Winners Never Quit, Quitters Never Grow : Using Text Mining to Measure Policy Volatility and its Link with Long-Term Growth in Latin America

Calvo-Gonzalez, Oscar ; Eizmendi, Axel ; Reyes, German

World Bank, Washington, DC

关键词: POLICY STABILITY;    ECONOMIC GROWTH;    VOLATILITY;    TEXT MINING;    PRESIDENTIAL SPEECHES;   

DOI  :  10.1596/1813-9450-8310
RP-ID  :  WPS8310

学科分类：社会科学、人文和艺术（综合）

来源: World Bank Open Knowledge Repository

【 摘 要 】

Although there is wide recognition ofthe negative consequences of policy volatility forcountries' long-term economic growth, there is limitedempirical work on this subject. One of the reasons is thedifficulty of measuring policy volatility over long periodsof time, especially in developing countries. This papercontributes to this literature by constructing a proxy forpolicy volatility that exploits the information content ofthe priorities conveyed in presidential speeches. The studycreates a policy volatility measure using a Latent DirichletAllocation algorithm on a novel data set of 953 presidentialspeeches in 10 Latin American countries and Spain. The papershows that the proxy for policy volatility is negativelycorrelated with long-term growth over 1940-2010. The resultsare robust to a large set of changes in the construction ofthe proxy for policy volatility.