Leader's National Identity Conceptions and their Effect on Military Spending: a Data Science Approach to Leadership Typologies

With the growing awareness of the realist paradigm’s limitations to address the ‘real world’ proliferation cases, which often contradict the ‘national interests’ centered forecasts, alternative theoretical frameworks were devised. One of the most ambitious theoretical novelties to address states’ (un)willingness to increase military capabilities – ‘National Identity Conception’ (NIC) by Hymans (2006) suggests the primary driver for such policy trajectories to be the cognition of the leaders. As claimed by the NIC framework, statesmen/women shape the national self-identification, based on their nationalist psychological archetypes, which directly influence their (un)willingness to militarize. In this paper we construct a natural language analysis model, which clusters the leaders based on their UN General Assembly speeches with the intentions to test the NIC’s hypothesis on the direct relationship between the emotional-linguistic manifestations of the leaders’ typologies and military spending. Having accounted for both internal (powers of the leaders in the domestic politics) and external (engagement in high intensity armed conflicts and/or wars) factors we empirically found the following correlations to be consistent for both selected datasets: i) the application of our unsupervised categorization model yields cluster distances, which correlate significantly with the military spending and ii) in both dataset the leaders’ discourse on “cooperation”, “common values” and other ‘liberal’ topics correlates negatively with the military expenditure of their respective states. Thus, these two robust findings indicate observable relationship between the political discourse and military expenditure, opening new possibilities for further research on the nature of securitization.