us,other}\))=\(\beta_1\)ln(\(Pop_{other}\))+\(\beta_2\)Distanc\(e_{US,other}\)+\(\beta_3\) \(ln(Y_{other})\)+ \(\beta_4\)pct_common_rel+\(\beta_5\)dummy for commong lang+ \(\beta_6\)Landloc\(k_{other}\)+\(\gamma\)Cables+\(\alpha_t\)+\(\epsilon\)

The gravity model has seen multiple implementations since its beggining in 1962, searches for “Gravity Model Trade” in google scholars return over 659.000 articles related. This means that a multiple and diverse variables have been theorized and implemented as ways for explaining bilateral trade volumes. The following section will introduce the ones choosen in this paper, providing the reader with their sources and justification.

Gravity Variables

The dependent variable is \(V_{Us,other}\), its source is the UN World trade data as compiled by Fennstra et all(2005).\(^3\) The dataset covers the years 1962-2000, and reports trade volumes in nominal tousands of dollars which presents an advance from other sources that had their lower limit as 100.000 dollars which biased the sample towards underrepresenting flows to smaller economies. A number of modifications in the original dataset were done in order to standarize the sample in accordance with the countries available in the other databases. As a result, of the original 153 countries that traded with the US only 145 were left representing 99% of all trade also countries not in the database for lack of trade but relevant such as North Korea were added to the sample. As the logarithm of Volumes is used instead of flows by themselves this presents an issue as it would be impossible to include logarithm of 0, therefore consistent with the literature a one is added to all the sample, the reason for this being done is that after inspecting the sample it is clear that the lack of data represents no trade and not lack of information.\(^4\) Another minor issue arises from the fact that the Trade data base merges some minor countries such as San Marino and the Vatican inside Italy etc. Please refer to the appendix for a complete list.

The next three variables are all related, the logarithm of Population, area and Gross domestic product(GDP). Population data is reported in Millions of habitants and has been accessed from the CEPII’s gravity dataset that was retrieved from the World Development Indicators from the World Bank. Finally, GDP data is reported in nominal values and was retrieved from the UNdata available online. This three variables are related to the work of Krugman(1980) and Melitz & Ottaviano(2008), both these papers deal with the issues of national per capita wealth, market size and productivity. Their conclusion is that bigger economies would tend to have bigger companies that would benefit from higher productivity therefore engaging less in global trade as a share of their economies. However, this effect is mitigated by the fact that wealthier economies measured by their GDP per capita would tend to have a love for variety therefore would tend to consume more foreign products. The combination of these papers recommends for the introduction of population, GDP and Area data in order to measure these elements. \(^5\)

The distance variable is related to \(\Omega\), meanign the trade potential for a country as the product of the other countries GDP and their distance adjusted by population. This data set was retrieved form the CEPII distance data set which uses the following formula in order to compute national distance computed by population:

\(d_{US,other}\)=(\(\sum_{k \epsilon i}^{25}\)(\(pop_k/pop_i\)) \(\sum_{l \epsilon j}^{25}\)(\(pop_l/pop_j\))\(d_{kl}^\phi\))\(^{1/\phi}\)

Where, \(pop_k\) and \(pop_l\) represent the population of one of the top 25 cities of the two countries, \(pop_i\) and \(pop_j\) are the total population of both each country respectively. \(d_{kl}\) is the distance between the pair of cities and \(\phi\) is the parameter of sensitivity to trade flows and is set equal to 1. Therefore, the weighted distance results from the weigthed sum of the distance between the countries top 25 cities by their share of each country total population.\(^6\)

The final variables are related to \(\tau_{US,other}\) represents the bilateral trade cost factor being natural or man made. Numerous papers have dealt with this factor focusing on

Commong religion is measured as the chance that two random persons from each country will share the same religion and ranges from 0 to 1, the dataset was obtained from the CEPII gravity dataset. This variable has been used by multiple authors among them Helpman et all(2008) who perform diverse robustness checks in order to probe that common religion seems to affect trade costs and increase the chances that two countries will trade after controlling for other factors.\(^8\) The formula for this variable would be the following:

\(comrelig_{US,other}\)=(%\(protestants_{US}\).\(protestants_{other}\))+(% \(Catholics_{US}\).\(Catholics_{other}\)) + (% \(Muslims_{US}\) . \(Muslims_{other}\))

Common language is a dummy that measures whoever the pair US and other country share the same official language and also comes from the CEPII gravity dataset. This variable is associated with lowering transaction costs between the two nations and as a results increases the chances of tradding among the pair of countries.\(^9\) A final variable related to trade costs is Landlocked, which is a dummy variable that measures as 1 if the country is landlocked(no access to a sea) or 0 if it is not. The fact that a country has no access to a sea has been identified as increasing its trade costs as one of the possible ways of moving its merchandise in and out of the territory is not available.\(^{10}\)

The following presents summary statistics of the used variables:

Dummy language	Distance	Pct Common religion	GDP	Dummy Landlock	Log EXp	Log Imp	Log Total
Min. :0.00	Min. : 2079	Min. :0.00	Min. : 0.31	Min. :0.00	Min. : 0	Min. : 0	Min. : 3.1
1st Qu.:0.00	1st Qu.: 7457	1st Qu.:0.01	1st Qu.: 6.54	1st Qu.:0.00	1st Qu.:10	1st Qu.:10	1st Qu.:10.1
Median :0.00	Median : 9706	Median :0.16	Median : 24.73	Median :0.00	Median :12	Median :12	Median :11.8
Mean :0.28	Mean : 9473	Mean :0.16	Mean : 215.51	Mean :0.13	Mean :11	Mean :12	Mean :11.6
3rd Qu.:1.00	3rd Qu.:12267	3rd Qu.:0.28	3rd Qu.: 141.01	3rd Qu.:0.00	3rd Qu.:13	3rd Qu.:13	3rd Qu.:13.3
Max. :1.00	Max. :16466	Max. :0.43	Max. :4630.73	Max. :1.00	Max. :17	Max. :17	Max. :17.5

