Democracy and Economic Development: A Time-Varying Parameter Model Approach

• In Event: Poster Session: Democracy & Autocracy

Abstract

Revisiting a controversy on the relationship between economic development and democratization, this paper explores the structural changes in the relationship using a new estimation methodology with time-varying parameters. While some argue that there is no empirical evidence that per capita income causes the creation and the consolidation of democracy, others demonstrate that the effect of income on democratization is conditional on income levels and time periods. However, existing work, which uses time-series cross-sectional data, ignores the time-varying changes in the relationship between variables. A simple econometric model that assumes the effect on a coefficient is constant over time may have a shortcoming as it would have some time variation. A time-dummy coefficient (time fixed-effects) would aim to capture the time variation, but it can produce abrupt fluctuation without any restriction over time.

To address this issue, we develop the Time-Varying-Parameter Probit (TVPP) model that enables us to capture possible changes in the underlying structural relationship between economic development and democracy. This model is incorporated as its time-varying parameter follows a random-walk process with a single variance of errors. A Bayesian estimation procedure is constructed for this purpose. A prior on rather smooth evolution of the time-varying parameters provides realistic smooth time variation in the relationship between variables. This also allows us to implement a counterfactual simulation with the time-varying parameter fixed at a level of any time point.

We show the magnitude of the causal relationship decreases over time, especially after the late 1990s. The coefficient of the income level in the TVPP model decreases roughly by half from 1960 to 2010, where the latest estimate in the 2010s is still statistically significant. The estimation confirms the existing theory that the marginal effects of income growth on democracy in already developed economies decrease, yet per capita income is statistically associated with the creation and the consolidation of democracy. In addition, our model shows the effects of social unrest and trade openness on democracy are not statistically significant over the estimation period. The findings of this study help understand the structural changes that contributed to the waves of democratization.

Authors

• Kenya Amano, University of Washington

• Jouchi Nakajima