When Are Democracies at Risk of Backsliding?

Abstract

In this thesis I examine the extent to which democracies are vulnerable to experiencing democratic backsliding through the emergence and survival of illiberal and undemocratic executives, who interfere with accountability institutions like the judiciary and media. Specifically, I evaluate the reasons executives interfere and the capacity of voters to punish undemocratic executives to show that many democracies are likely not at risk of sustained interference.

In long-standing democracies, where severe assaults on accountability institutions have not occurred, I find that voters dislike and heavily penalise illiberal and undemocratic policies. In reanalysing a large set of published experimental and survey data, primarily from the United States, I argue that negative attitudes towards undemocratic policies are strong enough to preclude politicians holding such polices from both being elected and then staying in office when voters learn about them.

Where interference frequently occurs, in new democracies, I argue that executives do so more for strategic reasons - to overcome procedural barriers to realising policies -- than any a priori hostility towards democracy or love for authoritarianism. I employ a severe case of country backsliding, the Philippines under Rodrigo Duterte, to show that Duterte's motivations to interfere, as well as those of his predecessors, were to realise policy priorities that they perceived were threatened by the courts and media. Data from a conjoint experiment I deployed before the 2022 presidential election demonstrates that Duterte's immense popularity alongside high copartisan bias -- the extent to which partisan voters are unwilling to defect to the pro-democratic opposition when exposed to information about copartisans' undemocratic positions -- explains the absence of electoral penalties despite his supporters preferring their own side to be pro-democratic. Such high levels of copartisan bias in so many voters are not exhibited in similarly designed conjoint experiments from the United States and other long-standing democracies.

I conclude that the risk of backsliding emerging and then persisting in many democracies is low -- especially in long-standing democracies where enough voters treated with undemocratic information are able to reliably punish the kinds of institutional violations found in severe backsliding cases like Ecuador, Hungary, and Turkey. A country is at high risk of persistent backsliding and an authoritarian transition where there is strong copartisan bias among voters, highly popular parties, and executives with strategic motivations to interfere with institutions. Show more