Regression avg of (log imp + log exp)


	Dependent variable:

	Log of Trade Flows
	Gravity	TAG 1	TAG 2	TAG 3	TAG 4
	(1)	(2)	(3)	(4)	(5)

log(pop_d)	-0.163^**	-0.232^***	-0.203^***	-0.232^***	-0.252^***
	(0.076)	(0.070)	(0.071)	(0.070)	(0.072)

log(distw)	-0.868^***	-0.822^***	-0.901^***	-0.871^***	-0.888^***
	(0.199)	(0.201)	(0.203)	(0.205)	(0.208)

log(GDP)	1.119^***	1.187^***	1.132^***	1.127^***	1.132^***
	(0.074)	(0.073)	(0.077)	(0.076)	(0.076)

comrelig	1.453^*	1.355^*	1.555^*	1.545^*	1.477^*
	(0.864)	(0.807)	(0.817)	(0.811)	(0.822)

comlang_off	0.574^***	0.523^***	0.537^***	0.470^**	0.489^**
	(0.205)	(0.202)	(0.205)	(0.207)	(0.206)

landlock	-1.186^***	-0.975^***	-0.938^***	-0.892^***	-0.848^***
	(0.309)	(0.281)	(0.279)	(0.282)	(0.281)

TAG_Investment		0.296^***	0.248^***	0.238^***	0.219^***
		(0.077)	(0.078)	(0.075)	(0.072)

TAG_Trade_Control		-0.144^***	-0.139^***	-0.152^***	-0.148^***
		(0.039)	(0.040)	(0.050)	(0.047)

TAG_Trade_Expansion			0.226^**	0.193^*	0.164^*
			(0.097)	(0.099)	(0.098)

TAG_Property_Protection_Services			-0.084^**	-0.091^**	-0.097^***
			(0.036)	(0.036)	(0.035)

TAG_Animal_Plants_and_Wood_Prod				0.160^***	0.152^**
				(0.061)	(0.059)

TAG_Export_import_Bank					0.117^*
					(0.064)


Observations	822	822	822	822	822
R²	0.780	0.800	0.804	0.807	0.808
Adjusted R²	0.777	0.796	0.800	0.803	0.804
F Statistic	477.922^*** (df = 6; 810)	402.970^*** (df = 8; 808)	330.117^*** (df = 10; 806)	305.579^*** (df = 11; 805)	282.659^*** (df = 12; 804)

Note:	p<0.1; p<0.05; p<0.01

Results

The above table presents 5 different models tried with the data explained, the first one is the restricted model that does not include any TAG variable while Models 2-5 include a combination of different TAGS that maximize the R square and are significangt. The inclusion of these models and its order is intended to give the reader an incremental exposure to the different models. The Sample comprehends the years 1974-1979 representing 137 different countries that had trade flows with the US.The regression above has time fixed effects, meaning that what is meassured is the association between the different variables and trade.For example when loking at the Population coefficent(log(pop)), in the first model, that has a value of -0.163, its meaning is that an incresase of 1% in the other country population would represent on average a reduction of 0.163 percent in trade flows. As can be seen from the above table all coefficients are significant and therefore all models are relevant, with the adjusted R square going the restricted 0.77 to the highest R square achieved is small 0.80. The sign of the variables in the restricted models is preveserved along the other unrestricted models but maginitude change as would be expected.

Three TAGs are present among all the unrestricted models as they are they have the higher frequency among a sample of the highest 297 Rsquare possible models with the restricted and four different TAGS. These are : Investment, Trade Expansion and Trade Controls. The remaining TAG’s are introduced as they are also present significant coefficents and are provide interesting information about the cables content.

Restricted Model

We find that the restricted model coefficents are consistent with prior findings such as Yakop(2008) both in direction and magnitude,therefore all models seem to be correct in terms of their parameters. The restriced model adjusted R square is 0.777, meanign that the traditional Gravity model explains almost 77% of all trade flows. the coefficient for the log of population is -0.163 significant at the 5% level, meaning that a 1 percent increase in population across countries would result in a decrease of 0.163 percent in trade flows keeping everything else constant. Our second variable is the log of distance weigthed by population, in this case a 1 percent increase in the distance would result on average in a reduction of 0.868 in trade flows significant at the 5% level. The third variable is GDP, in this case a 1 percent increase in GDP across countries would result on average in a increase of 1.119 percent in trade flows. Forth is the percentage of people with common religion, in this case a 1% increase in people with the same religion would result on average in a 0.014 percent increase in trade flows. Fifth, is the common official language dummy that means that on average the existence of a common language would result in a 0.574 percent increase in flows. Finally, the dummy for landlock countries, meaning that countries with no access to a sea would see their trade flows reduced by 1.186 percent on average. Other variables that are usually included such as the dummy for Island, was not included as the coefficient was positive, meanign that islands tend to trade more which is not consistent with the literature and therefore was not included in this study.

TAG models

This section will describe the TAG models 2-5 of the above table and their coefficents, for the gravity model coefficents please see the section above. The firs TAG model includes the two TAG’s that maximize the R square by themselves, it can be seen that the Adjusted R square increase by 2% from 0.77 to 0.79, meaning that the addition of this two variables increases the model explanatory power. The TAG investment should be employed according to the State Department “…for communications pertaining to international investment policy and planning.”. This coefficent is significant at the 1% level, and it can be interpreted as a 1% increase in the share of total cables with this TAG for a specific year would result in a 0.3 percent increase in trade flows. The second TAG is related to Trade Controls, is used intended to “Covers all matters related to the control of exports for economic warfare or economic defense purposes.”, also significant at the 1% level. It meanign is that a 1% increase in the share of total cables with this TAG for a specific year would result in a 0.14 percent decrease in trade flows.

