Fedesarrollo Indexes of Economic Policy Uncertainty for Colombia

Download Data                Annotated Fedesarrollo Colombia EPU Index

We are pleased to host Economic Policy Uncertainty indexes for Colombia (IPEC) developed by Fedesarrollo following the methodology in "Measuring Economic Policy Uncertainty" by Scott R. Baker, Nick Bloom and Steven J. Davis. The Fedesarrollo team uses articles from El Tiempo, the only Colombian newspaper with a digital archive covering the full period from 1990 onwards.

Index construction

Fedesarrollo identifies relevant articles using three sets of keywords:

• Economy: Economy, Economic
• Uncertainty: Uncertainty, Uncertain
• Policy: (a) Fiscal policy: Government, Fiscal policy, Budget, Fiscal deficit, Public debt, Tax, Tax authorities, Ministry of Finance, Public spending. (b) Monetary policy: Monetary policy, Central Bank of Colombia, Issuer. (c) Trade policy: Tariff/Tariffs, Trade policy.

An article is included in the IPEC if it contains at least one term from each of the three categories: Economy, Uncertainty, and Policy. Selected articles are then expressed as a proportion of the total number of articles published in the same newspaper and month, ensuring that the index captures relative changes in coverage.

Standardization and normalization

The raw dataset is standardized by dividing the monthly frequencies by the standard deviation of the series for the period January 2000 to December 2019, thus ensuring consistency over time while excluding disruptions from the COVID-19 pandemic. The series is further normalized to have a mean of 100 during the same period.

Sectoral classification

Fedesarrollo enhances the IPEC by categorizing selected news articles based on their primary economic sector. This classification provides deeper insights into how uncertainty impacts different areas of the economy. The process involves:

• Automated classification: Articles are first categorized using a Support Vector Classification (SVC) model trained on a dataset of pre-classified news articles. This machine learning approach identifies patterns in the text and assigns each article to an economic sector
• Manual validation: To ensure accuracy, Fedesarrollo experts review and validate the automated classifications, correcting potential misclassifications due to language ambiguities.
• Single sector assignment: Each article is linked to a single dominant sector to preserve clarity and facilitate sector-specific analysis. This avoids complications from overlapping classifications.

For additional information, see "Methodology of the Economic Policy Uncertainty in Colombia (IPEC)".