Read this article
- Access statistics
- 7,181 article downloads
- 1,650 complete issue downloads
- Total: 8,831
The present paper concerns two versions of an article by Andrew Bird and Stephen Karolyi titled “Governance and Taxes: Evidence from Regression Discontinuity.” One version is a manuscript working paper, dated September 2015; the other is the piece as published in The Accounting Review in 2017. Between the two versions, the numbers in all 11 tables stayed exactly the same (to three decimal places), so it is clear that the specifications generating the results were the same. But the description of the specifications changed. The 2015 working paper version says that, for their main results, the authors used one method of ranking firms by market capitalization, while the 2017 published version says that they used another method. Replication and investigation suggests that the working paper version accurately describes the main specification, and the published version misstates it. I conclude by demonstrating that the main result, when accurately described, fails a simple placebo test and is therefore spurious.