The TAG 2 model expands the first model adding two different TAG’s Propery protection Services and Trade Expansion, the increment in Adjusted R square is less just increasing 0.004 from the previous model but keeping all coeficents signicant. The Trade Expansion was employed according to the State Department for “…relating to activities of the Department of Commerce to promote the U.S. trade, including official U.S. trade exhibitions in the United States and abroad, trade fairs, missions, centers, Trade Development Trade Information Offices (TDTIO’s)”. This coefficent is significant at the 5% level, and it can be interpreted as a 1% increase in the share of total cables with this TAG for a specific year would result in a 0.22 percent increase in trade flows. The second TAG is related to Propery protection Services, which “Includes all matters pertaiing to protection of ownership, interests, or claims of U.S. citizens, ro personal or real property, including intangible property in foreign countries.”, also significant at the 5% level. It meanign is that a 1% increase in the share of total cables with this TAG for a specific year would result in a 0.08 percent decrease in trade flows.

The TAG 3 model expands the first model adding one additional TAG Plant, Animal, and Wood Products, the increment in Adjusted R square is less just increasing 0.003 from the previous model but keeping all coeficents signicant. This TAG was employed according to the State Department for “… processed and unprocessed plant, animal, and wood products and production or processing facilities.”. This coefficent is significant at the 5% level, and it can be interpreted as a 1% increase in the share of total cables with this TAG for a specific year would result in a 0.16 percent increase in trade flows.

The TAG 4 model expands the first model adding one additional TAG Technology, the increment in Adjusted R square is less just increasing 0.004 from the previous model but keeping all coeficents signicant. This TAG was employed according to the State Department for “…papers pertaining to policies and activities of governmental and private organizations involved in technology.”. This coefficent is significant at the 1% level, and it can be interpreted as a 1% increase in the share of total cables with this TAG for a specific year would result in a 0.137 percent decrease in trade flows.

Country Fixed Effetcs

The following section presents the results of diverse TAG’s that presented interesting results when running our regression with Country Fixed Effects. The main difference between the Time and Country fixed effects, is that the Country fixed effects control for everything that is constant along time, therefore variables such as distance, cultural factors(religion, language) and population are dropped from our regressions and what is left is what varies in time. Therefore the model includes Log of GDP and the TAG of interest, in order to give a direction in time the model is done with diverse time Lags from 0 to 3 years. This type of regression measures the changes and their effect, meaning that if a country gets more trade than its mean and also more cables with a specific TAG it would find a positive coefficent, it is relevant to say as has been explained before trade flows are greatly explained by non changing factors and as a result what is left to explain after removing this is small and therefore the R square loses its significance.


	Dependent variable:

	Log of Trade Flows
	No lag	Lag(1)	Lag(2)	Lag(3)
	(1)	(2)	(3)	(4)

OPIC	-0.004
	(0.020)

Lag 1 OPIC		0.029
		(0.023)

Lag 2 OPIC			0.041^*
			(0.022)

Lag 3 OPIC				0.029
				(0.025)

Log(GDP)	1.170^***	1.243^***	1.040^***	0.662^***
	(0.172)	(0.197)	(0.186)	(0.200)


Observations	822	822	822	822
R²	0.113	0.118	0.080	0.030
Adjusted R²	-0.066	-0.060	-0.106	-0.165
F Statistic (df = 2; 683)	43.612^***	45.837^***	29.730^***	10.736^***

Note:	p<0.1; p<0.05; p<0.01

The Table above shows the results of the regression of Trade flows on GDP and Overseas Private Investment Corporation(OPIC), which is intended for all cables regarding this organization according to the State Deparment. The coefficient is not significant with no lag or one year Lag, but it becomes significant at the 2 year lag at the 10%. The coeficient can be interpreted as an increase in on percentage of cables with this TAG results in an increase of 0.041 percent in trade flows after two years.

## 
## =======================================================================
##                          Model 1     Model 2     Model 3     Model 4   
## -----------------------------------------------------------------------
## log(GDP)                   1.14 ***    1.13 ***    0.86 ***    0.48 *  
##                           (0.19)      (0.25)      (0.23)      (0.21)   
## factor(deve)2              0.05        0.12        0.22        0.39 ***
##                           (0.18)      (0.16)      (0.13)      (0.11)   
## factor(deve)3              0.32        0.32        0.47 *      0.62 ***
##                           (0.20)      (0.22)      (0.21)      (0.18)   
## loop                      -0.00                                        
##                           (0.03)                                       
## factor(deve)2:loop         0.00                                        
##                           (0.07)                                       
## factor(deve)3:loop        -0.02                                        
##                           (0.07)                                       
## lag1.loop                              0.02                            
##                                       (0.03)                           
## factor(deve)2:lag1.loop                0.04                            
##                                       (0.05)                           
## factor(deve)3:lag1.loop                0.05                            
##                                       (0.06)                           
## lag2.loop                                          0.04                
##                                                   (0.03)               
## factor(deve)2:lag2.loop                            0.01                
##                                                   (0.04)               
## factor(deve)3:lag2.loop                            0.04                
##                                                   (0.05)               
## lag3.loop                                                      0.04    
##                                                               (0.03)   
## factor(deve)2:lag3.loop                                       -0.02    
##                                                               (0.05)   
## factor(deve)3:lag3.loop                                       -0.01    
##                                                               (0.05)   
## -----------------------------------------------------------------------
## R^2                        0.13        0.14        0.10        0.05    
## Adj. R^2                  -0.06       -0.05       -0.09       -0.15    
## Num. obs.                794         799         802         806       
## =======================================================================
## *** p < 0.001, ** p < 0.01, * p < 0.05


	Dependent variable:

