Explaining American Intervention in Foreign Wars from 1945 - 2019 Utilizing Petroleum Prices
***Tables and equations not included***
Abstract
The role of energy as a cause for war is often seen as controversial, or even conspiratorial. This research paper attempts to show the influence of oil in the American government’s decision to militarily intervene in a given conflict. This is done by measuring the effects of domestic oil prices between 1946 and 2019. I also measure the effect of having different levels of natural resources as a percentage of GDP and oil as a percentage of GDP. I also see if the ideology – measured by political party control – of the United States House of Representatives, United States Senate, and the United States Presidency effects the government’s decision to intervene in a foreign war, and what form that intervention will take.
Keywords
Third-party intervention, oil, natural resources, war
Introduction
A popular conspiracy theory that has developed since 9/11 is that the United States has been militarily intervening in third world countries that have oil to reduce the price of oil, or to increase the amount of oil available on global markets. This conspiracy theory is not as far-fetched as it might seem. Oil is the most important input in the production of goods and services, so it makes sense that the United States would want a lot of it – and at a good price.
Previous literature on the topic of oil and its influence in the decision-making process a country makes when intervening militarily in other countries’ affairs is fairly scarce. My research is model off the work of Mi Yung Yoon’s article, Explaining U.S. Intervention in Third World Internal Wars. Her work focuses on the American Government’s decision-making process by looking at the actions of the Soviet Union and its allies. My Research model is an attempt to modernize Mi Yung Yoon’s model by changing my focus from the Soviet’s actions to the domestic oil price in the United States and the reliance of foreign states on oil and natural resources.
This research is important because it sheds light on what goes on in the decision process a country goes through when deciding to intervene militarily. Mi Yung Yoon does this by explaining the decision-making process through the Soviet Union and its actions. She is assuming that the greatest influencer in the American decision of intervention is the Soviets. Due to the dissolution of the Soviet Union, this can no longer be true. Thus, researchers need to uncover what variables are key in the decision-making process of intervening in a foreign war.
I present a model that attempts to explain the influence of the price of oil on the American government’s decision to intervene in a foreign war. This model also considers how important oil revenues are for the intervened in-country and how important natural resources (including oil) is for the intervened country. The model also uses several binomial dummy variables, sequestered by year. These include the existence of the Soviet union – coded 1 for pre dissolution and 0 for after dissolution – and the political ideology of congress and the presidency. Where the majority ideology is coded as 1 for the Republicans and 0 for the Democrats.
I hypothesize that higher domestic oil prices in the United States encourage the United States government to intervene in foreign wars, and that nations where oil represents a large proportion of income will be intervened in more. I also expect that nations who have a large proportion of income being generated from natural resources will be intervened in more. I do not expect the political ideology of the American Congress or the American Presidency to make a big influence on the decision to intervene in a foreign war.
Theoretical Framework
The theoretical framework for my paper is derived from Jeff D. Colgan’s article Fueling the fire: Pathways from oil to war and Peter Zeihan’s book, The Accidental Superpower. Both works show the importance of petroleum in both global and American foreign affairs.
Jeff D Colgan creates three casual mechanisms that all influence conflicts. Within these mechanisms are individual mechanics that influence that conflicts and what type of conflict will occur. Table 1 is Colgan’s table that shows each of these mechanisms and mechanics (Colgan 2013).
Several of these mechanisms need to be recognized for my paper research. The first two within the casual mechanism, “Ownership and Market Structure”, are especially important. Oil reserves increase the payout for any given country, such as the United States, for militarily intervening in the foreign affairs of an oil-producing country. Oil importers also have an incentive to protect the oil markets from external conquests, which can create problems such as one country getting to large a market share of global oil supplies or by the conquest causing short term fluctuations in oil prices.
