Local and National Economic Policy Uncertainty in US States
Download Data
We are pleased to host three new EPU indices developed by Scott R. Baker, Steven J. Davis, and Jeffrey Levy. These indices measure Economic Policy Uncertainty within US states.
To construct our state level measures of EPU, we select sets of daily and weekly newspapers for Washington DC and every state in the United States, excluding national papers that are published in a given state such as the NY Times or Wall Street Journal. All newspaper archives are accessed through the Access World News Newsbank service.
In all, we cover around 3,500 papers, with the median state being represented by 49 newspapers in our sample period (maximum of 279 papers from Illinois and minimum of 4 papers in Washington DC). These papers range from highly local to wide-ranging state flagship newspapers. The median reported circulation of a newspaper in our sample is 9,000 and the mean is 23,000. Our panel of papers is unbalanced, though the majority of papers extend coverage to the present day.
Term Sets and EPU Index Construction
We construct EPU indexes by recording the fraction of articles that contain terms from term sets regarding the economy, uncertainty, and policy. For each state, we then construct three separate EPU indexes distinguished only by variations in their policy term sets. That is, for inclusion in each index, an article must contain the word `economic' or `economy' as well as `uncertain', `uncertainties', or `uncertainty'.
The first state level index, EPUNational, aims to measure the level of uncertainty within a state that stems from specifically national policy-related sources. It includes terms related to national elections, elected officials, federal agencies, departments, and regulators.
A second index, EPUState, looks to measure the level of uncertainty within a state that comes from state and local policy issues. Mirroring the terms from EPUNational, this set of terms encompasses many facets of their legislative, executive, and regulatory institutions. Each state-specific policy term set includes terms that describe the names of their executive positions and legislative bodies at both state and local levels as well as terms that note policy initiatives put to a direct vote by citizens. We also include the names of the state bodies that deal with regulations spanning the environment, labor and unemployment, gambling, transportation, banking, energy and utilities, and other financial services. Consequently, this set of terms is unique to each state, since the names and titles of officials and regulators and departments vary across states. A full list of state specific terms can be found HERE
Finally, a third index, EPUComposite, is composed of articles that contain terms related to the economy and uncertainty and also a term from a composite set of terms that contains state-specific policy terms as well as the set of national policy terms.
To construct each index, we search newspaper archives at a state-month level (across all newspapers in the state) for articles containing a term from each of the three sets of keywords (economic terms, uncertainty terms, and policy terms). We conduct separate searches for each of the three policy term sets. We then calculate the fraction of articles meeting these requirements relative to the total number of newspaper articles in the same state-month.
Finally, we normalize each index by the average state level ratio of the first term set, EPUNational, prior to 2018. This normalization is done to preserve information and so that we can meaningfully compare levels of policy uncertainty across our index variants within the same state. In this manner, we can examine how policy uncertainty that stems from state and local events varies relative to uncertainty coming from national drivers.