While I argued a few days ago that war was good in Malthusian economies, there is little doubt that war is bad in modern economies, except possibly for winners. Thus on average wars generate welfare losses. Then why do we still have wars? One key aspect here is that wars could be instigated by leaders who disrespect the welfare of their underlings. The logic conclusion of this argument is that democracies should be much less at war as, following Kant, democracies need to be accountable.
Paola Conconi, Nicolas Sahuguet and Maurizio Zanardi test this hypothesis by exploiting the fact that in some democracies presidents face term limits and are thus not accountable in the last term. The idea is that there is a repeated prisoner's dilemma game, where the horizon of the policy maker does not necessarily coincide with the country. Policy makers are semi-benevolent, care about rents from reelection and election success is endogenous. There is a short-term electoral boost of war, following Gregory Hess and Athanasios Orphanides, but there is more importantly a long-term cost to the country.
Conconi, Sahuguet and Zanardi construct a dataset of term limits from 1816-2001 for many countries and find that presidents in a legally binding last term behave as aggressively as autocrats. Thus, democracy is not sufficient to bring peace, you need to avoid term limits as well. However, what to think of the electorate, as voting for a lame duck seems to have negative consequences? Are people stupid, or are they manipulated? Also, the paper does not explore the implications a lame duck to continue presiding through his party (exhibit A: Vladimir Putin). But even without these consideration, the empirical evidence should us to explore ways to make war more difficult for lame ducks, as they have war mongering inclinations.