方汉明教授谈AER 2006创作心得 Hanming Fang on AER (2006)
本文最初于 2021 年 8 月 10 日 发布于微信公众号 Impactful Research；2026 年 4 月 28 日 同步至本网站。
Originally published on the WeChat official account Impactful Research on 2021-08-10; mirrored to this website on 2026-04-28.
来源：新华网
这个公众号的第二篇文章，我们邀请到宾夕法尼亚大学的讲席教授、上海科技大学创管学院院长方汉明教授分享他和他当时在耶鲁大学的博士生 Shamena Anwar合著的关于警察对机动车搜查中的种族偏见的文章（An Alternative Test of Racial Prejudice in Motor Vehicle Searches: Theory and Evidence ）的创作历程。这篇文章发表在2006年的美国经济评论（American Economic Review），非常巧妙地结合了理论与实证研究，填补了当时文献中的一个重要空白，是公共经济学中研究种族歧视问题非常关键的一篇文献。
方老师从博士阶段就开始关注公共经济学中歧视方面的研究。他的博士论文以一个一般性的视角对为何工资差异可能带来效率改善提供了一个新的解释，随后以Social Culture and Economic Performance为题发表在了美国经济评论（American Economic Review）。在2006年这篇文章之后，方老师与合作者又发表了数篇与种族问题相关的文章。如Separate When Equal? Racial Inequality and Residential Segregation分析了黑人中产阶级的崛起与城市居住隔离；如 Testing for racial Prejudice in the Parole Board Release Process: Theory and Evidence讨论了假释决议中是否存在种族偏见问题；又如Testing for the Role of Prejudice in Emergency Departments Using Bounceback Rates讨论了急诊室医生对不同族裔的病人可能存在的种族偏见问题。
方老师的研究一直强调与经济社会相关的现实意义及可能间接产生的政策影响。他关注美国的平权法案及其对少数族裔所产生的影响。Does affirmative action lead to mismatch? A new test and evidence研究了当校方有私人信息的时候，平权法案所可能带来的错误匹配的问题。而在现实中，方老师和他的合作者们也为亚裔学生团体起诉哈佛大学在录取过程中的不公平问题提供了重要的专家证词。
除了种族歧视，方老师的研究还关注了福利制度改革、公共品提供、拍卖机制、医疗保险市场、人口老龄化、房地产市场等等问题，这些问题都是与现实息息相关的，他也对中国经济的研究颇有影响力。他的研究把严谨的理论与严格的实证相结合，为我们理解现实世界提供了很多新的视角。正如他所说，“经济学是一门社会科学，它的生命力最终还是来自于能够为我们的经济和社会所面临的重大挑战提供洞见”。
以下是方汉明教授分享的关于**An Alternative Test of Racial Prejudice in Motor Vehicle Searches: Theory and Evidence**(AER 2006) 这篇文章的创作历程。
Q: 您是怎样发现这一研究问题的？
Q: How did you identify this research question?
A: 正如今天一样，种族归纳 (Racial profiling) ，或译为种族定性，在上世纪九十年代末是一个很受关注的公共议题。地方和全国性的媒体经常会报道那些关于警察更频繁地搜查少数族裔司机以寻找违禁品的故事。理论上来说，这样的种族间的差异既可能是由于决策者（也就是本例中的警察）带有种族偏见导致的（偏好性歧视），也可能是由于非裔司机非法持有违禁品的比例更高导致的（统计性歧视）。在很多指控警察涉及种族归纳的法律诉讼中，原告坚持主张前者，而被告则主张后者。两方都会使用一些简单的统计证据来支持自己的论断，而这些证据却缺乏足够的说服力来解决这些争端。
A : Racial profiling was a hot public issue in the late 1990s, as it is again today. Stories about police officers disproportionally searching minority drivers for contrabands frequently appeared on local and national news. Theoretically, such racial disparities could result from either the decision makers, in this case, the police officers, being racially prejudiced, or from black drivers having a higher rate of carrying contraband. In numerous lawsuits filed against the police departments being accused of racial profiling, the plaintiff argues for the former, and the defendant argues for the latter. The two sides often use simple statistical evidence that could not convincingly resolve the disputes.
A : 当年我还是宾大的一位在读博士生时，宾大经济系彼时非常擅长将理论研究的洞见带入到公共政策辩论中去。例如，文献中一些很有影响力的文章如Coate and Loury (1993, AER)和Knowles, Persico and Todd (2001)都是由当时宾大的学者所著。早些时候，我就意识到，虽然Knowles, Persico and Todd (2001)中所采用的实证方法非常简洁，但是作者们其实意识到却回避了在应用Gary Becker的结果检验 （outcome test）这一想法的时候的一个关键问题 – 次边际问题 （infra-marginality problem）。我当时就想探究一下如果没有把次边际问题通过假设排除掉，我们能对造成种族差异的深层次原因有什么新的洞见。
A : When I was still a PhD student, Penn economics department was at the forefront of bringing theoretical insights to this important public policy debate; for example, important papers in the literature, Coate and Loury (1993, AER) and Knowles, Persico and Todd (2001), were written by scholars affiliated with Penn then. I realized early on that the empirical approach in Knowles, Persico and Todd (2001) was very neat, but it actually recognized and then avoided the key infra-marginality problem in the application of Gary Becker’s outcome test idea. I wanted to see what one can say about the underlying causes of racial disparity when the infra-marginality problem is not assumed away.