The second causal mechanism, Producer Politics has a mechanism involving civil wars. It states large oil deposits create conditions for civil war. This agrees with the findings of Michael G. Findley’s article, Lootable Resources and third-party intervention into civil wars. His article shows that when lootable natural resources are being affected by civil conflict, foreign nations will intervene to secure those resources for themselves (Findley 2015). Combining both Findley and Colgan’s theories, I will say that oil is a relatively lootable resource that can help encourage third-party intervention into a civil conflict.
The other important mechanic for my paper is within the “Consumer Access Concerns” mechanism. The transit of oil and the struggle to control the transit of oil creates tensions that escalate to open conflicts. The United States, as the “largest importer of energy for at least the last thirty years (Zeihan 2016), has taken the responsibility of protecting global shipping lanes from threats. Thus, allowing for the smoother transit of oil across global markets.
Peter Zeihan’s Accidental Superpower shows how some of Jeff Colgan’s mechanisms are true. Important for my research paper is that “the Americans do not protect the Persian Gulf kingdoms and emirates so that the Americans can use Middle Eastern oil, but so that their Bretton Woods partners in Japan, Korea, China, Taiwan, Thailand, India and Pakistan can“ (Zeihan 2016). The United States, prior to the fall of the Soviet Union, relied on oil from producer nations so that the United States and its NATO allies could build-up, or maintain, their economies and militaries to combat the Soviets. As time went on and the United States began to rely on oil imported from nearby NAFTA countries and itself (Zeihan 2016). After the fall of the Soviet Union in 1991, the reasons for protecting the transit of oil has been steadily decreasing. This also lowers the payout from Colgan’s resource war mechanic, because the payout for a country that is less reliant on oil imports is less than the payout for a country that relies heavily on the importation of petroleum.
Concerning the United States, the following points should be considered to be affecting the American Government’s decision to intervene militarily in the foreign affairs of another country.
1) That country has large oil reserves and/or can produce large quantities of oil.
2) Any country that attempts to conquer an oil-producing state, any country that has a large market share of oil production, and any country that gains a large oil production capability from conquest.
3) Oil creates conditions for civil war, and natural resource-rich – oil included – nations are more likely to be intervened in by a foreign power.
4) Protection of oil transport routes. This is a factor that decreases in importance as time goes on.
The economic theory to back up the above theoretical framework is that the supply of American petroleum was greatly lower than the demand for petroleum between 1946 and 2018. However, the supply of petroleum has been increasing steadily since 1946. Even so, the most economically viable option for the United States was to import oil from oil producers around the world, mostly OPEC nations. This gave the United States a direct incentive to the United States to militarily intervene in any oil-producing nation that caused leftward supply shifts in oil. Since the United States military was built up from being in the cold war, the United States did not have to incur extra military buildup costs. This made it easy to act on those incentives to intervene for the sake of petroleum. This intervention was the most economically viable option the United States had to shift the global supply curve of oil to the right.
Data
I used a combination of panel data and binomial dummy variables for my data. I also used data from Mi Yung Yoon’s article, Explaining U.S. Intervention in Third World internal Wars, 1945-1989. The data I used was adequate for what I wanted to do with this project and was adequate to test the hypotheses outlined in the above sections. This section will outline what data I used, problems with the data, and missing variables.
Mi Yung Yoon listed every war that occurred during that period and the level of American Involvement in each war. The level of intervention was coded as 0 for no intervention, 1 for economic intervention, 2 for indirect military intervention, and 3 for direct military intervention. If multiple types if intervention was used by the United States over the course of the war, then the highest level recorded was used. Covert military operations, such as intervention via the Central Intelligence Agency, was not included into the dataset due to complications with collection and coding. Civil wars for independence were also excluded from her data set. I used Mi Yung Yoon’s list of wars and the level of American intervention for my project. I extended her dataset from to include up to 2016 by attempting to follow her data guidelines. I used Wikipedia as a baseline to see what wars have occurred, as well as Infoplease and Britannica. The following tables are Mi Yung Yoon’s table that shows the level of intervention in each war.
The focus of my project was to see if the price of oil was an influence in the American Government’s decision to intervene in a foreign war. I used data that gave the average yearly domestic price of oil in the United States, inflation adjusted to February 2019 prices. This data is from inflationdata.com, who have compiled the domestic inflation adjusted price for a barrel of oil from 1946 to today.
