首页 > 最新文献

arXiv - ECON - Econometrics最新文献

英文 中文
Your MMM is Broken: Identification of Nonlinear and Time-varying Effects in Marketing Mix Models 你的 MMM 坏了:识别营销组合模型中的非线性和时变效应
Pub Date : 2024-08-14 DOI: arxiv-2408.07678
Ryan Dew, Nicolas Padilla, Anya Shchetkina
Recent years have seen a resurgence in interest in marketing mix models(MMMs), which are aggregate-level models of marketing effectiveness. Oftenthese models incorporate nonlinear effects, and either implicitly or explicitlyassume that marketing effectiveness varies over time. In this paper, we showthat nonlinear and time-varying effects are often not identifiable fromstandard marketing mix data: while certain data patterns may be suggestive ofnonlinear effects, such patterns may also emerge under simpler models thatincorporate dynamics in marketing effectiveness. This lack of identification isproblematic because nonlinearities and dynamics suggest fundamentally differentoptimal marketing allocations. We examine this identification issue throughtheory and simulations, wherein we explore the exact conditions under whichconflation between the two types of models is likely to occur. In doing so, weintroduce a flexible Bayesian nonparametric model that allows us to bothflexibly simulate and estimate different data-generating processes. We showthat conflating the two types of effects is especially likely in the presenceof autocorrelated marketing variables, which are common in practice, especiallygiven the widespread use of stock variables to capture long-run effects ofadvertising. We illustrate these ideas through numerous empirical applicationsto real-world marketing mix data, showing the prevalence of the conflationissue in practice. Finally, we show how marketers can avoid this conflation, bydesigning experiments that strategically manipulate spending in ways that pindown model form.
近年来,市场营销组合模型(MMMs)再次引起人们的关注,这些模型是市场营销效果的综合模型。这些模型通常包含非线性效应,并且或隐或显地假定营销效果随时间而变化。本文表明,非线性效应和时变效应往往无法从标准营销组合数据中识别出来:虽然某些数据模式可能暗示了非线性效应,但这些模式也可能出现在包含营销效果动态变化的简单模型中。由于非线性效应和动态效应会带来根本不同的最优营销分配,因此缺乏识别性是个问题。我们通过理论和模拟研究了这一识别问题,探索了两类模型之间可能发生冲突的确切条件。在此过程中,我们引入了一个灵活的贝叶斯非参数模型,使我们能够灵活地模拟和估计不同的数据生成过程。我们表明,在存在自相关营销变量的情况下,混淆这两类效应的可能性尤其大,而自相关营销变量在实践中很常见,特别是考虑到股票变量被广泛用于捕捉广告的长期效应。我们通过对现实世界营销组合数据的大量实证应用来说明这些观点,从而显示出混淆问题在实践中的普遍性。最后,我们展示了营销人员如何通过设计实验来避免这种混淆,这些实验可以战略性地操纵支出,从而打破模型形式。
{"title":"Your MMM is Broken: Identification of Nonlinear and Time-varying Effects in Marketing Mix Models","authors":"Ryan Dew, Nicolas Padilla, Anya Shchetkina","doi":"arxiv-2408.07678","DOIUrl":"https://doi.org/arxiv-2408.07678","url":null,"abstract":"Recent years have seen a resurgence in interest in marketing mix models\u0000(MMMs), which are aggregate-level models of marketing effectiveness. Often\u0000these models incorporate nonlinear effects, and either implicitly or explicitly\u0000assume that marketing effectiveness varies over time. In this paper, we show\u0000that nonlinear and time-varying effects are often not identifiable from\u0000standard marketing mix data: while certain data patterns may be suggestive of\u0000nonlinear effects, such patterns may also emerge under simpler models that\u0000incorporate dynamics in marketing effectiveness. This lack of identification is\u0000problematic because nonlinearities and dynamics suggest fundamentally different\u0000optimal marketing allocations. We examine this identification issue through\u0000theory and simulations, wherein we explore the exact conditions under which\u0000conflation between the two types of models is likely to occur. In doing so, we\u0000introduce a flexible Bayesian nonparametric model that allows us to both\u0000flexibly simulate and estimate different data-generating processes. We show\u0000that conflating the two types of effects is especially likely in the presence\u0000of autocorrelated marketing variables, which are common in practice, especially\u0000given