An empirical stochastic model of Argentina’s Impossible Game (1955–1966)

Abstract

Argentine politics from 1955 to 1966 was characterized by the conflict between the Peronists and the anti-Peronists. While each camp could veto the other’s project, neither could advance their own agenda. In his canonical interpretation, O’Donnell (Modernization and Bureaucratic-Authoritarianism. Berkeley, CA: Institute of International Studies, University of California, 1973) concluded that party democracy during this era was tantamount to an ‘impossible game.’ While we recognize the significance of O’Donnell’s analysis, we believe that it presents a number of problems. To address its main shortcomings we consider a spatial model that emphasizes the importance of voters’ judgments about the characteristics of party leaders. We recover the positions of Argentine parties using a mixed logit stochastic model and an original dataset of recorded votes in the Argentine Chamber of Deputies during this era. Our results suggest that the electoral logic forced the Peronist party to adopt relatively radical positions away from the center in order to maximize its support. In turn, non-Peronist parties had little incentive to seek the support of moderate, and thus ‘unrepresented,’ Peronist voters by locating themselves at the electoral mean. In particular, valence differences associated with Peronism prevented larger parties from converging toward the center. We thus conjecture that the rules of the impossible game were a constraint imposed by the populace on Argentine political elites rather than a choice made by the latter behind the people’s back.