	Log of Trade Flows
	No lag	Lag(1)	Lag(2)	Lag(3)
	(1)	(2)	(3)	(4)

Emergency Evac	-0.018
	(0.011)

Lag 1 Emergency Evac		-0.035^*
		(0.021)

Lag 2 Emergency Evac			-0.029^*
			(0.017)

Lag 3 Emergency Evac				-0.017
				(0.010)

Log(GDP)	1.146^***	1.207^***	1.002^***	0.645^***
	(0.173)	(0.193)	(0.183)	(0.201)


Observations	822	822	822	822
R²	0.118	0.134	0.090	0.034
Adjusted R²	-0.060	-0.041	-0.094	-0.162
F Statistic (df = 2; 683)	45.772^***	52.903^***	33.668^***	11.864^***

Note:	p<0.1; p<0.05; p<0.01

The regressions above shows the results of the regression of Trade flows on GDP and TAG Emergency and Evacuation, which is intended for “Use for all emergency and evacutation matters…” according to the State Deparment. The coefficient is significant at all time Lags, becoming significant at the 1% on one and two year lags. The coeficient can be interpreted as an increase in one percentage of cables with emergency and evacuation in an decrease of 0.035 percent in trade flows after one year and 0.029 percent after two years.


	Dependent variable:

	Log of Trade Flows
	No lag	Lag(1)	Lag(2)	Lag(3)
	(1)	(2)	(3)	(4)

Trade Complaints	-0.013
	(0.014)

Lag 1 Trade Complaints		-0.083^**
		(0.041)

Lag 2 Trade Complaints			-0.054^*
			(0.031)

Lag 3 Trade Complaints				-0.037
				(0.025)

Log(GDP)	1.177^***	1.251^***	1.012^***	0.650^***
	(0.172)	(0.197)	(0.186)	(0.201)


Observations	822	822	822	822
R²	0.114	0.140	0.088	0.034
Adjusted R²	-0.065	-0.034	-0.096	-0.161
F Statistic (df = 2; 683)	43.853^***	55.369^***	32.919^***	12.127^***

Note:	p<0.1; p<0.05; p<0.01

The regressions above shows the results of the regression of Trade flows on GDP and TAG trade Complaints and disputes, which is intended for “Use for communications, including followups, concerning a specific trade dispute between a U.S. and a foreign business, and for trade complaints regarding U.S. firms or products.” according to the State Deparment. This variable is significant at Lags for the second adn third year. The coeficient can be interpreted as an increase in one percentage of cables with foreign assistance in a decrease of 0.083 percent in trade flows after one year and 0.054 percent after two years.

Selected TAG’s cables content and overlapping

A final step before discussing the results will be done in order to better undestand what is the content of the selected cables and briefly describe them. The Technolog TAG that by the name itself seems to be quite broad and does not provide an idea about its content, seems to be used to talk about nuclear technology which can be infered from the main words “nuclear”, “iaea” that stands for International Atomic Energy Agency, “Reactor” which all seem to be related to nuclear energy. When looking at the overlapping of TAG’s this furthers reinforces this idea with Energy, IAEA, Arms control and Disarmament and Military Nuclear Applications being the major TAG associated with Technology. Altough this is not a definitive content description both methods seem to indicate that the fact that this TAG is dealing with Nuclear issues and military applications it produces a negative coefficent for what would seem to be a non related topic.

The second TAG to be discussed is “Investment” which wil be described together with OPIC’s TAG as the later has a 60% overlapping being a subtopic of Investment. Both TAG’s seem to be related with Investment projects with “investment”,“project”,“company” among the main words in their wordclouds, with the cabeat that “agreement”,“loan”, “insurance” appear often on OPIC which is consistent with OPIC’s work in development projects. When looking at the TAG’s overlapping OPIC TAG is the biggest overlapper in Investment but not the only one, with Metal and Minerals, Energy, Foreign TRade and Financial and Monetary Affiars this leads to think that Investment is a general TAG that is being used for investment issues that are related to other major topics, while OPIC seems to be a subtopic of investment that specifically deals with the organization projects.

Trade Expansion wordcloud has trade, equipment, commercial, project among its main words that when combined with the overlapping TAG’s Travel by Businessmen, foreign trade, investment opportunities and engineering and construction services builds in the idea that State Department officals were talking about the implementation of projects and information about them. A special focus seems to be placed on major infraestructure or machinery projects with Construction and equipment appearing on top on both visualizations. This would justify the positive coeffcient seen on the previous regressions.

Analyzing the Strategic Trade Controls, the most used word is CoCom which stands for Coordinating Committee for Multilateral Export Controls, as was explained before this was an European forum for the limitations of exports to the Socialist Block Countries. The next words are license, equipement, computer, firm etc… which seems to reinforce the idea that these cables are discussing specifics about the work CoCom’s work, which is further reinforced by the big overlapp of over 50% with CoCom’s TAG, and that altough not included in the reverse case 90% of CoCom’s cables have the Stratefic Trade Controls TAG included. All of this seems to confirm that CoCom is a subtopic of Strategic Trade Controls and the negative sign on the coefficient observed.

The TAG “Plant, Animal and Wood Products” has on its wordcloud the names of multiple commodities such as coffee, meat, sugar, wheat etc.. and data that seems to be related to them such as price and quantities, also it also contains agreement, trade and meeting. On the overlapping we can identify foreign trade and also Agriculture and Forestry, all of this leads to think that the cables are describing issues related to commodity trade between the US and the other countries. This would be consistent to the positive coefficent as more cables related to this TAG would be meaning more trade in commodities.

The TAG “Property Protection Services”, has the words extradition, case, request, property, court, attorney which seem to indicate to be in the direction of legal issues. When looking at the overlapping Political Affairs, which is a general TAG apears as one of the big overlappers, together with operations and Assistance to citizens and Narcotics, while also the word Aricraft and the TAG Aeronautica are present. This TAG is not as clear as the above described but seems to be realted to legal issues affecting citizens and also property, which could explain the negative coefficient as higher number of legal cases could be representing a more difficult business enviroment.

