Economic inequality continues to thrive in democracies across the world. This is puzzling because theory predicts that the ideals of democracy should induce more equal societies. At the same time, high levels of inequality tend to undermine democratic institutions. It may be that many democracies are trapped in a vicious cycle of persistent economic inequality and democratic backsliding, which may be difficult to alter. The purpose of this dissertation is to examine the general question of why inequality persists in democracies with a focus on the means by which the poor can exert political influence to remedy persistent inequality and why they may fail to do so. The dissertation consists of three independent chapters. Each addresses the links between democracy, redistribution and inequality from different angles. In Chapter 1, I examine the general link between democracy and economic inequality by analysing the distributional dynamics of economic growth in democracies relative to autocracies. The chapter shows that democracies are, on average, no better than autocracies when it comes to inclusive economic growth (growth that is associated with a reduction in economic inequality). However, when countries are classified according to their de facto implementation of democratic institutions, liberal democracies do foster inclusive growth in both relative and absolute terms. Chapter 2 and Chapter 3 focus on the political power of the poor to influence redistributive policies in electoral democracies. Each chapter examines different forms of political action that are available to the poor. Chapter 2 focuses on voting and on the impact of clientelism, which may shape how elections serve to channel demand for programmatic redistribution into the political system. The chapter shows that clientelism adversely affects voter support for political candidates regardless of efforts to provide programmatic redistribution. Chapter 3 focuses on protest as an alternative (or complementary) form of political action to influence redistribution when the electoral process is curtailed. The chapter shows that the poor may hold some de facto political power by virtue of protests. However, the magnitude of that power depends on the choice of protest tactics. The dissertation arrives at these novel insights by employing both observational and experimental quantitative research designs with an emphasis on causal inference. The research questions are analysed both across countries and among individual voters in two unequal electoral democracies; Brazil and South Africa. The dissertation adds to the understanding on the causes of equilibria of persistent inequality, low programmatic redistribution, and democratic backsliding, which are liable to emerge when poor citizens are deprived of political power - at the ballot box or in the streets.