![Sample our Politics & International Relations journals, sign in here to start your FREE access for 14 days]({"type":"image" , "src" : "/sda/5947/TOC-Subject-Sample-2021-Politics-International-Relations.jpg"})
Abstract
This paper explores the correlates of Russia’s aggressive international policy and argues that rising oil revenues increase the aggressiveness of presidential foreign-policy rhetoric. Using content analysis and machine-learning techniques, I generate a measure of aggressive discourse as the share of anti-Western sentences in Russian presidential speeches delivered between 2000 and 2016. These are analyzed using OLS regression with lagged dependent variables. I conclude that the aggressiveness of foreign-policy rhetoric in Russian presidential speeches positively correlates to oil prices. I also find no support for alternative explanations linking hawkish foreign policy to NATO expansion or domestic legitimacy concerns.
ACKNOWLEDGMENTS
I would like to thank Mikhail Nefedov, Ulyana Sentsova, and John Lyell from the National University Higher School of Economics, as well as Neal R. Lewis from Bloomberg L.P. for their assistance and collaboration.
Notes
1. Countries with an annual net oil export of at least 10 percent of GDP (Colgan Citation2014a, 199).
2. Colgan (Citation2014a) argues that the combination of oil income and Putin’s aggressive preferences, along with his ambition to return Russia to its status as a superpower, make Russia more likely to instigate international conflict.
3. The growth in defense expenditure is concealed by using secret expenditure sections in the state budget, while the share of classified articles is increasing annually (Jushkin Citation2018). In 2009, 10 percent of federal budget expenses were classified (including defense); in 2015, 20 percent of federal budget expenses were classified; in 2016, 22 percent of federal budget expenses were classified (nearly doubling since 2014); in 2017, of the proposed increases of total budget expenditures, only 53 percent were non-classified.
5. The data are structured quarterly since most of my explanatory variables are available on a quarterly basis, including oil and gas exports, presidential approval ratings, and Russia’s economic growth. In addition, the average quarterly share of aggressive speeches provides for more easily interpretable results, since there are not enough observations for monthly averages.
11. An alternative way to test for my hypotheses is to structure the dependent variable not as a share of aggressive sentences per quarter, but rather as a count variable (number of aggressive sentences in a given speech). In this case the dependent variable contains 457 observations (based on the number of speeches in the dataset). When structured as a count variable, my dependent variable becomes a discrete dependent variable, and the use of least squares is not appropriate in this case. Because of overdispersion of my dependent variable, I used the negative binominal regression rather than the poisson distribution for my analysis. In this specification of my dependent variable, the “Oil” hypothesis again found the strongest confirmation. The coefficients to the lagged oil and export variables were positive and statistically significant at 95 percent and 90 percent confidence levels in most models. Combined with the above results, these findings confirmed my original expectation: higher oil prices are associated with the more assertive foreign-policy rhetoric of Russian presidents.