Contagious Protests

Introduction

History provides examples of media incentivizing street protests, which spread across borders. The popular discontent that ended communism in Eastern Europe began in Poland in the early 1980s, and swept through Hungary, East Germany, Bulgaria, Czechoslovakia, and Romania. Radio Free Europe played a critical role as a vehicle for the spread of the information about protests and catalyzed its spillovers across borders (Puddington 2000), as did a network of radio stations of the Catholic Church (Stehle 1982). Fast forward 30 years to the wave of street protests that sparked the Arab Spring in 2011 following the death of a Tunisian street vendor who set himself ablaze—after having been harassed and humiliated by municipal officials. These protests established social media as the new conduit for the spread of protests. In 2019, a second wave of protests that started in Sudan and Algeria spread to other Arab countries, including the Arab Republic of Egypt, Lebanon and Iraq, eventually resulting in a global contagion of protests spanning from Chile to the Russian Federation to Hong Kong SAR, China. Just as the global pandemic put a damper on street protests, the death of George Floyd in US police custody powered a global protest movement against police brutality and racial injustice. Again, social media was the main channel of protest engagement.

In this paper we systematically explore spillovers of spontaneous street protests across countries, with social media acting as the catalyst. We use data on nonviolent and unorganized demonstrations for 200 countries for the period 2000 to 2020. Using an autoregressive spatial model, we find strong evidence for contagious protests with a catalyzing role of social media. Social media penetration in the source and destination of protests leads to protest spillovers between countries. There is evidence for parallel learning between streets of nations alongside the previously documented learning between governments.

Our research is related to three strands of previous work. First, our paper is related to the emerging literature on social media and protests. Battaglini (2017) argues that protests are a way to signal private information to policy makers, in which social media is a good medium for policy makers to aggregate information. A field experiment in Hong Kong SAR shows that one’s turnout decision depends on others’ turnout decisions (Cantoni et al, 2019). Broadband internet is shown to be positively associated with local online grassroots protest movements in Italy (Campante et al, 2018). The diffusion of VK, Russia’s dominant online social network, increased protest turnout in Russia (Enikolopov et al, forthcoming). Where internet coverage is low, such as in Africa, mobile phones’ diffusion increased protest turnout (Manacorda and Tesei, 2020).

Our paper is also related to the literature on learning and networks. Buera et al. (2011) study how countries learn from the successes of neighbors’ past economic policies. Chen and Suen (2016) theorize that a successful revolution could dramatically lead to revisions in beliefs of both protesters and governments and hence may lead to a series of revolutions in other countries. König et al. (2017) examine how networks (of military alliances and enmities) affect the intensity of the conflict in the Democratic Republic of Congo. Harari and Ferrara (2018) examine how local conflicts in one gridded cell in Africa spill over to neighboring cells.

A strand of the literature studies spillovers from protests. Several papers have documented that protests in Tunisia and Egypt have inspired protesters elsewhere (Bamert et al, 2015; Hale, 2013; Lynch, 2013; Saideman, 2012). Similar spillovers were observed from other revolutionary periods, such as the Eastern European democratization in the 1990s and the revolutionary wave of 1848 in Europe (Hale, 2013; Weyland, 2010, 2012).

The remainder of the paper is organized as follows. Section II presents the data and the empirical strategy. Section III shows the results. Section IV presents robustness checks. Section V concludes.

Download the full version: