New Evidence on Employee Noncompete, No Poach, and No Hire Agreements in the Franchise Sector

This repository holds data, replication files, and a machine learning classifier to detect no poach clauses for "New Evidence on Employee Noncompete, No Poach, and No Hire Agreements in the Franchise Sector," forthcoming at Research in Labor Economics. Abstract. This paper presents new evidence on anti-competitive practices in the franchise sector. Drawing from a corpus of Franchise Disclosure Documents (FDDs) filed by 3,716 franchise brands in years 2011- 2023 (partial), I report new information on franchise brands’ use of inter-firm non-solicitation (“no poach”) clauses barring recruitment between firms, no hire clauses barring employment, and franchisor requirements that franchisees use employee non-compete clauses barring workers from joining competitors. Regulatory actions that restricted the enforceability of anti-competitive clauses began to appear in FDDs in 2018. While non-solicitation and no hire clauses have declined in use, the use of non-competes remained stable over time. While prior evidence on anti-competitive practices largely draws from individual complaints, survey data, and limited hand-coded samples, this paper spotlights new methods for finding barriers to worker mobility in large, unstructured text corpora. The process to create the machine learning classifier from unstructured text is described in "Creating Data from Unstructured Text with Context Rule Assisted Machine Learning (CRAML)" with Stephen Meisenbacher. Replication materials are released under a Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International License. CRAML software is at https://github.com/sjmeis/CRAML_Beta/. To acccess the text files used, I request you complete a short form. You may also use this form to notify the author of any issues/ questions. PDFs uploaded to DocumentCloud are available by searching documentcloud.org: +user:peter-norlander-103369 +access:public . I am grateful for support from the Economic Security Project Anti-Monopoly Fund; Loyola Rule of Law Institute; Loyola Quinlan School of Business; and Loyola University Chicago. I thank: Stephen Meisenbacher for development of the CRAML software, document collection, and computational assistance; Patricia Tabarani, Chloe Clark, Kayleigh Currier, Zach Nelson, and Damian Orozco for research assistance; Kate Bahn, Michael Lipsitz, Ioana Marinescu, Eric Posner, Todd Sorensen, Evan Starr, Spencer Weber Waller, David Weil, and two anonymous reviewers and the editors for feedback on earlier versions of the manuscript; participants at the 86th Midwest Economic Association Annual Meeting, the 74th Annual Labor and Employment Relations Association, 42nd Annual Strategic Management Society, the 82nd Academy of Management, and at Michigan State University School of Human Resources and Labor Relations. Errors are mine.本仓库包含用于检测禁止挖人条款（no poach clauses）的机器学习分类器、相关数据及复现文件，对应论文为《特许经营领域员工竞业禁止、禁止挖人与禁止雇佣协议的新证据》，即将发表于《劳动经济学研究（Research in Labor Economics）》。 摘要：本文针对特许经营领域的反竞争行为提供了全新实证证据。本研究基于2011年至2023年（部分年份）3716个特许经营品牌提交的特许经营披露文件（Franchise Disclosure Documents, FDDs）语料库，披露了特许经营品牌使用企业间禁止招揽（即“禁止挖人”）条款（禁止企业间互相招募员工）、禁止雇佣条款（no hire clauses），以及特许人要求受许人使用竞业禁止条款（noncompete clauses）以禁止员工入职竞争对手企业的相关新信息。2018年起，限制反竞争条款可执行性的监管举措开始出现在FDDs中。尽管禁止招揽与禁止雇佣条款的使用率有所下降，但竞业禁止条款的使用比例始终保持稳定。以往关于反竞争行为的实证研究大多基于个人投诉、调查数据以及规模有限的人工编码样本，本文则提出了在大规模非结构化文本语料库中识别员工流动障碍的新方法。 依托上下文规则辅助机器学习（Context Rule Assisted Machine Learning, CRAML）从非结构化文本构建该机器学习分类器的具体流程，详见与Stephen Meisenbacher合著的《Creating Data from Unstructured Text with Context Rule Assisted Machine Learning (CRAML)》一文。 本研究的复现材料采用知识共享署名-非商业性使用-相同方式共享4.0国际许可协议（Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International License）发布。 CRAML软件的开源地址为：https://github.com/sjmeis/CRAML_Beta/。 若需获取本研究使用的文本文件，请填写一份简短问卷。您也可通过该问卷向作者反馈任何问题或咨询相关事宜。 上传至DocumentCloud的PDF文件可通过在documentcloud.org搜索"+user:peter-norlander-103369 +access:public"获取。 本研究感谢经济安全项目反垄断基金（Economic Security Project Anti-Monopoly Fund）、洛约拉法治研究所（Loyola Rule of Law Institute）、洛约拉昆兰商学院（Loyola Quinlan School of Business）以及芝加哥洛约拉大学的资助。感谢Stephen Meisenbacher开发CRAML软件、收集文本语料并提供计算支持；感谢Patricia Tabarani、Chloe Clark、Kayleigh Currier、Zach Nelson与Damian Orozco提供研究协助；感谢Kate Bahn、Michael Lipsitz、Ioana Marinescu、Eric Posner、Todd Sorensen、Evan Starr、Spencer Weber Waller、David Weil以及两位匿名审稿人与编辑对本文初稿提出的修改意见；感谢第86届中西部经济协会年会、第74届劳资关系协会年会、第42届战略管理学会年会、第82届管理学年会以及密歇根州立大学人力资源与劳资关系学院的参会者提供的有益讨论。文责自负。