Protest and repression in democratic systems: a comparative analysis with a focus on Brazil

This thesis focuses on whether protest affects the levels of repression in electoral democracies and, if so, in what manner. After an overview of the literature, Chapter 2 contains an empirical analysis of the relationship between protest and repression at a global level, using a dataset of 71 democracies over 10 years. The results point to a positive association between protest and repression that is driven primarily by post-1974 democracies. The chapter then develops a theoretical model of the costs and benefits accruing to a democratic leader when deciding whether to repress a protest. The model yields a number of testable hypotheses about which factors will affect the likelihood that repression will be chosen, which are then tested for using cross-national and sub-national data. The impact of constitutional constraints is examined first using the cross-national dataset, which reveals that executives in new democracies centralise power in response to protest. Chapter 4 is a quantitative study of the 27 Brazilian states over a 9-year period using data on the repression of land protesters and political variables. The results indicate that governors with precarious political positions are less likely to promote repressive policing strategies. Chapter 5 uses data drawn from five Brazilian national newspapers to identify whether under-reporting of land protest events might contribute to the level of state repression. Chapter 6 is a qualitative comparison of two states – São Paulo and Pará – and suggests that while tight political control over the police explains repression in the former, the unaccountability of the police and the ideology of the main opposition parties in the state assembly may explain why the latter has a much higher level of repression than would be predicted by political factors alone. Chapter 7 revisits the cross-national dataset of 71 democracies to test whether additional determinants of repression identified in Chapter 6 have an effect at the global level.