Ireland Monthly Indexes (Rice)

Download Data                                        Download Annotated Index

We are pleased to host two indexes of economic policy uncertainty for Ireland developed by Jonathan Rice.

Rice constructs two indices of Economic Policy Uncertainty for Ireland, as detailed in his paper on "Economic Policy Uncertainty in Small Open Economies: A Case Study of Ireland," which follows the index construction methods in "Measuring Economic Policy Uncertainty" by Scott R. Baker, Nicholas Bloom and Steven J. Davis.

His headline EPU index for Ireland begins in January 1982 and his second EPU index begins in January 1997 and aims to isolate the domestic aspect of economic policy uncertainty in Ireland (purged of foreign policy uncertainty).

The headline EPU series compiles newspaper articles from three leading Irish newspapers - the Irish Times, Irish Independent and Irish Examiner. The Irish Independent is Ireland's most widely circulated newspaper, shortly followed by the Irish Times. In line with Baker, Bloom and Davis (2016), Rice conducts a keyword search within the electronic archives of each newspaper for articles containing one or more of "uncertainty" or "uncertain", one or more of "economic" or "economy", and finally, one or more of a range of Irish policy-related terms as detailed in the paper. For the headline series, the resulting monthly count is scaled by the total number of same-month articles published in the three papers, and the data are normalised to an average value of 100.

The final compilation of newspaper articles contained in the series was manually audited to ensure accuracy and test for errors, as detailed in the paper. As a further check, Rice tested the Baker, Bloom and Davis (2016) method by constructing a series containing the keywords "volatility" or "volatile" and "Irish stock exchange" or "ISEQ" and comparing it to actual data on Irish equity price volatility.

For the Irish-domestic EPU series, the headline Irish EPU series was purged of foreign uncertainty shocks. To do this, a global EPU series was constructed by extracting principal components from a range of other countries' EPU series (excluding Ireland). Next, Rice regressed the headline Irish EPU series on these principal components and their lags, recovering the residuals. This residual series may be interpreted as movements in the Irish EPU series that are not predicted by current and lagged values of global EPU, thereby isolating an Irish-domestic EPU series. This series is available from 1997 and normalised to a mean of 100.

Although Rice's index construction methods are similar to those of Ryan Zalla (2016), the time-series correlation between the Rice and Zalla EPU indexes for Ireland is relatively low at 0.6. In the paper Rice contrasts the two series, suggesting reasons for the differences, which are also included below:

• Corpus of newspaper archives: While the approach in Rice's paper is based on three major Irish newspapers, Zalla (2017) is based on one - The Irish Times. A larger number of newspapers included is likely to boost the robustness of the outcome due to the reduced potential for small sample bias. This is particularly true during periods of low uncertainty, where the count of relevant papers is observed to be low.
• Keywords used to identify the newspapers: Zalla (2017) contains fewer Irish policy specific keywords in its newspaper-based search (7). Rice includes 32 keywords, as summarised in the paper. The inclusion of more keywords improves the likelihood that a relevant article will be flagged in the keyword search, reducing the potential for small sample bias for months with low uncertainty.
• Coding algorithm: The algorithm used for the Boolean search varies between the two papers. In Rice's paper he attempts to reconstruct the approach used by Zalla. Using Rice's code with identical keywords to those used by Zalla, and using the Irish Times archives only, the resulting series has a correlation coefficient of 0.75 with the series developed by Zalla, and appears to be less volatile.