A : 说到这里，也许我应该先解释一下什么是“结果检验”。我之前说到“警察更频繁地搜查少数族裔司机以寻找违禁品”的时候，这一论断是基于警察所采取的“行动”的。而由Gary Becker首次提出的结果检验关注的是这些搜查到底产生了什么“结果”。Gary Becker的想法其实很简单：如果警察是由于种族偏见才更频繁地检查非裔司机的话，我们就应该可以预见到搜查非裔司机会有更低的成功率（也叫“命中率”）。但是，理论上说，上述论断只当我们比较白人司机与非裔司机的“边际”搜查命中率时才能成立。（编者注：即使没有种族偏见，白人司机与非裔司机是否被检查出持有违禁品的平均概率还是会受到他们各自种族不同的风险分布的影响，例如有可能非裔相比白人本身就更不倾向于持有违禁品。这就导致即使警察不存在种族偏见，非裔持有违禁品的平均概率仍然显著低于白人，就导致基于平均概率而非边际概率的结果检验不能很好地识别种族偏见。）但是，在数据中我们并无法判断哪些司机才是处于搜查“边际”的司机，因此我们也无法比较边际命中率 ，而只能比较平均命中率 。这个问题就是著名的结果检验中的次边际问题。我和我在耶鲁的博士学生Shamena Anwar所写的这篇文章就是通过利用非裔警察和白人警察作为共同的评判基准，部分解决了这个次边际问题。具体来说，如果非裔和白人警察有同等程度的种族偏见，那么对每个族裔，非裔和白人警察应该具有相同的边际搜查命中率。从而，我们就可以根据对不同种族司机的平均搜查命中率的排序对不同种族的警察是否相同来设计一个种族偏见的检验。
A : It may be useful to explain what outcome test is. When I earlier said that “police officers disproportionally searching minority drivers for contrabands”, the statement is about “actions” taken by the police officers. Outcome test, first proposed by Gary Becker, instead focuses on what happened to the searches. The idea is very simple: if police officers search black drivers more because they were racially prejudiced against them, then you would expect a lower search success rate (or known as the “hit rate”) from black searches. However, theoretically this statement is true only when one compares the marginal hit rates against black and white drivers. However, in the data we do not know which drivers are marginal drivers, and thus we are unable to compare the marginal hit rates, only the average hit rates. This problem is known as the infra-marginality problem. My paper with Shamena Anwar, who was my PhD student at Yale, aimed to partially address the infra-marginality problem by using black officers and white officers as mutual benchmarks. If black officers and white officers are equally racially prejudiced, then they should have the same marginal hit rates across different racial groups of drivers. Thus we are able to design a test of racial prejudice based on whether the rank order of averages hit rates across groups of drivers with different races is preserved across officers of different races.
Q: 从您的角度看，这篇文章具有如此重大影响力的主要原因是什么？
Q: From your perspective, what are the main reasons that make this paper a high impact research?
A : 这有几方面的原因。首先，区分种族偏见和统计性歧视是一个非常有现实意义的、经典的计量识别问题。这在当时是一个重要的公共议题，而今又重新回到了讨论的焦点。第二，次边际问题是一个大家都意识到的经典问题，而这个问题在当时还没有得到很好的解决。如我上面所提到的，Knowles, Persico and Todd (2001, JPE)通过一个理论模型来巧妙地回避了这个问题。在某种意义上，我们的文章填补了当时文献研究中一个大家都公认的空白。第三，这篇文章在理论上很严谨，实证研究的设计也与理论部分非常贴合。
A : There are several reasons. First and foremost, the issue of racial prejudice vs. statistical discrimination is a classical identification problem with high real-world relevance. It was an important public issue then as it is now. Second, the infra-marginality problem is a well-recognized problem, and there was no great solution at the time. As I mentioned, Knowles, Persico and Todd (2001, JPE) used a model to finesse the problem away. In a sense, our paper fills a gap that is well known by experts in the literature. Third, the paper is theoretically rigorous and empirically relevant.
Q: 在这篇文章写作与修改过程中最大的挑战是什么？
Q: What was the greatest challenge during the writing and revision of the paper?
