Revolutions and International Cooperation: a Quantitative Analysis

Within international relations analysis, there is a growing body of literature that has shifted its focus onto regime-specific conditions and events that impact state behavior on the international stage. These newer studies provide a more nuanced way to understand interstate behavior than did the broad structural analyses (see Walz 1979 and Walt 1987) that previously dominated the literature on this topic. For my project, I select a specific domestic event, political revolution, and analyze the ways in which this occurrence affects alliance formation. I plan to examine whether the similar-regime effect (whereby states are less likely to pursue alliances with potential partners whose polity scores differ by a greater amount) is exacerbated in states that have undergone revolution. I will draw primarily off of two influential studies that attempt to analyze similar relationships.

In “Reliability, Reputation, and Alliance Formation,” authors Crescenzi, Kathman, Kleinberg, and Wood argue that the heads of state and crucial to the formation of that state’s reputation. Potential partners, when the state seeks alliances in the future, will factor the state’s reputation by considering its past performance on alliance commitments. The authors find that a positive reputation, that is a reputation deemed “reliable,” has a positive and significant effect on that state’s probability of being sought out by other states that are pursuing alliances. This indicates that perceptions of potential partners play a provable role in the formation of alliances. They also note that polity difference has a negative effect on alliance formation when the sample included in the ATOP data set, as well as a negative and significant effect on alliances in models that included all alliances from 1816-1913, all alliances from 1914-1945, and bilateral alliances from 1816-2000. The authors look at this variable which they call polity difference, as well as a variable they call joint democracy, which codes for whether all states involved in the alliance have a polity score of 5 or higher. Beyond these two variables, however, the authors do not endeavor to uncover the alliance behavior of post-revolutionary states, which could be different from that of other states contained within this dataset.

In fact, Colgan and Lucas argue in their paper “Revolutionary Pathways: Leaders and the International Impacts of Domestic Revolutions” that revolutions do have both short and long-term impacts on a state’s international relationships. These authors explore the multiple pathways through which revolutions can affect three dependent variables that will affect its international behavior: the onset of economic sanctions against the state, the domestic economic growth of the state following revolution, and the state of its alliances. For both alliance onset, meaning the initiation of new alliances after a revolution, as well as alliance change, which denotes whether the revolutionary states withdraws from or breaks off existing agreements, the authors find that revolutionary leadership negatively affect existing alliances, and that revolutionary events, in the short term (coded as within 3 years of the revolution) motivate leaders to seek new alliances as perceptions change and new friends are sought. This finding indicates that revolutions have a significant effect on the state’s international positioning and relationships.

Whereas Colgan and Lucas find that a state that has just undergone a revolution is more likely to and Crescenzi et al. reveal that states are more likely to form alliances with states who are more similar in their regime type, I am interested in tying these two ideas together to examines whether revolutionary states in particular are more likely to form alliances with other revolutionary states. I define revolutionary states in both cases with the 3 Year Revolutionary Period variable (revperiod3), which takes a value of 1 for the year in which a revolution took place, as well as for the two subsequent years, and 0 for all remaining years. I use this variable rather than that of Revolutionary Leader (coded revolutionaryleader) in order to identify revolutionary states because Colgan and Lucas find that revolutionary leaders are in fact 31.1% less likely to form an alliance, whereas states that experienced a revolution within the critical 3-year period was 30.1% more likely to form an alliance. This leads me to believe that the effects of revolutions are diluted over time, as the immediate interests that drive leaders to form alliances in the aftermath of the revolutionary event are muted by their growing accustomed to the existing international system. The reputation of a state then also becomes steadier and more predictable in the long-term aftermath of a revolutionary event. This may therefore indicate that revolutionary states, after passing the critical three year mark are more able to develop a reliable reputation that would then provide with a greater variety of potential partners when it comes to alliance formations. In the immediate aftermath, however, revolutionary states, having just experienced a radical leadership change, are interested in forming new alliances, but are unable to engage with states whose reputations are well-established, and therefore are more likely to try their luck with similarly fresh revolutionary regimes. I believe that I will find that alliance onset becomes more probable when both involved states are revolutionary according to the 3-year time frame.

Hypothesis: Alliance onset occurs at a higher rate when both involved states are revolutionary according to the 3-year time frame established in Colgan and Lucas, 2016.

I conduct this analysis by using a generalized linear model to examine the effect that two revolutionary states have on alliance onset rate. I utilize the Alliance Treaty Obligations and Provisions database (ATOP; Leeds, Ritter, Mitchell, and Long 2002), which contains dyadic entries and records military alliances between 1816 and 2010, with variables such as allaince history, alliance reputation (examined by Crescenzi et al.), and polity score differences. I will explain in further detail below the variables that are of interest for my analysis.

Download my RStudio code and dataset here. To read the full paper, get in touch with me!


Democratic v. Revolutionary Alliance Onset Rates

Interestingly, the alliance onset rate between two revolutionary states is also higher than that of two democratic states, which indicates that although Crescenzi et al. found that regime type does not have any significant effect on the rate of alliances between states (see Crescenzi et al., 2012:268 and model_4 above), similarity between regimes does seem to be significant when it comes to joint democracies or joint revolutionary states. The results are shown below, with the alliance onset rate for the dataset as a whole included for reference.


©2020 by Juliann Li. Proudly created with