Two other important datasets are oil rents as a percentage of GDP and total natural resource rents as a percentage of GDP. These two datasets allow me to see how much a country relies on either oil or natural resources for their incomes. It also allows me to see if a country has multiple natural resources that have value. For example, if natural resource rents are 20% while oil rents are 10% then there must be other valuable resources in that country that could make up 10% of that country’s income. This is important because it could mean that another resource other than oil is responsible for the intervention.
Both of these datasets are from the world bank and thus were created similarly; the same collection methods. This was a problem because these datasets had a timeframe from 1970 to 2017. Different countries also had different data ranges where data was collected from and some countries had missing data if that country was experiencing some sort of trauma. For example, many of the post-Soviet states, including Russia, did not have data for the immediate years following the collapse of the Soviet Union. This means that for some outlier cases, due to extreme conflict or traumas, data was not available for my econometrics. I also had to deal with missing data from 1945 to 1970. This meant that STATA had to run regressions with data with a large number of missing values for two variables. I had to destring both the variables that were created with this data; turning them both into numeric variables so that STATA could know how to regress them.
I have sequestered my data into two different types, political and economic. The above data sets have been for my economic variables. I measured the political ideology of the United States House of Representatives, the United States Senate, and the Presidency of the United States by creating binomial dummy variables for each one. If one of them had a republican majority, it was coded as 1 while a democratic majority was coded as a 0. I then sorted it by year and attached them to the other data. I got the United States house data from house.gov and the data on the United States senate from senate.gov. The data on the United States Presidency is from the comprehensive list created by the guardian.
The data collected is more than adequate to test my hypotheses and allowed me to see the influence of several different economic and political variables on American intervention in foreign states.
Econometric model
I created two separate econometrics models to see how each variable interacts with the dependent variable of intervention. The first econometrics model that I used was an OLS regression model where instead of regressing a model where intervention is measured the same way Mi Yung Yoon does; where intervention is coded either 0, 1, 2, or 3. I created another intervention variable where no intervention is coded as 0 and any form of intervention, whether 1, 2, or 3 is coded as 1. This regression should tell us the influence of each variable has on whether or not the United States will intervene at all.
The other econometrics model that I created was an ordered probit model. I used this model because it was used by Mi Yung Yoon in her article about the American Government intervention in foreign wars. Since my dependent variable is essentially just an extended version of her depended variable, I thought it would be best to use the same type of model that she employed in her work. An ordered probit model successfully sequesters the levels of intervention and should give me a better understanding of what influences the American government’s decision to intervene in another country.
These two econometric models were appropriate at analyzing the data that I had available. The econometric models were also suitable for conducting tests on the hypotheses that were presented earlier in this paper.
Limitations to econometrics
Due to the nature of my dependent variable, any econometrics model that I create will be insufficient to meaningfully produce a result that shows a definitive influencer into why the American Government intervenes in any given foreign war. The decision-making process most likely incorporates so many variables that it would be impossible to create a model with the capability of adequately, or reliably, predicting what causes the American Government to intervene in a war. The variables that cause that decision are far more numerous and change drastically with every nation or conflict. Previous works, such as Oil Above Water by Vincenzo bove, discuss similar problems with methodology. Certain key variables were emphasized while others weren’t, or even kept out; the model that was created was limited in the scope of what could and could not be included (Bove 2016).
That being said, I think that it is still important to attempt to find variables that could potentially be common among many conflicts. This is why I decided to choose domestic oil prices over many other variables. There also is previous literature on the effect of both petroleum and other natural resources on the decision-making process of intervening in another country.
Limitations to data
My data contains a lot of dummy variables. Things like the political ideology of Congress and the presidency and intervention. This causes variance problems when doing econometric regressions. Another problem, as stated in a previous section, is that two of the data sets that are were not used to create dummy variables lacked data before 1970. This means that I have a 25-year time period with very no variance in the data, which could cause problems. This could cause the results of my project to be off target. However, I think the results should give a good idea of the trends of the variables that are affecting the American government’s decision-making.