A : 在那之前，我有关于这篇文章的想法已经很久了。我遇到的第一个挑战是如何获取数据。我们需要能够将搜查行为与搜查结果和警察的信息（尤其是警察的种族的信息）合并在一起的数据。我在网上搜索了很久，直到我发现了一个佛罗里达高速公路巡视的网站提到有可能有这样的信息。我邮件联系了网站上公布的联系人，询问是否有可能获得这个数据，非常让我惊喜的是他同意了我的请求。在我得到了数据之后，写作和修改这篇文章的过程实际上非常顺利。我们夏天完成了第一版初稿，之后我在一些讲座会议上宣讲了几次。然后我们就向AER投稿了，仅时隔三个月我们就收到了审稿人非常积极的审稿意见以及编辑的修改后再投稿（R&R）的通知。我们修改了一遍文章就被接受了。
A : I had the idea for the paper for quite some time. The first challenge was to obtain a data set that has information of the stopped and searched drivers linked to the information of the police officers (particularly the race of the officer). I searched the internet for quite some time until I came across a website of Florida Highway Patrol that mentioned that such information might be available. I emailed the contact person listed on the website to inquire about the possibility of obtaining the data set and was pleasantly surprised when he said yes. The writing and revision of the paper were actually very smooth once I obtained the data set. We completed a first draft in the summer and I presented it at a couple of seminars. We submitted to the AER and within three months received an R&R with very positive reports. We revised it and it got accepted.
Q: 您对于青年学者创作有影响力的高质量研究有什么建议？
Q: Do you have any advice for junior scholars on how to produce impactful high quality research?
A : 提供如何写作高质量论文的指南可不是容易的事情，事实上我想也很难有这样一个普适性的指南。如果我尝试着提一些提纲挈领的建议的话，我想，第一点就是要了解文献的前沿领域，并且精确地了解文献中最主要的需要填补的空白在哪里。影响力高的研究一定要能够为填补现有研究的空白做出一定贡献 。第二，好的研究应该是与重要的公共政策讨论紧密相关的 ，不论是我和Shamena这篇文章中讨论的关于种族偏见的问题，还是关于金融危机、人口老龄化、卫生服务体系改革、气候变化、自动化等等问题。经济学是一门社会科学，它的生命力最终还是来自于能够为我们的经济和社会所面临的重大挑战提供洞见。第三，高质量的研究应该兼具严谨的理论和实证事实或实证动机 。当然了，以上这些都只是高质量研究的必要而非充分条件。
A : It is not easy to provide a recipe for producing high-quality research, and I doubt there is any universal recipe. If I have to venture some high-level advice, I would say, first, one has to know the frontier of the literature and know precisely what are the major gaps in the literature. High impact research must make a contribution in filling the major gaps in the literature. Second, the research must be relevant to some important public policy discussions, whether it is racial profiling in the case of my paper with Shamena, or financial crisis, population aging, health care reform, climate change, or automation, etc. Economics is a social science, and its vitality comes from being able to shed insights on important challenges our economy and society face. Third, high quality research should combine theoretical rigor with empirical facts or empirical motivations. Of course, these are necessary, but not sufficient conditions.
学者简介：方汉明，宾夕法尼亚大学经济系Joseph M. Cohen讲席教授，上海科技大学创业与管理学院（海外）院长、特聘教授。方汉明教授是世界计量经济学会（The Econometric Society）会士（Fellow），美国国家经济研究局（NBER）研究员，并于2014至2016年担任中国经济工作组代理主任。他是人口研究中心和人口老龄化研究中心研究员、伦纳德·戴维斯卫生经济研究所（Leonard Davis Institute of Health Economics）高级研究员，以及当代中国研究中心执行委员会成员。他还曾担任新南威尔士大学澳中人口老龄化研究中心科学主任，新加坡亚洲经济和金融研究局高级研究员以及德国IZA研究员。他曾任Journal of Public Economics，International Economic Review等国际权威经济学期刊的联合编辑，目前是Journal of Risk and Insurance的资深编辑，也是Annual Review of Economics (2020-2024)编委员会成员，同时也是学术分享网站VoxChina的联合创始人。方汉明教授注重理论和实证的紧密结合，研究课题涵盖歧视等社会经济问题心理与经济、福利改革、拍卖、健康保险市场和人口老龄化。他在医疗保险市场方面的研究获得了第17届国际卫生经济学会（iHEA）颁发的最佳健康经济学肯尼斯·阿罗奖（Kenneth Arrow Prize）。
参考文献：
Anwar, S. and Fang, H., 2006. An alternative test of racial prejudice in motor vehicle searches: Theory and evidence. American Economic Review , 96(1), pp.127-151.
Fang, H., 2001. Social culture and economic performance. American Economic Review , 91(4), pp.924-937.
Bayer, P., Fang, H. and McMillan, R., 2014.Separate when equal? Racial inequality and residential segregation. Journal of Urban Economics , 82 , pp.32-48.
Anwar,S. and Fang, H., 2015. Testing for racial prejudice in the parole board release process: Theory and evidence. The Journal of Legal Studies , 44(1), pp.1-37.
Anwar,S. and Fang, H., 2012. Testing for the role of prejudice in emergency departments using bounceback rates. The BE Journal of Economic Analysis & Policy, 13(3).
| 中文翻译 | 李明 |
| 责任编辑 | 秦雨、韩亚婕、战昶威 |
| 校对 | 方汉明 |