Netherlands Monthly EPU Indexes
We are pleased to host two new indices of economic policy uncertainty for The Netherlands developed by Suzanne Kok, Lars Kroese and Jante Parlevliet.
These researchers construct EPU indices for the Netherlands based on frequency counts of articles in leading Dutch newspapers, following the methods in "Measuring Economic Policy Uncertainty" by Scott Baker, Nick Bloom and Steven J. Davis. The work by Kok, Kroese and Parlevliet is published as Beleidsonzekerheid in Nederland in Economisch Statistische Berichten, 100 (4715), 464-467, August 2015. Their Dutch EPU indices are monthly and run from March 2003 to the present.
As a small open economy, the Netherlands has much exposure to policy developments in the rest of the world, especially other EU countries. This exposure is clearly reflected in Dutch newspaper articles on policy uncertainty, and foreign-sourced policy uncertainty has a large impact on the Dutch economy. Therefore, Kok, Kroese and Parlevliet develop two EPU indexes for the Netherlands: a strict version (EBO-NL) that seeks to measures domestic policy uncertainty only, and a broader version (EBO) that seeks to capture all policy-related economic uncertainty discussed in Dutch newspapers.
To construct their Dutch EPU indices, they obtain frequency counts of articles in the five largest Dutch newspapers (Algemeen Dagblad, NRC Handelsblad, De Telegraaf, Trouw, and De Volkskrant) that contain terms pertaining to uncertainty ('onzeker' or 'onzekerheid') and economics ('economisch' or 'economie') and one or more policy-related terms such as 'beleid' (policy), 'minister', 'begroting' (budget) or 'belasting' (tax). After obtaining these monthly counts by newspaper, they scale by the number of articles published in the first seven days of the month, with an adjustment for public holidays on which Dutch newspapers do not publish. They standardize each paper's scaled frequency count to have a unit time-series standard deviation through December 2014 and then sum across papers by month. As a final step, they normalize the overall series to mean 100 prior to January 2015.
Kok, Kroese and Parlevliet select their policy-related terms based on a small-scale manual audit. They identify domestic sources of policy uncertainty as follows. First, for general policy terms such as 'beleid' (policy), they count an article toward their EBO-NL index only if the article includes a signal term for the Netherlands in close proximity to its mention of 'beleid'. These signal words include the Dutch equivalents of The Netherlands, Dutch, Holland, and The Hague. They do not impose this proximity requirement for policy-related terms that already invoke Dutch policy matters - examples include 'torentje' (office of the Dutch prime minister) and 'binnenhof' (location of the Dutch parliament). Other terms that are mostly used in an international context count towards their EBO index but not their EBO-NL index - examples include 'parlement' (parliament) and 'senaat' (senate). The EBO-NL index is a stricter version of the EBO index. That is, all articles that count toward their EBO-NL index also count towards their EBO index.