Results
I will first present the summary statistics of my data. I will then show and discuss my binomial regression statistics, followed by the results of the ordered probit model.
The descriptive statistics tell us several things about the data. It tells us that the average level of intervention is somewhere between economic intervention and indirect military intervention. The binomial intervention variable shows that there is much more intervention, of any kind, than no intervention. The mean average domestic oil price does not have much importance when just looking at the descriptive statistics. The average oil price is more important when analyzing the regressions. Average natural resource rents and average oil rents show that on average a country has more natural resources that have a value that is not oil, however, oil still constitutes a majority of the natural resources sold on the market.
The table on the following page is an OLS regression with the dependent variable being a binomial of intervention. Where 0 is no intervention and 1 is any form of intervention. The table shows that only gives a minuscule impact on the American government's decision to intervene in another country’s foreign war. It is telling us that it does not affect intervention. The higher natural resource rents decrease the likelihood of any form of intervention, while higher oil rents very slightly increase the likelihood of any form of intervention. The interaction effect is very small and can be ignored. The existence of the Soviet Union decreased the likelihood of any form of intervention. The political ideology of the Presidency and the United States Senate do not influence the likelihood of intervention at all. While the house has a slightly higher likelihood to intervene if the United States House is Republican.
The ordered probit model gives slightly different results, shown in the table below. The average oil price has a larger effect in the ordered probit model, but it still has the possibility of giving a negative effect. The highest positive effect that could be reasonably given to the variable is still not that much. Natural resource rents as a percentage of GDP tend to have a negative effect on the intervention. Oil rents as a percentage of GDP tends to have a large effect on intervention, and the oil interaction effect is also large. The Soviets exists also has a negative effect on intervention, just like in the regression binomial intervention. The United States Senate has a large effect on intervention when the Senate is ideologically Republican. The United States House of Representatives, having a majority Republican ideology decreases the likelihood of intervention. The United States Presidency does not have any effect when it’s held by either party.
Conclusions
My conclusion is that the average domestic oil price in the United States does not affect the United States Government’s decision to intervene in a foreign war. This conclusion is contradicting what I thought would happen. According to my theory, the United States Government should be intervening more in oil producing countries when the supply of oil is not enough to satisfy American oil demand. The results of my project show that the United States is more willing to intervene in nations that are more reliant on oil revenues, which is consistent with what I thought would happen. This is consistent with findings of other researchers, while also being contradictory.
Jean- Francois Maystadt’s paper, Mineral Resources and Conflicts in the Democratic Republic of Congo and Michael Findley’s paper, Lootable resources and third-party intervention into civil wars both show that natural resources are an important factor in the decision to militarily intervene in a foreign country’s affairs (Finley 2015)(Maystadt 2013). Both found that natural resource boosted the payout from intervening. This is not found in my research. My research indicates that having a high level of income as a percentage of GDP from natural resource decreases the likelihood of intervention or has no notable effect. What I do find is that a country that has a high level of income as a percentage of GDP from oil revenues has a higher likelihood of intervention. This is consistent with Jeff Colgan’s article, Fueling the Fire: Pathways from Oil to War, where he notes that oil creates a higher payout for the intervening party due to petroleum’s importance in the modern economy (Colgan 2013).
One possible explanation for why percentage of GDP from oil revenues promoted intervention and the percentage of GDP from natural resources did not is that I lacked data on both variables before the 1970s. It could be that with the variance from before 1970, I would have gotten results that would be more expected. Future research should attempt to rectify this with more data.
Another possible conclusion is that there are so many variables that go into the American Government’s decision that any one variable that is analyzed will also result in a small coefficient. Because there are so many different variables effecting the decision, a positive coefficient between (-0.00061, 0.00535) might actually be a sizeable effect. This, however, is unlikely, as the coefficients on the percentage of GDP from oil revenues would be astronomically high in this case.