The TAG “Trade Complaints and Disputes”, has as its main words contract, company, firm which would indicate that this is related to issues concerning comapnies and their activities. Also the TAG overlapping Investment, Commerce, Foreign Trade would reafirm what was seen in the word cloud, what is intersting is the difference between the nature of discussion between Trade Disputes and Property protection, while the first one seems to be dealing with trade business issues the later was more focused on legal concerns. In conclusion, this would confirm the negative sign of the coefficient seen as a higher number of trade disputes would dificult trade and lower it.

Finally, Emergency and Evacuation has as its main words personnel, evacution, american, employess, plan etc.. this would indicate that these cables are discussing issues related to emergency plans or situations in embassies which seens to correspond with the TAG name and also the overlapping that has as its main TAG’s Operations, Consular Affairs, personel, Assistance to citizens all issues that would be related to the evacuation of an embassy. We can then infer that when an embassy uses this TAG is discussing its plans in an emergency case and therefore could have a negative effect on trade.

Word Clouds Selected TAG’s

TreeMaps Overlapping Selected Cables

Discussion

This paper intent hs been to build on the existing literature by introducing the Gravity model in order to use a established model that explains trade flows, however it seems to be the case that there is further space for introducing the issue of government intervention in order to explain trade. We started wit the hypothesis that government could affect \(tau_{US,other}\), meaning the transaction costs that firms faced when deciding to trade or not with other country. What comes first to mind is the appearence of both positive and negative coefficents and issues that seems to be circumstancial and other that are related to the government activity.

For example the Investment TAG that as it has been explained has a clear relation with OPIC’s activities seems to be increasing trade flows. We could possibly find an economic basis for this issue, as OPIC engaged in the business of insurance, we could conclude that it is solving a market imperfection as described by Stiglitz & Rothschild (1976) were lack of perfect information and size prevent insurers from engaging in the insurance business and this limits sthe ability of market players to develop new business, and therefore OPIC could be solving this issue and allowing for more trade. This is not to say that OPIC is free to develop its investments in random countries, as there is a clear case of endogenity were countries that already trade with the US receive the organization atention and therefore trade more. However the case that the more Investment TAG’s the more a country trades and the case that the more OPIC’s cables a country receives the more it will trade seem to reinforce the idea that the government can play a positive role in trade by lowering transaction costs.

A second case is related to the negative effects on trade triggered by the government as seen in the Trade controls TAG which as it has been explained has a great overlapp with CoCom TAG. In this case we would be evidencing the increase in transaction costs by the government with companies that intended to trade with countries that the US government does not want to trade. These increases in transaction costs would not be directed at the specific products but penalties or fines that could incur if violating the rules established by CoCom. An extreme case was the Toshiba-konsberg incident, were a Japanese company with a Norwegian partner exported computers to the Soviet Union which an item prohibited by CoCom, as a result these companies were barred from exporting their products to the US which would greatly affect their business beyond any profits that could be made in the specific deal breaching CoCom’s rules.

Another TAG that seems to be realted to government intervention is Trade Expansion, which describes projects and visits by businesmen in which the State Deparment reports. Altough it is not possible to determine that this is clearly the effect of government activity there seems to be a positive relation between trade promotion activities and trade flows, however this does not mean that trade is generated but could mean that trade with an already exisiting tradding partner is increased.

Other TAG’s seem to be more circumstancial and less related to government activities, such as the emergency and evacuation TAG, in this case an increase in the number of cables containing this TAG would result in a decrease in trade. We could interpret this as the increase number of Cables would be reflecting a deterioration in the security level in the related country, and therefore a less secure enviroment for American citizens and embassies also diminishes trade flows with the US.

Another TAG that seems to affect trade flows in time, is trade disputes, meaning that the more cables associated with this TAG and a country tend to decrease trade over time. Meaning that trade disputes do affect volumes and therefore have a real effect besides rethoric and denounces, this could probe an important issue taking into account today’s US China relations and the rethoric involved.

Conclusion

TILL HERE IS DONE

Breusch-Pagan for Homoskedasticity

## 
##  Breusch-Pagan test
## 
## data:  reg0
## BP = 130, df = 6, p-value <0.0000000000000002

predicted = Actual - residual or Residual = Actual - predicted

##      iso3_d year   CoCom
## 2890    RUS 1976 18.0296
## 2740    POL 1976 15.5505
## 1215    GBR 1976 10.8822
## 815     CSK 1976 10.6890
## 1540    HUN 1976  8.3065
## 2865    ROU 1976  7.3084
## 1165    FRA 1976  6.5035
## 215     BEL 1976  4.1533
## 615     CHN 1976  3.8313
## 315     BGR 1976  3.5415
## 1815    JPN 1976  2.2215
## 1740    ITA 1976  2.0605
## 2490    NLD 1976  1.4488
## 2515    NOR 1976  0.8371
## 540     CAN 1976  0.7727
## 3140    SWE 1976  0.7083
## 165     AUT 1976  0.6439
## 940     DNK 1976  0.6439
## 1615    IRL 1976  0.6117
## 565     CHE 1976  0.4185

