Behavioral Responses to Taxation: Evidence from Field Experiments and Tax Policy Variation

本提案的目的是使用四种新颖的设计来估计对税收和转移计划的行为反应：(1)使用美国大型税务编制人员的随机现场实验，以估计有关大型转移计划（劳动所得税抵免）的信息对接受者随后的收入行为的影响。(2)利用丹麦税务机关开展的随机现场实验，分析税务审计和未来税务审计公告对税收合规的影响。(3)使用一个基于队列的工资税改革在希腊的工资税上限在1992年显著增加(但只对那些开始工作后,1992年)的发病率估计长期工资税对收入和劳动供给行为(4)使用非常大的房地产税率的变化程度的关系在欧洲大陆国家(如法国)相对于英语国家(如美国)评估房地产的影响对财富积累征税。所提议的活动的潜在智力价值是显著地推进我们对税收和政府转移的经验行为反应的认识。这种理解对于评估政府税收和转移政策的成本和收益至关重要，从而有助于决策者设计更好的税收和转移制度。为了在这个问题上取得进展，该提案提供了四种创新的研究设计。前两个设计涉及分析大型随机现场实验，而其余两个利用实际税法产生的变化。信息对税收行为反应的影响以前几乎没有被探索过，并且可能是反应大小的重要决定因素。如果这种影响很大，政府可以以适度的成本增加旨在促进某些行为的项目的信息。劳动所得税抵免信息实验在这个方向上提供了一个开创性的尝试。税收执法是成功实施税收的关键。税务机关通常对税务执法工作保密，使其特别难以分析。丹麦税务管理部门进行了一项大型随机税务审计现场实验，并将为我们的分析提供完整的数据访问。这种伙伴关系提供了一个独特的机会来严格评估税务执法的效果，并测试具体的执法战略。从长期来看，评估工资税的发生率效应以及劳动力供给对所得税的反应是出了名的困难，因为比较税制改革前后的结果的研究只衡量了对税收的短期反应。希腊基于队列的工资税改革是一项不同寻常的政策实验，因为它在同一个劳动力市场上创造了两组工人，他们将永远面临非常不同的工资税时间表，这仅仅取决于他们进入劳动力市场的日期。希腊社会保障局一直愿意提供详尽的行政数据来研究这项改革。最后，财富积累对遗产税的反应是有关遗产税政策辩论的核心。使用基于关系程度的税率变化（在一些国家，而不是其他国家）提供了一种在这个困难的实证问题上取得进展的新方法。产生的更广泛影响提出活动是开发一个模型,一个政府税收机构之间的合作关系(如丹麦税收执法的情况下项目或希腊工资税改革)或私人公司(如大的情况下对劳务所得收入抵免的税收填表人信息项目),为推进科学知识也可以富有成效的改进设计和实际税收和转移政策的实施。就劳动所得税抵免项目而言，我们还计划尝试与美国国税局建立合作关系，以获得更多的数据访问权限。访问公司或政府的大型管理数据库以及进行实地实验的可能性可能会变得越来越普遍。这项NSF拨款申请背后的一个更广泛的目标是促进斯堪的纳维亚国家等最先进国家为研究目的访问数据的经验，以改善美国等目前研究访问受限的国家的数据访问。

The objective of this proposal is to estimate behavioral responses to taxation and transfer programs using four novel designs: (1) Use a randomized field experiment with a large tax preparer in the United States in order to estimate the effect of information about a large transfer program--the Earned Income Tax Credit--on the subsequent earnings behavior of recipients. (2) Use a randomized field experiment carried out by the tax administration in Denmark in order to analyze the effects of tax audits and announcements about future tax audits on tax compliance. (3) Use a cohort based payroll tax reform in Greece where the ceiling on payroll taxes was significantly increased in 1992 (but only for those starting to work in 1992 and after) to estimate the long-run incidence of payroll taxes on earnings and labor supply behavior (4) Use the very large variation in estate tax rates by degree of relationship in continental European countries (such as France) vis-à-vis English speaking countries (such as the United States) to estimate the effect of estate taxation on wealth accumulation. The potential intellectual merit of the proposed activity is to significantly advance our knowledge of empirical behavioral responses to taxes and government transfers. Such an understanding is crucial to evaluate the costs and benefits of government tax and transfer policies and hence to help policy makers design better tax and transfer systems. The proposal offers four innovative research designs in order to make progress on this question. The first two of the designs involve analyzing large randomized field experiments, while the remaining two exploit variation created by actual tax laws. The effect of information on behavioral responses to taxation has hardly been explored before and could be a significant determinant to the size of the responses. If such effects are large, the government could increase information of programs designed to promote certain behaviors at a modest cost. The Earned Income Tax Credit information experiment offers a pioneering attempt in that direction. Tax enforcement is critical for successful tax implementation. Tax enforcement efforts are most often kept secret by tax authorities making them particularly difficult to analyze. The Danish tax administration carried out a large randomized tax audit field experiment and will give us full data access for the analysis. This partnership offers a unique opportunity to evaluate rigorously the effects of tax enforcement as well as test specific enforcement strategies. Evaluating the incidence effects of payroll taxes as well as the labor supply responses to earnings taxation in the long-run is notoriously difficult because studies comparing outcomes before and after a tax reform measures only the short-term response to taxation. The cohort based payroll tax reform in Greece is an unusual policy experiment because it creates two groups of workers in the same labor market which face permanently very different payroll tax schedules based solely on their date of entry in the labor force. The Greek Social Security Administration has been willing to supply exhaustive administrative data to study this reform. Finally, the response of wealth accumulation to estate taxation is central in the policy debate on the estate tax. Using variation in tax rates based on the relationship degree (in some countries but not others) offers a novel way to make progress on this difficult empirical question. The broader impact resulting from the proposed activity is to develop a model for partnerships between either a government tax agency (as in the case of the Danish tax enforcement project or the Greek payroll tax reform) or a private company (as in the case of the large tax preparer for the Earned Income Tax Credit information project) that could be fruitful both for advancing scientific knowledge but also for improving the design and the practical implementation of tax and transfer policies. In the case of the Earned Income Tax Credit project, we are also planning on trying to develop a partnership with the US Internal Revenue Service for additional data access. Access to company or government large administrative databases as well as the possibility to run field experiments is likely to become more and more prevalent. A broader goal behind this NSF grant application is to promote the experiences for data access for research purposes from the most advanced countries such as Scandinavian countries in order to improve data access in countries such as the United States where access for research is currently more restricted.