Resources for the paper: "The Political Sustainability of Fuel Subsidy Reform: Rent Appropriation versus Public Grievance"

This repository contains the resources in our paper "The Political Sustainability of Fuel Subsidy Reform: Rent Appropriation versus Public Grievance" Ramon Villa-Cox, Luis A. Guamán L., Michelle Tello and Juan F. Rumbea P. This paper examines how different types of fuel price reforms affect protest dynamics in Ecuador, a country with varied experiences in subsidy adjustments. Leveraging a novel methodological approach that combines administrative police data and protest indicators from ACLED and GDELT via neural networks and ARIMAX models, we construct a comprehensive daily protest-intensity series from 2018 to 2024. Using a Vector Autoregressive (VAR) model, we test whether changes in diesel, regular gasoline, and premium gasoline prices Granger-cause shifts in protest activity. Our findings demonstrate that diesel price increases trigger substantial and persistent protests, whereas regular gasoline reforms generate moderate, short-lived responses, and premium gasoline reforms have no significant impact. This repository is comprised of the datasets necessary for the replication of the results of the study. Input Datasets: Contain raw datasets form third party sources Protest\protestas_original.csv: Original protest series with missing values. This dataset contains the original daily protest records with missing values. The data is sourced from the National Police of Ecuador, documenting instances of people mobilizing in public spaces. It reflects direct administrative records of collective actions, providing a ground-truth benchmark for validation and comparison with alternative protest measures derived from media and event-based sources. GDELT\RSU_index.csv: GDELT-derived indices. This dataset contains indices derived from the Global Database of Events, Language, and Tone (GDELT), constructed using the methodological framework proposed by Barrett et al. (2022). The RSU index aggregates signals of unrest, stability, and urgency as reported in global media coverage, offering a media-based proxy to detect periods of social tension and protest activity. ACLED\acled_final_new.csv: Indicators built from ACLED protest/violence events. Gasoline_Pump_Prices\Precios_Final.csv: Ecuador's daily fuel prices (Diesel, Regular, Premium). Gasoline_Pump_Prices\Precios_USA_diario_act.csv: USA's average daily fuel prices (Diesel, Mid-grade, Premium). Gasoline_Pump_Prices\WTI_final_act: International oil prices (WTI) 2. Intermediate Datasets: Datasets that contain intermediate results Protest\Protesters Imputation - Tranche 2\protestas_imputadas_personas.csv: Protest series constructed by averaging ARIMAX and LSTM imputed values for the second segment and combining them with the original protest series. Protest\personas_imputada_final.csv: Final version of the imputed protest series, averaged from ARIMAX and LSTM results. VAR_data\df_var: The time series dataframe constructed in Var_Vecm.R, containing: o Diesel_Gasolinera o Eco_Extra_Gasolinera (Regular) o Super_Gasolinera (Premium) o final_imputation (Protests)