Does climate change cause social mobilization? Exploring the linkages between protests events and climate change 1990-2020

When and why do accelerating processes of environmental change lead to social mobilization in the form of mass protests? The effects of environmental problems such as climate change, biodiversity loss and resource depletion are significantly impacting the daily lives of people and communities around the world. As a result, environmental problems are being increasingly contested and environmental mass protests (demonstrations, marches, public manifestations) are thus a key expression of how environmental problems are creating contestation and mobilization in local communities, as well as a premonition of how accelerating environmental degradation will affect the stability of communities in the future. The aim of this paper is to investigate if, how, and under which circumstances global environmental change causes environmental protests events. We investigate the effect of two types of environmental change –droughts and heat waves—that might have direct affect the livelihood of communities and individuals and thus are more likely to lead to social mobilization. To our knowledge, this is the first study to systematically assess the link between environmental degradation and social mobilization. We use a new global dataset of approximately 17 million protest events for the period 1990-2022. Our main source data is Dow Jones/Factiva, which has the most comprehensive global full-text newspaper archive available with more than 40 thousand multi-language news sources from around the globe. We minimize potential Western bias in news reporting by including many different newswire sources from beyond the Anglophone sphere (currently Spanish, French, German, Portuguese, and Chinese). We apply machine-learning techniques to filter relevant environmental protest events from other protest events. Building on Alsaedi et al. (2017), we apply supervised machine learning algorithms to detect larger scale events and unsupervised approaches to cluster, disambiguate, and detect smaller events. Once we have identified environmental protest events, we utilize Stanford Core NLP tools to extract the fine-grained event information. We utilize linear-chain conditional random fields (CRF)-based NER (Finkel et al. 2005) for entity extraction. The end result is a dataset with georeferenced and time-stamped (daily) protest events where it will be possible to distinguish protesters (NGOs, local resource users, citizens, etc.), protest target (national/local government, companies, the public), protest issue (e.g. water usage, land rights/development, climate change, infrastructural development, mining, biodiversity protection/habitat loss, pollution, etc.). To facilitate integration control variables from with other georeferenced datasets, the protest data is imputed into the PRIO Grid data format which contains grids of 50x50 km2 at the equator (Tollefsen et al. 2012). In a final step we combine our dataset of protests with two high-resolution environmental change data sets: the “CPC Global Daily Temperature” and the “CPC Global Unified Gauge-Based Analysis of Daily Precipitation” provided by NOAA/OAR/ESRL PSL for data on temperature and precipitation anomalies, and the Global Biodiversity Information Facility (GBIF) for deforestation data. With the help of spatial regression models (Gleditch and Ward 2008), this data will allow us to investigate, with a high degree of temporal and geographic resolution, how changes in natural systems cause social unrest and mobilization of dissent.