the widespread use of stock variables to capture long-run effects of\u0000advertising. We illustrate these ideas through numerous empirical applications\u0000to real-world marketing mix data, showing the prevalence of the conflation\u0000issue in practice. Finally, we show how marketers can avoid this conflation, by\u0000designing experiments that strategically manipulate spending in ways that pin\u0000down model form.","PeriodicalId":501293,"journal":{"name":"arXiv - ECON - Econometrics","volume":"24 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-08-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142184221","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
引用次数: 0
Quantile and Distribution Treatment Effects on the Treated with Possibly Non-Continuous Outcomes 对结果可能不连续的受治疗者的定量和分布治疗效果
Pub Date : 2024-08-14 DOI: arxiv-2408.07842
Nelly K. Djuazon, Emmanuel Selorm Tsyawo
Quantile and Distribution Treatment effects on the Treated (QTT/DTT) fornon-continuous outcomes are either not identified or inference thereon isinfeasible using existing methods. By introducing functional index paralleltrends and no anticipation assumptions, this paper identifies and providesuniform inference procedures for QTT/DTT. The inference procedure applies underboth the canonical two-group and staggered treatment designs with balancedpanels, unbalanced panels, or repeated cross-sections. Monte Carlo experimentsdemonstrate the proposed method's robust and competitive performance, while anempirical application illustrates its practical utility.
对于非连续性结果的定量效应和分布效应(QTT/DTT),现有方法要么无法识别,要么无法推断。本文通过引入函数指数平行趋势和无预期假设,确定并提供了 QTT/DTT 的统一推断程序。该推断程序适用于平衡面板、非平衡面板或重复横截面的典型两组设计和交错处理设计。蒙特卡洛实验证明了所提方法的稳健性和竞争力,而经验应用则说明了该方法的实用性。
{"title":"Quantile and Distribution Treatment Effects on the Treated with Possibly Non-Continuous Outcomes","authors":"Nelly K. Djuazon, Emmanuel Selorm Tsyawo","doi":"arxiv-2408.07842","DOIUrl":"https://doi.org/arxiv-2408.07842","url":null,"abstract":"Quantile and Distribution Treatment effects on the Treated (QTT/DTT) for\u0000non-continuous outcomes are either not identified or inference thereon is\u0000infeasible using existing methods. By introducing functional index parallel\u0000trends and no anticipation assumptions, this paper identifies and provides\u0000uniform inference procedures for QTT/DTT. The inference procedure applies under\u0000both the canonical two-group and staggered treatment designs with balanced\u0000panels, unbalanced panels, or repeated cross-sections. Monte Carlo experiments\u0000demonstrate the proposed method's robust and competitive performance, while an\u0000empirical application illustrates its practical utility.","PeriodicalId":501293,"journal":{"name":"arXiv - ECON - Econometrics","volume":"34 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-08-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142184220","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
引用次数: 0
What are the real implications for $CO_2$ as generation from renewables increases? 随着可再生能源发电量的增加,CO_2$ 的实际影响是什么?
Pub Date : 2024-08-09 DOI: arxiv-2408.05209
Dhruv Suri, Jacques de Chalendar, Ines Azevedo
Wind and solar electricity generation account for 14% of total electricitygeneration in the United States and are expected to continue to grow in thenext decades. In low carbon systems, generation from renewable energy sourcesdisplaces conventional fossil fuel power plants resulting in lower system-levelemissions and emissions intensity. However, we find that intermittentgeneration from renewables changes the way conventional thermal power plantsoperate, and that the displacement of generation is not 1 to 1 as expected. Ourwork provides a method that allows policy and decision makers to continue totrack the effect of additional renewable capacity and the resulting thermalpower plant operational responses.
