Economic keywords in political communications and financial markets

This paper investigates the interplay between the use of economically significant keywords in political speeches - indicative of their economic content - and observed trends in financial markets. To this end, we implement a statistical/bibliometric study on the speeches stated by the US Presidents and the Standard and Poor's 500 index (S&P 500). The corpus covers 1951-2017, including 376 discourses. The S&P 500 daily returns, prices, and volumes are considered in the same period. Our analysis utilises Kendall's $\tau$ correlation to assess the relationship between S\&P 500 variables and the frequency of economic terms in the speeches. Additionally, four distance measures are applied to evaluate the similarities among these time series. Lastly, we compare Shannon entropies computed for each variable. The values of the index are observed the day before, the same day, and the day after the Presidents conducted their speeches. The joint evaluation of the results shows that the speeches' economic content and the S&P 500 volume have the most meaningful relationship. In particular, the volume shows some changes interrelated to the speeches before and after the days of the talks. The price has similar behaviour but with a lighter magnitude than volume. The daily returns are poorly responsive to the speeches' economic content, but entropy suggests that they present some distributional similarities.