Monday, January 11, 2021
Anthony M. Diercks (Board of Governors, Federal Reserve), Daniel Soques (University of North Carolina-Wilmington) & William Waller (Tulane), The 283 Days of Stock Returns after the 2016 Election:
Conventional wisdom suggests that the promise of tax legislation played an important and positive role in the 25% increase in the stock market that began on November 9, 2016 and continued through December 22, 2017 (the day TCJA was signed into law). Our comprehensive and exhaustive forensic analysis confirms its positive effect. With that said, we find that its net impact is relatively modest.
To come to this conclusion, we first construct a novel daily human-based attribution by carefully reading the news on each of the 283 days. This exercise shows the 52 days in which tax-related news was important make up less than 1% of the total observed return. We attribute large gains to tax-related news immediately after the election as well as the build-up to passage in late 2017. However, key events in the summer of 2017 decreased the prospects for tax legislation, which wiped out most of the gains that we attributed to tax policy over the full sample. This "up, down, and up again" narrative is corroborated across a wide-range of alternative approaches, including (1) a machine-driven textual analysis based on over 1,500 possible specifications, (2) a novel probability measure tied to the passage of tax legislation constructed from prediction markets, (3) the relative performance of high tax firms compared to low tax firms, (4) a daily attribution based on firm-level regressions, and (5) several macroeconomic financial indicators.
The relatively modest estimated effects are consistent with the market potentially being more driven by strong global growth, changes in other policies, a weaker dollar (which coincided with a reduction in the likelihood of passage of tax legislation), and numerous below-expectations inflation prints (keeping monetary policy at bay) that fortuitously occurred over this time period.