风能和太阳能发电占美国总发电量的 14%,预计未来几十年还将继续增长。在低碳系统中,可再生能源发电取代了传统的化石燃料发电厂,从而降低了系统排放和排放强度。然而,我们发现可再生能源的间歇性发电改变了传统火力发电厂的运行方式,而且发电量的替代并非如预期的 1 比 1。我们的工作提供了一种方法,使政策和决策者能够继续跟踪新增可再生能源发电能力的影响以及由此产生的火力发电厂运行响应。
{"title":"What are the real implications for $CO_2$ as generation from renewables increases?","authors":"Dhruv Suri, Jacques de Chalendar, Ines Azevedo","doi":"arxiv-2408.05209","DOIUrl":"https://doi.org/arxiv-2408.05209","url":null,"abstract":"Wind and solar electricity generation account for 14% of total electricity\u0000generation in the United States and are expected to continue to grow in the\u0000next decades. In low carbon systems, generation from renewable energy sources\u0000displaces conventional fossil fuel power plants resulting in lower system-level\u0000emissions and emissions intensity. However, we find that intermittent\u0000generation from renewables changes the way conventional thermal power plants\u0000operate, and that the displacement of generation is not 1 to 1 as expected. Our\u0000work provides a method that allows policy and decision makers to continue to\u0000track the effect of additional renewable capacity and the resulting thermal\u0000power plant operational responses.","PeriodicalId":501293,"journal":{"name":"arXiv - ECON - Econometrics","volume":"1 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-08-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141947516","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
引用次数: 0
Difference-in-Differences for Health Policy and Practice: A Review of Modern Methods 卫生政策与实践中的差异:现代方法综述
Pub Date : 2024-08-08 DOI: arxiv-2408.04617
Shuo Feng, Ishani Ganguli, Youjin Lee, John Poe, Andrew Ryan, Alyssa Bilinski
Difference-in-differences (DiD) is the most popular observational causalinference method in health policy, employed to evaluate the real-world impactof policies and programs. To estimate treatment effects, DiD relies on the"parallel trends assumption", that on average treatment and comparison groupswould have had parallel trajectories in the absence of an intervention.Historically, DiD has been considered broadly applicable and straightforward toimplement, but recent years have seen rapid advancements in DiD methods. Thispaper reviews and synthesizes these innovations for medical and health policyresearchers. We focus on four topics: (1) assessing the parallel trendsassumption in health policy contexts; (2) relaxing the parallel trendsassumption when appropriate; (3) employing estimators to account for staggeredtreatment timing; and (4) conducting robust inference for analyses in whichnormal-based clustered standard errors are inappropriate. For each, we explainchallenges and common pitfalls in traditional DiD and modern methods availableto address these issues.
差异推断法(DiD)是卫生政策领域最常用的观察因果推断方法,用于评估政策和项目在现实世界中的影响。为了估计治疗效果,差分法依赖于 "平行趋势假设",即在没有干预措施的情况下,治疗组和比较组的平均轨迹是平行的。本文为医学和卫生政策研究人员回顾并总结了这些创新。我们重点关注四个主题:(1) 评估卫生政策背景下的平行趋势假设;(2) 在适当的时候放宽平行趋势假设;(3) 使用估计器来考虑治疗时间的错开;(4) 在基于正态分布的聚类标准误差不合适的分析中进行稳健推断。我们将分别解释传统 DiD 中的挑战和常见陷阱,以及解决这些问题的现代方法。
{"title":"Difference-in-Differences for Health Policy and Practice: A Review of Modern Methods","authors":"Shuo Feng, Ishani Ganguli, Youjin Lee, John Poe, Andrew Ryan, Alyssa Bilinski","doi":"arxiv-2408.04617","DOIUrl":"https://doi.org/arxiv-2408.04617","url":null,"abstract":"Difference-in-differences (DiD) is the most popular observational causal\u0000inference method in health policy, employed to evaluate the real-world impact\u0000of policies and programs. To estimate treatment effects, DiD relies on the\u0000\"parallel trends assumption\", that on average treatment and comparison groups\u0000would have had parallel trajectories in the absence of an intervention.\u0000Historically, DiD has been considered broadly applicable and straightforward to\u0000implement, but recent years have seen rapid advancements in DiD methods. This\u0000paper reviews and synthesizes these innovations for medical and health policy\u0000researchers. We focus on four topics: (1) assessing the parallel trends\u0000assumption in health policy contexts; (2) relaxing the parallel trends\u0000assumption when appropriate; (3) employing estimators to account for staggered\u0000treatment timing; and (4) conducting robust inference for analyses in which\u0000normal-based clustered standard errors are inappropriate. For each, we explain\u0000challenges and common pitfalls in traditional DiD and modern methods available\u0000to address these issues.","PeriodicalId":501293,"journal":{"name":"arXiv - ECON - Econometrics","volume":"28 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-08-08","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141947554","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
引用次数: 0
Vela: A Data-Driven Proposal for Joint Collaboration in Space Exploration Vela:以数据为导向的太空探索联合合作建议书
Pub Date : 2024-08-08 DOI: arxiv-2408.04730
Holly M. Dinkel, Jason K. Cornelius
