The SEC's Short-Sale Experiment: Evidence on Causal Channels and on the Importance of Specification Choice in Randomized and Natural Experiments
78 Pages Posted: 21 Jul 2020 Last revised: 25 Jan 2022
Date Written: October 11, 2020
During 2005-2007, the Securities and Exchange Commission (SEC) conducted a randomized trial in which it removed short-sale restrictions from one-third of the Russell 3000 firms (pilot firms). Early studies found modest market microstructure effects of removing the restrictions but no effect on short interest, pilot firm returns, or price efficiency. More recently, many studies have attributed a wide range of indirect outcomes to this experiment, mostly without assessing the causal channels for those outcomes. We examine the three most often cited causal channels for these indirect effects: short interest, share returns and managerial fear. We find no evidence to support any of these channels. We then reexamine the principal findings in four recent studies using a pre-specified research design (similar across the four reexaminations) and a larger sample that closely matches the actual experiment, and find no support for the reported outcomes in any of these papers. We then switch to best-match specifications that closely match the samples and specifications reported in each paper, and still find only minimal support for the reported results. For two papers, we have the authors’ original data and code; the reported results technically replicate but are highly fragile. Our findings highlight the importance of confirming a causal channel in randomized trials or natural experiments as well as the importance of sample selection and other aspects of specification choice for the statistical significance of reported results.
The Internet Appendix is available at https://ssrn.com/abstract=3657200.
Our pre-specified analysis plan is available at https://ssrn.com/abstract=3415529.
Keywords: natural experiments; causal channels; specification choice; Regulation SHO; SEC experiment
Suggested Citation: Suggested Citation