Using Natural Language Processing to Measure COVID-19-Induced Economic Policy Uncertainty for Canada and the US

We develop an economic policy uncertainty (EPU) index for Canada and the US using natural language processing (NLP) methods. Our EPU-NLP index is based on an application of several algorithms, including a rapid automatic keyword extraction algorithm (RAKE), a combination of the RoBERTa and the Sentence-BERT algorithms, a PyLucene search engine, and the GrapeNLP local grammar engine. For comparison purposes, we also develop an index based on a strictly Boolean index. We find that the EPU-NLP index captures COVID-19 related uncertainty better than the Boolean index. Using a structural VAR approach, we found that an economic policy uncertainty shock with EPU-NLP results in larger declines in real GDP, employment, industrial production and the TSX index than with EPU-Boolean for Canada. Similar results were also found for the US: an EPU-NLP shock led to larger declines in industrial production, employment, real personal consumption expenditure, and S&P500 than EPU-Boolean. The SVAR model showed an abrupt contraction in economic variables both for Canada and the US in line with the COVID-19 impact. Moreover, an uncertainty shock (with the EPU-NLP) caused a much larger contraction in economic variables for the period including the COVID-19 pandemic, than for the period before COVID-19.