The UN Office of Outer Space Affairs identifies synergy of space developmentactivities and international cooperation through data and infrastructuresharing in their Sustainable Development Goal 17 (SDG17). Current multilateralspace exploration paradigms, however, are divided between the Artemis and theRoscosmos-CNSA programs to return to the moon and establish permanent humansettlements. As space agencies work to expand human presence in space, economicresource consolidation in pursuit of technologically ambitious spaceexpeditions is the most sensible path to accomplish SDG17. This paper compilesa budget dataset for the top five federally-funded space agencies: CNSA, ESA,JAXA, NASA, and Roscosmos. Using time-series econometric anslysis methods inSTATA, this work analyzes each agency's economic contributions toward spaceexploration. The dataset results are used to propose a multinational spacemission, Vela, for the development of an orbiting space station around Mars inthe late 2030s. Distribution of economic resources and technologicalcapabilities by the respective space programs are proposed to ensureprogrammatic redundancy and increase the odds of success on the given timeline.
联合国外层空间事务办公室在其可持续发展目标 17(SDG17)中确定了通过数据和基础设施共享实现空间发展活动和国际合作的协同作用。然而,目前的多边空间探索范式分为阿尔忒弥斯计划和俄罗斯航天局-中国国家航天局重返月球计划,以及建立人类永久居住地计划。随着太空机构努力扩大人类在太空的存在,通过经济资源整合来追求技术上雄心勃勃的太空探索,是实现可持续发展目标 17 的最明智途径。本文汇编了联邦政府资助的五大航天机构的预算数据集:中国国家航天局(CNSA)、欧洲航天局(ESA)、日本宇宙航空研究开发机构(JAXA)、美国国家航空航天局(NASA)和俄罗斯航天局(Roscosmos)。利用 STATA 中的时间序列计量经济学分析方法,本文分析了各机构对太空探索的经济贡献。数据集的结果被用于提出一个多国太空发射计划--"维拉 "计划,以在 2030 年代末开发一个环绕火星的轨道空间站。提出了各太空计划的经济资源和技术能力分配方案,以确保计划的冗余性,提高在既定时间内取得成功的几率。
{"title":"Vela: A Data-Driven Proposal for Joint Collaboration in Space Exploration","authors":"Holly M. Dinkel, Jason K. Cornelius","doi":"arxiv-2408.04730","DOIUrl":"https://doi.org/arxiv-2408.04730","url":null,"abstract":"The UN Office of Outer Space Affairs identifies synergy of space development\u0000activities and international cooperation through data and infrastructure\u0000sharing in their Sustainable Development Goal 17 (SDG17). Current multilateral\u0000space exploration paradigms, however, are divided between the Artemis and the\u0000Roscosmos-CNSA programs to return to the moon and establish permanent human\u0000settlements. As space agencies work to expand human presence in space, economic\u0000resource consolidation in pursuit of technologically ambitious space\u0000expeditions is the most sensible path to accomplish SDG17. This paper compiles\u0000a budget dataset for the top five federally-funded space agencies: CNSA, ESA,\u0000JAXA, NASA, and Roscosmos. Using time-series econometric anslysis methods in\u0000STATA, this work analyzes each agency's economic contributions toward space\u0000exploration. The dataset results are used to propose a multinational space\u0000mission, Vela, for the development of an orbiting space station around Mars in\u0000the late 2030s. Distribution of economic resources and technological\u0000capabilities by the respective space programs are proposed to ensure\u0000programmatic redundancy and increase the odds of success on the given timeline.","PeriodicalId":501293,"journal":{"name":"arXiv - ECON - Econometrics","volume":"10 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-08-08","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141947518","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
引用次数: 0
Semiparametric Estimation of Individual Coefficients in a Dyadic Link Formation Model Lacking Observable Characteristics 缺乏可观测特征的二元联系形成模型中个体系数的半参数估计
Pub Date : 2024-08-08 DOI: arxiv-2408.04552
L. Sanna Stephan
Dyadic network formation models have wide applicability in economic research,yet are difficult to estimate in the presence of individual specific effectsand in the absence of distributional assumptions regarding the model noisecomponent. The availability of (continuously distributed) individual or linkcharacteristics generally facilitates estimation. Yet, while data on socialnetworks has recently become more abundant, the characteristics of the entitiesinvolved in the link may not be measured. Adapting the procedure of citet{KS},I propose to use network data alone in a semiparametric estimation of theindividual fixed effect coefficients, which carry the interpretation of theindividual relative popularity. This entails the possibility to anticipate howa new-coming individual will connect in a pre-existing group. The estimator,needed for its fast convergence, fails to implement the monotonicity assumptionregarding the model noise component, thereby potentially reversing the order ifthe fixed effect coefficients. This and other numerical issues can beconveniently tackled by my novel, data-driven way of normalising the fixedeffects, which proves to outperform a conventional standardisation in manycases. I demonstrate that the normalised coefficients converge both at the samerate and to the same limiting distribution as if the true error distributionwas known. The cost of semiparametric estimation is thus purely computational,while the potential benefits are large whenever the errors have a stronglyconvex or strongly concave distribution.