First regression log of exp


	Dependent variable:

	log_exp
	(1)	(2)	(3)	(4)	(5)

log(pop_d)	-0.200^***	-0.282^***	-0.251^***	-0.285^***	-0.313^***
	(0.062)	(0.062)	(0.063)	(0.064)	(0.065)

log(distw)	-0.586^***	-0.532^***	-0.623^***	-0.589^***	-0.612^***
	(0.161)	(0.158)	(0.162)	(0.162)	(0.162)

log(GDP)	1.197^***	1.278^***	1.224^***	1.218^***	1.226^***
	(0.057)	(0.060)	(0.067)	(0.067)	(0.067)

comrelig	3.355^***	3.238^***	3.435^***	3.423^***	3.327^***
	(0.581)	(0.567)	(0.575)	(0.573)	(0.573)

comlang_off	0.749^***	0.689^***	0.702^***	0.624^***	0.651^***
	(0.147)	(0.144)	(0.144)	(0.147)	(0.147)

landlock	-1.079^***	-0.827^***	-0.785^***	-0.730^***	-0.669^***
	(0.208)	(0.206)	(0.206)	(0.206)	(0.207)

TAG_Investment		0.352^***	0.305^***	0.294^***	0.268^***
		(0.080)	(0.085)	(0.084)	(0.085)

TAG_Trade_Control		-0.173^***	-0.165^***	-0.181^***	-0.175^***
		(0.037)	(0.038)	(0.038)	(0.038)

TAG_Trade_Expansion			0.230^**	0.192^*	0.151
			(0.102)	(0.103)	(0.104)

TAG_Property_Protection_Services			-0.101^*	-0.110^**	-0.118^**
			(0.053)	(0.053)	(0.053)

TAG_Animal_Plants_and_Wood_Prod				0.187^***	0.176^**
				(0.072)	(0.072)

TAG_Export_import_Bank					0.165^**
					(0.072)


Observations	822	822	822	822	822
R²	0.622	0.642	0.646	0.649	0.651
Adjusted R²	0.617	0.636	0.639	0.642	0.644
F Statistic	222.100^*** (df = 6; 810)	181.300^*** (df = 8; 808)	146.900^*** (df = 10; 806)	135.100^*** (df = 11; 805)	124.900^*** (df = 12; 804)

Note:	p<0.1; p<0.05; p<0.01

First regression log of imp


	Dependent variable:

	log_imp
	(1)	(2)	(3)	(4)	(5)

log(pop_d)	-0.126^***	-0.182^***	-0.155^***	-0.179^***	-0.191^***
	(0.036)	(0.036)	(0.036)	(0.037)	(0.037)

log(distw)	-1.150^***	-1.113^***	-1.178^***	-1.154^***	-1.163^***
	(0.094)	(0.092)	(0.094)	(0.094)	(0.094)

log(GDP)	1.042^***	1.095^***	1.039^***	1.035^***	1.038^***
	(0.033)	(0.035)	(0.039)	(0.038)	(0.038)

comrelig	-0.449	-0.527	-0.324	-0.332	-0.373
	(0.341)	(0.330)	(0.332)	(0.330)	(0.331)

comlang_off	0.399^***	0.358^***	0.372^***	0.317^***	0.328^***
	(0.086)	(0.084)	(0.083)	(0.085)	(0.085)

landlock	-1.293^***	-1.123^***	-1.092^***	-1.053^***	-1.027^***
	(0.122)	(0.120)	(0.119)	(0.119)	(0.120)

TAG_Investment		0.240^***	0.190^***	0.182^***	0.171^***
		(0.046)	(0.049)	(0.049)	(0.049)

TAG_Trade_Control		-0.116^***	-0.112^***	-0.124^***	-0.121^***
		(0.022)	(0.022)	(0.022)	(0.022)

TAG_Trade_Expansion			0.222^***	0.194^***	0.177^***
			(0.059)	(0.059)	(0.060)

TAG_Property_Protection_Services			-0.066^**	-0.073^**	-0.076^**
			(0.031)	(0.031)	(0.031)

TAG_Animal_Plants_and_Wood_Prod				0.133^***	0.128^***
				(0.041)	(0.041)

TAG_Export_import_Bank					0.070^*
					(0.042)


Observations	822	822	822	822	822
R²	0.798	0.812	0.816	0.818	0.819
Adjusted R²	0.795	0.809	0.812	0.815	0.815
F Statistic	531.700^*** (df = 6; 810)	435.900^*** (df = 8; 808)	357.100^*** (df = 10; 806)	329.300^*** (df = 11; 805)	302.800^*** (df = 12; 804)

Note:	p<0.1; p<0.05; p<0.01

Dickey Fuller for Unit Root(stationarity)

## 
##  Augmented Dickey-Fuller Test
## 
## data:  Panel.set$y
## Dickey-Fuller = -35, Lag order = 2, p-value = 0.01
## alternative hypothesis: stationary

## 
##  Breusch-Pagan test
## 
## data:  reg3
## BP = 79, df = 10, p-value = 0.0000000000007

regression with lags

## 
## ======================================================================================================
##                                 Lag 0       Lag 1       Lag 2       Lag 3       Lag 4       Lag 7     
## ------------------------------------------------------------------------------------------------------
## Dummy for diplomatic relations    6.70 ***    6.44 ***    6.07 ***    5.96 ***    5.86 ***    5.27 ***
##                                  (0.21)      (0.22)      (0.22)      (0.23)      (0.23)      (0.26)   
## log of population                -0.12 **    -0.13 **    -0.12 *     -0.10 *     -0.05        0.06    
##                                  (0.04)      (0.05)      (0.05)      (0.05)      (0.05)      (0.05)   
## log of Area sq Km                -0.07 **    -0.06 *     -0.06 *     -0.07 **    -0.09 ***   -0.15 ***
##                                  (0.02)      (0.03)      (0.03)      (0.03)      (0.03)      (0.03)   
## Log of distance                  -0.72 ***   -0.66 ***   -0.64 ***   -0.66 ***   -0.63 ***   -0.69 ***
##                                  (0.10)      (0.10)      (0.10)      (0.10)      (0.11)      (0.11)   
## log of GDP                        1.08 ***    1.09 ***    1.08 ***    1.08 ***    1.06 ***    1.06 ***
##                                  (0.04)      (0.04)      (0.04)      (0.04)      (0.04)      (0.04)   
## % of common religion              0.62        0.61        0.69        0.71        0.94 *      1.33 ** 
##                                  (0.36)      (0.38)      (0.39)      (0.39)      (0.40)      (0.41)   
## Dummy for Common language         0.41 ***    0.44 ***    0.45 ***    0.44 ***    0.43 ***    0.45 ***
##                                  (0.09)      (0.10)      (0.10)      (0.10)      (0.10)      (0.10)   
## Dummy landlock                   -0.82 ***   -0.92 ***   -1.02 ***   -1.07 ***   -1.16 ***   -1.22 ***
##                                  (0.12)      (0.13)      (0.14)      (0.14)      (0.14)      (0.14)   
## Tag Trade controls               -0.14 ***   -0.15 ***   -0.14 ***   -0.14 ***   -0.14 ***   -0.14 ***
##                                  (0.02)      (0.02)      (0.03)      (0.03)      (0.03)      (0.03)   
## ------------------------------------------------------------------------------------------------------
## R^2                                                                                                   
## Adj. R^2                                                                                              
## Num. obs.                       870         870         870         870         870         870       
## ======================================================================================================
## *** p < 0.001, ** p < 0.01, * p < 0.05

