Overview
We have developed 18 economic policy uncertainty (EPU) indices for the Latin American region: Argentina, Brazil, Chile, Colombia, Mexico, Peru, Bolivia, Ecuador, Paraguay, Uruguay, Venezuela, Costa Rica, El Salvador, Guatemala, Honduras, Nicaragua, Panama, and the Dominican Republic.
These indices will be regularly updated on policyuncertainty.com or Banco de España's website.
Main references
Using Newspapers for Textual Indicators: Guidance Based on Spanish- and Portuguese-Speaking Countries (E. Andres-Escayola, C. Ghirelli, L. Molina, J.J. Pérez, and E. Vidal). Computational Economics. Volume 62, Issue 2, Aug. 2023.
How Economic Policy Uncertainty Spreads Across Borders: the Case of Latin America (E. Andres-Escayola, L. Molina, J.J. Pérez, and E. Vidal). Banco de España Working Paper. No 2549, Dec. 2025.
Economic Policy Uncertainty in Central America and the Dominican Republic (M. Diakonova, C. Ghirelli, J. Quiñonez). Latin American Journal of Central Banking. pp. 100166, Feb. 2025.
Methodology
The methodology closely follows the procedure described in the paper "Measuring Economic Policy Uncertainty" by Scott R. Baker, Nicholas Bloom, and Steven J. Davis (QJE, 2016).
The indices are constructed by searching digital news archives (via the Dow Jones Factiva database) for articles about each country that include keywords related to "uncertainty", "economy", and "policy". Articles unrelated to policy topics such as sports or entertainment are excluded using Factiva's indexing. Keywords are defined in Spanish, Portuguese, and English to accurately capture the concept of economic policy uncertainty:
Uncertainty: uncertain, uncertainty/uncertainties, unstable, instability/instabilities, risk(s).
Economy: economic(s), economy.
Policy: (keywords related to institutions and policy measures, including each country's central bank name; the seat of government or parliament; terms like government, treasury, budget(s), deficit(s), public spending, debt (public or sovereign), fiscal policy/policies, monetary policy/policies, tax(es), reform(s), law(s), regulation(s), etc.).
To construct the monthly EPU indices, we proceed as follows. First, we compute the share of articles containing keywords related to "uncertainty", "economy", and "policy" in each newspaper and month (for local newspapers, the denominator is the total articles published that month in that newspaper; for foreign newspapers, it is the total number of articles about the target country published by that newspaper in that month). Second, we standardize each newspaper's monthly series to have unit standard deviation. Third, we average the standardized series across newspapers for each country. Finally, we rescale the resulting index to have a mean of 100 over the sample period.
Newspaper sources
The indices are based on the following sets of newspaper sources, by country:
Argentina: Clarín, La Nación, Infobae.
Brazil: O Globo, Folha de S. Paulo, O Estado de S. Paulo, Correio Braziliense, Estado de Minas, Agência Brasil.
Chile: El Mercurio, La Tercera, El Diario Financiero, Pulso.
Colombia: El Espectador, El Nuevo Siglo, Portafolio.
Mexico: Reforma, El Universal, La Jornada, El Financiero, El Economista, Agencia Mexicana de Noticias.
Peru: El Comercio, La República, Gestión, Agencia Peruana de Noticias Andina.
Bolivia: Opinión, La Razón, Los Tiempos.
Ecuador: El Comercio, El Diario, Metro.
Paraguay: ABC Paraguay, Agencia Paraguaya de Noticias, Última Hora.
Uruguay: El País Uruguay, La República.
Venezuela: El Universal, Tal Cual, Últimas Noticias.
Costa Rica: La Nación, El Financiero, El Jornal, Supranational Press (Reuters America Latina, Reuters España, CE Noticias Financieras).
El Salvador: La Prensa Gráfica, Supranational Press (Reuters America Latina, Reuters España, CE Noticias Financieras).
Guatemala: Prensa Libre.com, Supranational Press (Reuters America Latina, Reuters España, CE Noticias Financieras).
Honduras: Criterio, La Tribuna, Supranational Press (Reuters America Latina, Reuters España, CE Noticias Financieras).
Nicaragua: La Prensa, Supranational Press (Reuters America Latina, Reuters España, CE Noticias Financieras).
Panama: ANPanamá, Critica Online, Critica, Decisiones Panama, El Siglo, La Estrella, Panamá América, La Prensa, La Verdad Panamá, Supranational Press (Reuters America Latina, Reuters España, CE Noticias Financieras).
Dominican Republic: Supranational Press (Reuters America Latina, Reuters España, CE Noticias Financieras).
For the following countries (Argentina, Brazil, Chile, Colombia, Mexico, Peru, Bolivia, Ecuador, Paraguay, Uruguay, Venezuela), three versions of the EPU index are available, based on different groups of news sources:
Local Press EPU: Constructed from that country's leading local newspapers (in Spanish, or Portuguese for Brazil).
Foreign Press EPU: Constructed from major foreign newspapers, both Anglophone (U.S. and UK/Canada: Anglo-Saxon Press: Los Angeles Times, The Boston Globe, The Globe and Mail, The New York Times, The Telegraph (UK), The Times (UK), Chicago Tribune, The Guardian (UK), The Wall Street Journal, The Washington Post, The Economist) and Spanish-language (Spain: El País, El Mundo, Expansión, ABC, Cinco Días, El Economista, La Vanguardia), that cover news about the country.
All Press EPU: An index based on all the above sources combined, weighting each source equally regardless of location.
Breadth of press coverage: IRFs of GDP growth to EPU shocks.
Each plot shows the median impulse responses of GDP growth to a one standard deviation shock to the EPU index of Brazil, along with 84% credible bands. In each iteration, we compare the baseline impulse response function (IRF), shown in grey and constructed using all available newspapers, with alternative IRFs shown in green, which are based on different combinations of local and foreign newspaper sources. This comparison allows us to assess whether GDP growth responses converge toward the baseline as additional newspapers are included in the construction of the EPU index. Overall, the results illustrate that using a larger set of newspapers leads to more stable and robust results.