二元网络形成模型在经济研究中具有广泛的适用性,但在存在个体特定效应和缺乏模型噪声成分分布假设的情况下,很难进行估计。个人或链接特征(连续分布)的可用性通常有助于估算。然而,虽然社会网络的数据最近变得越来越丰富,但参与链接的实体的特征可能无法测量。通过改编 citet{KS} 的程序,我提议仅使用网络数据对个体固定效应系数进行半参数估计,该系数承载了对个体相对受欢迎程度的解释。这就意味着有可能预测一个新来的个体将如何在已有群体中建立联系。快速收敛所需的估计器未能实现关于模型噪声成分的单调性假设,因此有可能颠倒固定效应系数的顺序。我采用数据驱动的新方法对固定效应系数进行归一化处理,可以方便地解决这一问题和其他数值问题。我证明了归一化系数收敛于同一水平,并收敛于相同的极限分布,就像已知的真实误差分布一样。因此,半参数估计的成本纯粹是计算成本,而当误差具有强凸或强凹分布时,半参数估计的潜在优势则非常大。
{"title":"Semiparametric Estimation of Individual Coefficients in a Dyadic Link Formation Model Lacking Observable Characteristics","authors":"L. Sanna Stephan","doi":"arxiv-2408.04552","DOIUrl":"https://doi.org/arxiv-2408.04552","url":null,"abstract":"Dyadic network formation models have wide applicability in economic research,\u0000yet are difficult to estimate in the presence of individual specific effects\u0000and in the absence of distributional assumptions regarding the model noise\u0000component. The availability of (continuously distributed) individual or link\u0000characteristics generally facilitates estimation. Yet, while data on social\u0000networks has recently become more abundant, the characteristics of the entities\u0000involved in the link may not be measured. Adapting the procedure of citet{KS},\u0000I propose to use network data alone in a semiparametric estimation of the\u0000individual fixed effect coefficients, which carry the interpretation of the\u0000individual relative popularity. This entails the possibility to anticipate how\u0000a new-coming individual will connect in a pre-existing group. The estimator,\u0000needed for its fast convergence, fails to implement the monotonicity assumption\u0000regarding the model noise component, thereby potentially reversing the order if\u0000the fixed effect coefficients. This and other numerical issues can be\u0000conveniently tackled by my novel, data-driven way of normalising the fixed\u0000effects, which proves to outperform a conventional standardisation in many\u0000cases. I demonstrate that the normalised coefficients converge both at the same\u0000rate and to the same limiting distribution as if the true error distribution\u0000was known. The cost of semiparametric estimation is thus purely computational,\u0000while the potential benefits are large whenever the errors have a strongly\u0000convex or strongly concave distribution.","PeriodicalId":501293,"journal":{"name":"arXiv - ECON - Econometrics","volume":"63 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-08-08","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141947519","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
引用次数: 0
Robust Estimation of Regression Models with Potentially Endogenous Outliers via a Modern Optimization Lens 通过现代优化透镜对具有潜在内生异常值的回归模型进行稳健估计
Pub Date : 2024-08-07 DOI: arxiv-2408.03930
Zhan Gao, Hyungsik Roger Moon
This paper addresses the robust estimation of linear regression models in thepresence of potentially endogenous outliers. Through Monte Carlo simulations,we demonstrate that existing $L_1$-regularized estimation methods, includingthe Huber estimator and the least absolute deviation (LAD) estimator, exhibitsignificant bias when outliers are endogenous. Motivated by this finding, weinvestigate $L_0$-regularized estimation methods. We propose systematicheuristic algorithms, notably an iterative hard-thresholding algorithm and alocal combinatorial search refinement, to solve the combinatorial optimizationproblem of the (L_0)-regularized estimation efficiently. Our Monte Carlosimulations yield two key results: (i) The local combinatorial search algorithmsubstantially improves solution quality compared to the initialprojection-based hard-thresholding algorithm while offering greatercomputational efficiency than directly solving the mixed integer optimizationproblem. (ii) The $L_0$-regularized estimator demonstrates superior performancein terms of bias reduction, estimation accuracy, and out-of-sample predictionerrors compared to $L_1$-regularized alternatives. We illustrate the practicalvalue of our method through an empirical application to stock returnforecasting.