From the table above we can see that when lagging the Trade controls cables as an index of every country with itself the results are positive meaning that the more cables the more trade. Also there seems to be a higher impact 3 years after the increase, meaning increases in cables tend to have their bigger impact after 3 years, altough the difference in magnitude do not seem to be highly relevant 0.11 vs 0.15. The most interesting issue, comes from the fact that when comparing the impact of the tags when comapred in a relative form meaning how many tags per country as % of total the coefficent was negative, meaning the more they were talking about you in comparrison to the others the less trade you were getting, while in this case it is the opposite effect.

Here I am trying to do a kind of granger test, the problem is that future trade volumes are kind of perfectly explained by past ones.

## 
## =============================================================================================
##                     Model 1      Model 2     Model 3      Model 4     Model 5      Model 6   
## ---------------------------------------------------------------------------------------------
## lag1.log_total_avg     0.98 ***    0.99 ***                                                  
##                       (0.01)      (0.01)                                                     
## lag1.TAG_90                        0.01                                                      
##                                   (0.01)                                                     
## lag2.log_total_avg                              0.97 ***    0.98 ***                         
##                                                (0.01)      (0.01)                            
## lag2.TAG_90                                                 0.01                             
##                                                            (0.02)                            
## lag3.log_total_avg                                                       0.95 ***    0.96 ***
##                                                                         (0.01)      (0.01)   
## lag3.TAG_90                                                                          0.02    
##                                                                                     (0.02)   
## ---------------------------------------------------------------------------------------------
## R^2                    0.94        0.94         0.91        0.91         0.87        0.88    
## Adj. R^2               0.93        0.94         0.91        0.91         0.87        0.88    
## Num. obs.           1015         870         1015         870         1015         870       
## =============================================================================================
## *** p < 0.001, ** p < 0.01, * p < 0.05

Robustness check

In order to check that the TAG’s selected are special in the sense that they reflect some topic that it is indeed related to trade and economcis and not just a random issue that contains no value, an excercise was done with all TAG’s. There is a fair amount of clustering among the TAG’s specially the ones dealing with European security issues such as NATO, Warsaw pact, confences on security. These TAG’s put a special emphasis on the Socialist block nations and therefore do not provide much value besides the fact that there was a high level of tension and a blockade and therefore no trade.

##    Min. 1st Qu.  Median    Mean 3rd Qu.    Max. 
##   0.765   0.777   0.778   0.780   0.782   0.806

##     Var1 Freq.x Freq.y tot atg_idi
## 89    18    181     15 196     245
## 167   71    124     68 192     172
## 196   98     95     94 189     257
## 8    106     86    102 188     240
## 161   66    123     63 186     169
## 95   185     12    170 182     311
## 103  192      6    176 182     171
## 136   43    142     40 182      90
## 141   48     96     34 130      83
## 35   130     34     76 110     243
## 61   154     10     43  53     160
## 79   170      5     41  46    1863
## 117   26     37      5  42     111
## 135   42     29      5  34     128
## 46   140      7     18  25     123
## 187    9     24      0  24      53
## 131   39     20      3  23      64
## 10   108      7     15  22     161
## 138   45     18      4  22     140
## 67    16     20      1  21     135
##                                                title
## 89                North Atlantic Tready Organization
## 167 Conference on Secutiry and Cooperation in Europe
## 196          Overseas Private Investment Corporation
## 8          Mutual and Balances Force Reduction Talks
## 161        Coordinating Committee on Export Controls
## 95                          Warsaw Pact Organization
## 103            Council of Mutual Economic Assistance
## 136                         Strategic Trade Controls
## 141                                      Investments
## 35                     North Atlantic Council (NATO)
## 61        Conference of the Committee on Disarmament
## 79                                  Voice of America
## 117                     Arms Control and Disarmament
## 135                                      Environment
## 46           Propaganda and Psychological Operations
## 187                    Trade Expansion and Promotion
## 131                      Consular Affairs -- General
## 10     Committee on the Challenges of Modern Society
## 138                Technology and Science -- General
## 67                                          Refugees

All Word clouds

Trade Missing Values

Trade data refers as Imported by the United States and exported by the US to from/other countries.

