Inexorable Force or Dying Wave? Forecasting Regime Change until 2050

• In Event: Poster Session: Democracy & Autocracy

Abstract

The last few years have seen a decreased of democratization and increased autocratization among the regimes in the world. However, is there enough evidence to suggest that the current wave of autocratization has spelled the end of the third wave of democratization? Or, are these democratic and autocratic 'waves' simply the results of a highly variable, but stationary and consistent, data generating process? This paper uses a dynamic monte carlo simulation approach to assess whether or not the current trends of regime democratization and autocratization and democratic backsliding are consistent with such a stationary data generating process, and if so how we can expect the the state of the regimes in the world to develop in the next thirty years. The use of a monte carlo simulation approach is especially useful when aiming to assess the impact of 'waves', as the observed waves could have grown stronger or subsided faster depending on the number of countries swept along in each wave, and single changes that did not turn into 'waves' could in certain cases have done so. Using a monte carlo simulation allows for a credible evaluation of these counterfactual dynamics which allows the paper to provide unique insights into the current trends of democratization and autocratization. To conduct this study, a two-level model is first fitted on data for all countries from 1946 to 2018. The first level of the model estimates the risk of regime change events based on regime characteristics as well as social, economic, and poltical data from the Varieties of Demcoracy project (VDEM), while the second level models different types of regime change using the same covariates. Using the two estimated models, the study then runs 10,000 dynamic monte carlo simulations from 1946 to 2050, using the data from VDEM in the period 1946 to 2018 and forecasted data from the Shared Socioeconomic Pathways (SSP) project from 2019 to 2050. In each of these simulations, the observed regime changes and regime characteristics are dynamically updated for each year by making random draws of regime change events and types of regime changes based on the predicted probabilities from the two-level model fitted on the entire data. This dynamic approach allows individual countries in each simulation to experience or avoid regime changes, which may influence their likelihood of them experiencing further regime changes or causing neighboring countries to also experience or avoid regime changes, thereby mimicking the behavior of 'waves'.

Author

• David Randahl, Department of Peace and Conflict Research, Uppsala University