本文探讨了存在潜在内生异常值时线性回归模型的稳健估计问题。通过蒙特卡罗模拟,我们证明了现有的 $L_1$ 规则化估计方法,包括 Huber 估计器和最小绝对偏差 (LAD) 估计器,在异常值是内生的情况下表现出明显的偏差。受此启发,我们对 L_0$ 规则化估计方法进行了研究。我们提出了系统的启发式算法,特别是迭代硬阈值算法和局部组合搜索改进算法,以有效解决(L_0)规则化估计的组合优化问题。我们的蒙特卡洛模拟得出了两个关键结果:(i) 与基于初始投影的硬阈值算法相比,局部组合搜索算法大大提高了求解质量,同时比直接求解混合整数优化问题具有更高的计算效率。(ii) 与 L_1$ 规则化替代方法相比,L_0$ 规则化估计器在减少偏差、估计精度和样本外预测误差方面表现出更优越的性能。我们通过股票回报预测的经验应用来说明我们方法的实用价值。
{"title":"Robust Estimation of Regression Models with Potentially Endogenous Outliers via a Modern Optimization Lens","authors":"Zhan Gao, Hyungsik Roger Moon","doi":"arxiv-2408.03930","DOIUrl":"https://doi.org/arxiv-2408.03930","url":null,"abstract":"This paper addresses the robust estimation of linear regression models in the\u0000presence of potentially endogenous outliers. Through Monte Carlo simulations,\u0000we demonstrate that existing $L_1$-regularized estimation methods, including\u0000the Huber estimator and the least absolute deviation (LAD) estimator, exhibit\u0000significant bias when outliers are endogenous. Motivated by this finding, we\u0000investigate $L_0$-regularized estimation methods. We propose systematic\u0000heuristic algorithms, notably an iterative hard-thresholding algorithm and a\u0000local combinatorial search refinement, to solve the combinatorial optimization\u0000problem of the (L_0)-regularized estimation efficiently. Our Monte Carlo\u0000simulations yield two key results: (i) The local combinatorial search algorithm\u0000substantially improves solution quality compared to the initial\u0000projection-based hard-thresholding algorithm while offering greater\u0000computational efficiency than directly solving the mixed integer optimization\u0000problem. (ii) The $L_0$-regularized estimator demonstrates superior performance\u0000in terms of bias reduction, estimation accuracy, and out-of-sample prediction\u0000errors compared to $L_1$-regularized alternatives. We illustrate the practical\u0000value of our method through an empirical application to stock return\u0000forecasting.","PeriodicalId":501293,"journal":{"name":"arXiv - ECON - Econometrics","volume":"7 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-08-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141947556","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
引用次数: 0
Robust Identification in Randomized Experiments with Noncompliance 有违规行为的随机试验中的稳健识别
Pub Date : 2024-08-07 DOI: arxiv-2408.03530
Yi Cui, Désiré Kédagni, Huan Wu
This paper considers a robust identification of causal parameters in arandomized experiment setting with noncompliance where the standard localaverage treatment effect assumptions could be violated. Following Li,K'edagni, and Mourifi'e (2024), we propose a misspecification robust boundfor a real-valued vector of various causal parameters. We discussidentification under two sets of weaker assumptions: random assignment andexclusion restriction (without monotonicity), and random assignment andmonotonicity (without exclusion restriction). We introduce two causalparameters: the local average treatment-controlled direct effect (LATCDE), andthe local average instrument-controlled direct effect (LAICDE). Under therandom assignment and monotonicity assumptions, we derive sharp bounds on thelocal average treatment-controlled direct effects for the always-takers andnever-takers, respectively, and the total average controlled direct effect forthe compliers. Additionally, we show that the intent-to-treat effect can beexpressed as a convex weighted average of these three effects. Finally, weapply our method on the proximity to college instrument and find that growingup near a four-year college increases the wage of never-takers (who representmore than 70% of the population) by a range of 4.15% to 27.07%.
