Insight from Michael Simpson
Regardless of your political standpoint, most people can agree that Donald Trump’s use of Twitter has been unconventional. The President’s personal account has become a public gateway into his personal and professional thoughts. All this news from Twitter is quickly digested and reflected in the financial markets, but how heavily does the stock market weigh information from the social media site? Has this new tactic of communication impacted uncertainty in the economy? I set out to answer these questions in my senior honors thesis, and hope to shed some light on the results of my analysis here.
At first glance, much of the information Trump shares on Twitter seems to have little to no economic value. To account for this, I began my research by reading all posts on Twitter from the official @POTUS and @RealDonaldTrump accounts from November 2016 to November 2017. After reviewing these, I classified the posts as either personal (neutral), increasing uncertainty (negative), or decreasing uncertainty (positive) based on general tone and expected effect. Below are a few examples:
Increasing Uncertainty - “The era of strategic patience with the North Korea regime has failed. That patience is over. We are working closely…”
Decreasing Uncertainty - “Starting to develop a much better relationship with Pakistan and its leaders. I want to thank them for their cooperation on many fronts.”
Personal - “Wishing everyone a safe and Happy Halloween! #Halloween2017”
I then assigned posts that were expected to increase uncertainty a value of -1 and decreased uncertainty a value of 1. Using these values, I ran univariate regressions with the S&P 500 and the VIX, commonly referred to as the “fear index” that measures uncertainty in the economy. My regression was straightforward; I hoped to determine whether or not the 1 or -1 sentiment values assigned to each post correlated—and predicted—changes in the indices. Here is what I found:
For both indices, it is evident that markets take notice and trade accordingly for the first 1-2 minutes after Trump posts to Twitter. The p-value indicates that results are statistically significant for durations of both 1 and 2 minutes. However, once 5 minutes has passed since the post, it appears that the financial markets have already processed and completely priced in the new information.
The data seems to show that traders are using this information, but is it reflected in the index prices? Below are the returns for these indices immediately following when Trump has posted to Twitter:
*Note that an increase in the VXX (VIX futures contract) represents an increase in uncertainty
As shown above, uncertainty tends to increase after negative posts and to decrease after positive posts. However, the regression indicates a strong relationship only in the 1 and 2 minute time frames. The SPX (an ETF for the S&P 500) tends to decrease following negative posts and to increase following positive posts. The opposite is true for VXX, since an increase in VXX denotes an increase in uncertainty (unlike in the S&P 500, where increases indicate greater certainty and confidence).
Another key indicator for investor action following news events is the change in trading volume. To measure the impact that Trump’s posts to Twitter might have from this perspective, I gathered all the trading volume for the same two indices and looked for changes in volume following posts to Twitter. Here’s what I found:
These are perhaps the most clear-cut results from the entire study. In almost every timeframe, volume increased dramatically after negative posts (a sign of uncertainty) and stayed fairly constant after positive posts (a sign of investor “calmness”).
The overall economic impact of the President’s unconventional Twitter use is still up for debate, but one things is clear: investors are watching the @POTUS and @RealDonaldTrump accounts very closely.
Michael Simpson graduated from UNH in 2018 with a major in economics and a minor in mathematics. He currently works as an Operations Analyst at Susquehanna International Group.