Winners never quit, quitters never grow : using text mining to measure policy volatility and its link with long-term growth in Latin America

Although there is wide recognition of the negative consequences of policy volatility for countries' long-term economic growth, there is limited empirical work on this subject. One of the reasons is the difficulty of measuring policy volatility over long periods of time, especially in developing countries. This paper contributes to this literature by constructing a proxy for policy volatility that exploits the information content of the priorities conveyed in presidential speeches. The study creates a policy volatility measure using a Latent Dirichlet Allocation algorithm on a novel data set of 953 presidential speeches in 10 Latin American countries and Spain. The paper shows that the proxy for policy volatility is negatively correlated with long-term growth over 1940-2010. The results are robust to a large set of changes in the construction of the proxy for policy volatility