Missing values per year
Year	Freq
1962	7
1963	7
1964	7
1965	7
1966	5
1967	4
1968	4
1969	4
1970	3
1971	2

Frequency missing countries
iso3c	Name	Freq
MWI	Malawi	4
ZWE	Zimbabwe	4
FJI	Fiji	5
BLZ	Belize	8
RWA	Rwanda	9
ARE	United Arab Emirates	10
BGD	Bangladesh	10

Malawi(MWI) gained independence in 1964, therefore only has trade data starting in 1966. Zimbawe(ZWE) gained independence in 1964, therefore only has trade data starting in 1966. Fiji(FJI) Missing data 1962-1966 not clear why. Equatorial Guinea(GNQ) several years missing, changed to 0 as it broke relations with the West in 1972. Belize(BLZ) Missing data 1962-1969 not clear why. Rwanda(RWA) Missing data 1962-1970 not clear why. United arab emirtates(ARE) missing data 1966-1971 year in which it gained independence. Bangladesh(BGD) missing data 1966-1971 year in which it gained independence.

Overlapping on TAGS and Documents

Cables with only one TAG
Title	TAG.id	Freq
Economic Alert List Reporting	71	1
General Agreement of Tariffs and Trade	190	1
International Bank for Reconstruction and Development	200	1
Export-Import Bank of the United States	312	1
Organization for Economic Cooperation and Development	255	6
Chemical Industry and Chemical Products	72	9
East-West Trade	78	14
Economic Integration and Collaboration	73	17
Coordinating Committee on Export Controls	169	20
Science Grants	108	40
National Development	113	65
Plant, Animal, and Wood Products	88	105
Strategic Trade Controls	90	159
Commerce, Industry, and Industrial Products	82	174
Budget Services and Financial Systems	15	179
Metals and Minerals	86	188
Waterborne Transportation	94	201
Economic Conditions, Trends, and Potential	74	211
Trade and Investment Opportunities	58	222
Trade Complaints and Disputes	51	248
Technology and Science – General	140	321
Investments	83	365
Military Assistance and Sales	97	393
Fishing	80	427
Financial Services	568	435
Technology	139	448
Environment	128	485
Economics – General	81	567
Agriculture and Forestry	68	715
Business Services – General	55	825
Foreign Assistance	69	1109
Travel by U.S. and Foreign Businessmen	59	1289
Financial and Monetary Affairs	79	1463
Foreign Trade	92	1790
Property Protection Services	66	1878
Trade Expansion and Promotion	53	2621
Business Proposals and Inquiries	56	3302
Consular Affairs – General	64	9787

obervations missing by Tags Cluster and Country
cluster	0 obs missing	1 obs missing	2 obs missing	3 obs missing	4 obs missing	5 obs missing	6 obs missing
1	24.12	13.78	11.43	11.55	13.17	12.20	13.74
2	84.37	6.15	3.47	1.95	1.66	1.16	1.29
3	52.81	15.58	9.45	7.30	5.31	4.86	4.67
4	10.80	5.97	5.07	4.40	6.00	22.07	45.77
5	0.00	0.70	37.90	31.70	12.40	5.50	11.70

obervations missing by Tags Cluster and year
cluster	1974	1975	1976	1977	1978	1979
1	58.79	43.68	38.39	40.29	43.97	42.18
2	7.31	7.03	6.34	5.84	5.70	6.71
3	27.46	23.16	20.32	20.16	19.80	19.07
4	79.77	72.18	71.26	68.97	74.25	71.49
5	100.00	99.31	37.93	33.10	33.79	15.17

Among the 5 clusters we can identify two main categories, Cluster 1 is compossed of Tags that contain all countries and are almost complete for all years. Cluster 2 is composed of Tags that are present almost completetly among all countries but not as widespread as in the case of Cluster 1, also the lack of presence over the years is constant around 20% of countries. Cluster 3 is compossed of TAG 33( U.S. Overseas Direct Employment Program) alone, this TAG will be dropped as it is not constant over the years just starting to be used in 1975. Cluster 4 is composed of TAGS that have constant use across time but vary across countries with just 25% of the countries being present completetly. Finally, cluster 5 is the opposite from Cluster 1, is composed of TAGS were for some countries there are compelte observation but the norm is that for the majority there are no observations at all. This seems to be a TAGS that deal with specific issues not common among all countries.

Tags U.S. Overseas Direct Employment Program(33), International Labor Organization(216), Business-Commercial Publications and Libraries (57), Business Services Reports(50), UN Conference on Trade and Development (273) will be dropped because they do not have consistent appearence along the years. In some cases there does not seems to be a clear cause for this other than probably changing patterns of use. For ILO, and UNCTAD this change is due to the fact that theh US left the ILO in 1977 and there were two UNCTAD conferences in 1976 and 1979 respectively. Therefore as this issues affect the TAGS they were dropped.

A Cable(comunication between two entities of the State deparment), could have attached more than one TAG. TAGS can be seen as Hashtags attached to a post on social media, the post which is the message can have 0 or more hastags. The more TAGS a message has the more topics been discussed in the Cable. The table above shows cables that have one TAG meaning that when they were classified they only deal with the topic of the TAG, this selection was done among all TAGS and not only the ones pertraning this study. As a rule of tumb we will drop all Tags that have less than 321 observations of cables were they are the only TAG. This criteria is based on the fact that the dataset has 6-7 years and 145 countries meaning that having at least one cable per year for each country would represent 870 cables therefore close to half of it seems to be a reasonable benchmark after reviewing the frequency on the table above.

From the dropped cables I will keep the ones with at least 100 cables were they are the unique cables, as they could be referring to specific topics that only refer to a limited number of countries.

Also Asian Development Bank(45), International Monetary Fund(219), Overseas Private Investment Corporation(257), Canal Zone(378) do not appear on the table above as they have no unique cable were their TAG is the only one.

Above all this needs to be done in order to prevent an issue of multicollinearity among the cables, as if all cables of one specific TAG are shared with other Tags this would amount to a linear combination of the other TAGS. Therefore behind the the rationale of dropping TAGS with the less number of distinct tables lies the idea that these TAGS represent less unique topics that are reflected or comprehended inside the other more “unique” TAGS.

Thesis

nathan fabius