本文考虑的是在随机实验设置中的因果参数稳健识别问题,在这种设置中,标准的局部平均治疗效果假设可能会被违反。继 Li、K'edagni 和 Mourifi'e (2024)之后,我们为各种因果参数的实值向量提出了一个误设稳健约束。我们讨论了两组较弱假设下的识别:随机分配和排除限制(无单调性),以及随机分配和单调性(无排除限制)。我们引入了两个因果参数:当地平均治疗控制直接效应(LATCDE)和当地平均工具控制直接效应(LAICDE)。在随机分配和单调性假设下,我们分别推导出了 "总是接受者 "和 "从不接受者 "的本地平均治疗控制直接效应,以及 "遵守者 "的总平均控制直接效应的尖锐界限。此外,我们还证明,意向治疗效果可以表示为这三种效果的凸加权平均值。最后,我们将我们的方法应用于靠近大学的工具上,发现在四年制大学附近长大的人,会使从不接受治疗者(占总人口的 70% 以上)的工资增加 4.15% 到 27.07%。
{"title":"Robust Identification in Randomized Experiments with Noncompliance","authors":"Yi Cui, Désiré Kédagni, Huan Wu","doi":"arxiv-2408.03530","DOIUrl":"https://doi.org/arxiv-2408.03530","url":null,"abstract":"This paper considers a robust identification of causal parameters in a\u0000randomized experiment setting with noncompliance where the standard local\u0000average treatment effect assumptions could be violated. Following Li,\u0000K'edagni, and Mourifi'e (2024), we propose a misspecification robust bound\u0000for a real-valued vector of various causal parameters. We discuss\u0000identification under two sets of weaker assumptions: random assignment and\u0000exclusion restriction (without monotonicity), and random assignment and\u0000monotonicity (without exclusion restriction). We introduce two causal\u0000parameters: the local average treatment-controlled direct effect (LATCDE), and\u0000the local average instrument-controlled direct effect (LAICDE). Under the\u0000random assignment and monotonicity assumptions, we derive sharp bounds on the\u0000local average treatment-controlled direct effects for the always-takers and\u0000never-takers, respectively, and the total average controlled direct effect for\u0000the compliers. Additionally, we show that the intent-to-treat effect can be\u0000expressed as a convex weighted average of these three effects. Finally, we\u0000apply our method on the proximity to college instrument and find that growing\u0000up near a four-year college increases the wage of never-takers (who represent\u0000more than 70% of the population) by a range of 4.15% to 27.07%.","PeriodicalId":501293,"journal":{"name":"arXiv - ECON - Econometrics","volume":"59 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-08-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141947555","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
引用次数: 0
Efficient Asymmetric Causality Tests 有效的非对称因果检验
Pub Date : 2024-08-06 DOI: arxiv-2408.03137
Abdulnasser Hatemi-J
Asymmetric causality tests are increasingly gaining popularity in differentscientific fields. This approach corresponds better to reality since logicalreasons behind asymmetric behavior exist and need to be considered in empiricalinvestigations. Hatemi-J (2012) introduced the asymmetric causality tests viapartial cumulative sums for positive and negative components of the variablesoperating within the vector autoregressive (VAR) model. However, since the theresiduals across the equations in the VAR model are not independent, theordinary least squares method for estimating the parameters is not efficient.Additionally, asymmetric causality tests mean having different causalparameters (i.e., for positive or negative components), thus, it is crucial toassess not only if these causal parameters are individually statisticallysignificant, but also if their difference is statistically significant.Consequently, tests of difference between estimated causal parameters shouldexplicitly be conducted, which are neglected in the existing literature. Thepurpose of the current paper is to deal with these issues explicitly. Anapplication is provided, and ten different hypotheses pertinent to theasymmetric causal interaction between two largest financial markets worldwideare efficiently tested within a multivariate setting.