In the binomial regression model, the political variables show the ideology of the United States Congress and the United States Presidency effect the decision to intervene. In the binomial regression model, the United States Presidency being held by a Republican slightly decreased the likelihood of intervention. The ideology of the United States Senate had no real effect and the ideology of the House of Representatives mostly increased the intervention when it was held by the Republicans. Because the 95% confidence interval can be positive or negative, it could depend, but for the majority of this interval it is a positive coefficient.
In the ordered probit model the United States Senate greatly increases the likelihood of the United States intervening in another country. This likelihood decreases when the United States House of Representatives was controlled by the Republicans. The United States Presidency controlled by the Republicans caused an effect that could be positive or negative in a 95% confidence interval. This means that the effect could be either for or against intervention in a foreign war.
One final possibility is that the United States intervenes on the behalf of other countries. Peter Zeihan’s book, the Accidental Superpower, demonstrates how the United States has been securing oil from oil rich nations for not only themselves, but also their NATO allies (Zeihan 2016). This would explain why the domestic price of oil in the United States does not affect the likelihood of American military intervention in a foreign war. It also would explain why oil rents are important in the decision-making process.
Future research
Future research should include a way to measure the intensity or devastation caused by the conflict. It could be that oil prices, large scale oil production, or large oil reserves makes the conflict more intense. It could be that the United States tends to intervene in countries that are currently in very intense conflicts, and my model does not account for that.
Bibliography
Colgan, Jeff D. "Fueling the Fire: Pathways from Oil to War." International Security 38, no. 2 (2013): 147-80. www.jstor.org/stable/24480933.
Findley, Michael G. and Josiah F. Marineau. 2015. "Lootable Resources and Third-Party Intervention into Civil Wars." Conflict Management and Peace Science 32 (5): 465-486. http://proxyau.wrlc.org/login?url=https://search-proquest-com.proxyau.wrlc.org/docview/1777459072?accountid=8285.
Zeihan, Peter. The Accidental Superpower: the next Generation of American Preeminence and the Coming Global Disorder. New York: Twelve, 2016.
Bove, Vincenzo, Kristian Skrede Gleditsch, and Petros G. Sekeris. “‘Oil above Water’: Economic Interdependence and Third-Party Intervention.” Journal of Conflict Resolution 60, no. 7 (October 2016): 1251–77. doi:10.1177/0022002714567952.
Jean-François Maystadt, Giacomo De Luca, Petros G. Sekeris, John Ulimwengu, Mineral resources and conflicts in DRC: a case of ecological fallacy?, Oxford Economic Papers, Volume 66, Issue 3, July 2014, Pages 721–749, https://doi-org.proxyau.wrlc.org/10.1093/oep/gpt037
The Word Bank, Total Natural Resource Rents (% of GDP). World Bank, 2011. The World Bank. https://data.worldbank.org/indicator/NY.GDP.TOTL.RT.ZS
The Word Bank, Oil Rents (% of GDP). World Bank, 2011. The World Bank. https://data.worldbank.org/indicator/NY.GDP.PETR.RT.ZS?end=2017&start=1971&view=chart
Tim McMahon, Annual Average Domestic Crude Oil Prices. Inflationdata.com, 2019. Inflationdata.com. https://inflationdata.com/articles/inflation-adjusted-prices/historical-crude-oil-prices-table/
“Party Division.” U.S. Senate: Party Division, January 3, 2019. https://www.senate.gov/history/partydiv.htm.
“Party Divisions of the House of Representatives, 1789 to Present: US House of Representatives: History, Art & Archives.” Party Divisions | US House of Representatives: History, Art & Archives. Accessed December 11, 2019. https://history.house.gov/Institution/Party-Divisions/Party-Divisions/.
Appendix
Intervention = B0 + B1Average Oil Price + B2Natural Resource Rents + B3Oil Rents + B4Senate Ideology + B5House Ideology + B6President Ideology + B7Soveits +e
