Many important provisions of US securities law—most notably, crucial elements of the Sarbanes-Oxley (SOX) legislation enacted in 2002—apply only to firms that have a public float of at least $75 million. Public float (i.e., the market value of shares held by non-insiders) is not comprehensively reported in standard databases, so I “scrape” public float data from firms’ 10-K filings for an extensive sample of reporting entities over fiscal years 1993–2015. I use a bunching approach that compares the number of observations immediately below the $75 million threshold to a smooth counterfactual density. Prior to SOX (i.e., over 1993–2002), there is no detectable bunching. Following SOX (i.e., over 2003–15), there is statistically significant evidence of bunching. However, the magnitude of bunching is relatively modest. Moreover, bunching is concentrated in the early post-SOX years (2003–09) and is virtually absent in later years (2010–15). The magnitude of bunching is not a sufficient statistic for the compliance costs of securities regulation because the costs of managing public float are unobservable. Nonetheless, the results of the bunching analysis cast some doubt on widespread claims that the regulatory burdens of these securities law provisions are large.

中文翻译：

---

使用聚类方法估计公司对证券监管的反应

美国证券法的许多重要条款——最著名的是 2002 年颁布的萨班斯-奥克斯利法案 (SOX) 立法的关键要素——仅适用于公众持股量至少为 7500 万美元的公司。公众持股量（即非内部人持有的股票的市场价值）在标准数据库中没有全面报告，因此我从公司 10-K 文件中“抓取”了 1993 财年报告实体的大量样本的公众持股量数据–2015。我使用了一种聚类方法，将低于 7500 万美元阈值的观测值数量与平滑的反事实密度进行比较。在 SOX 之前（即 1993-2002 年期间），没有可检测到的聚束。在 SOX 之后（即 2003-15 年期间），有统计上显着的聚束证据。然而，聚束的幅度相对适中。而且，聚束集中在 SOX 之后的早期（2003-09 年），而在后来的几年（2010-15 年）几乎不存在。由于管理公众持股量的成本是不可观察的，因此对于证券监管的合规成本来说，集群的规模不足以统计。尽管如此，聚类分析的结果对普遍认为这些证券法规定的监管负担很大的说法产生了一些怀疑。