The Determinants of Macroeconomic Volatility: A Bayesian Model Averaging Approach

Abstract

Bayesian model averaging is applied to robustly ascertain the determinants of various output volatility measures, including the downside semi-deviation of growth rates. Financial sophistication variables are found to have qualitatively different effects on volatility. The ratio of government expenditure to GDP exhibited a significant positive relationship with volatility and the trade share of GDP was positively related for a balanced dataset of developed and developing countries between 1960-89, and negatively related for developing countries between 1974-89. Other significant determinants were the black market premium, civil liberties, political rights, rule of law, and ratios of short-term debt and taxation to GDP.

Keywords: Macroeconomic volatility, Growth, Government policy, Bayesian model averaging, Model selection

JEL Classification: C11, C52, E32, E60, F00, O47