非对称因果关系检验在不同科学领域越来越受欢迎。这种方法更符合实际情况,因为非对称行为背后存在逻辑原因,需要在实证研究中加以考虑。Hatemi-J (2012)引入了非对称因果检验,即对向量自回归(VAR)模型中运行变量的正负分量进行部分累积求和。此外,非对称因果检验意味着有不同的因果参数(即正负分量),因此,不仅要评估这些因果参数是否单独具有统计意义,还要评估它们之间的差异是否具有统计意义。本文旨在明确处理这些问题。本文提供了一个应用,并在多变量环境下有效检验了与全球两个最大金融市场之间非对称因果互动相关的十个不同假设。
{"title":"Efficient Asymmetric Causality Tests","authors":"Abdulnasser Hatemi-J","doi":"arxiv-2408.03137","DOIUrl":"https://doi.org/arxiv-2408.03137","url":null,"abstract":"Asymmetric causality tests are increasingly gaining popularity in different\u0000scientific fields. This approach corresponds better to reality since logical\u0000reasons behind asymmetric behavior exist and need to be considered in empirical\u0000investigations. Hatemi-J (2012) introduced the asymmetric causality tests via\u0000partial cumulative sums for positive and negative components of the variables\u0000operating within the vector autoregressive (VAR) model. However, since the the\u0000residuals across the equations in the VAR model are not independent, the\u0000ordinary least squares method for estimating the parameters is not efficient.\u0000Additionally, asymmetric causality tests mean having different causal\u0000parameters (i.e., for positive or negative components), thus, it is crucial to\u0000assess not only if these causal parameters are individually statistically\u0000significant, but also if their difference is statistically significant.\u0000Consequently, tests of difference between estimated causal parameters should\u0000explicitly be conducted, which are neglected in the existing literature. The\u0000purpose of the current paper is to deal with these issues explicitly. An\u0000application is provided, and ten different hypotheses pertinent to the\u0000asymmetric causal interaction between two largest financial markets worldwide\u0000are efficiently tested within a multivariate setting.","PeriodicalId":501293,"journal":{"name":"arXiv - ECON - Econometrics","volume":"90 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-08-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141947517","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
引用次数: 0
A nonparametric test for diurnal variation in spot correlation processes 定点相关过程日变化的非参数检验
Pub Date : 2024-08-05 DOI: arxiv-2408.02757
Kim Christensen, Ulrich Hounyo, Zhi Liu
The association between log-price increments of exchange-traded equities, asmeasured by their spot correlation estimated from high-frequency data, exhibitsa pronounced upward-sloping and almost piecewise linear relationship at theintraday horizon. There is notably lower-on average less positive-correlationin the morning than in the afternoon. We develop a nonparametric testingprocedure to detect such deterministic variation in a correlation process. Thetest statistic has a known distribution under the null hypothesis, whereas itdiverges under the alternative. It is robust against stochastic correlation. Werun a Monte Carlo simulation to discover the finite sample properties of thetest statistic, which are close to the large sample predictions, even for smallsample sizes and realistic levels of diurnal variation. In an application, weimplement the test on a monthly basis for a high-frequency dataset covering thestock market over an extended period. The test leads to rejection of the nullmost of the time. This suggests diurnal variation in the correlation process isa nontrivial effect in practice.
根据高频数据估算的现货相关性衡量的交易所交易股票对数价格增量之间的关联,在当日范围内呈现出明显的向上倾斜和几乎成片的线性关系。与下午相比,上午的正相关性明显较低,平均较低。我们开发了一种非参数检验程序来检测相关过程中的这种确定性变化。在零假设下,检验统计量的分布是已知的,而在备择假设下,检验统计量的分布是偏离的。它对随机相关性具有稳健性。我们通过蒙特卡罗模拟发现了测试统计量的有限样本特性,即使在样本量较小和昼夜变化水平较高的情况下,测试统计量也接近于大样本预测值。在应用中,我们按月对覆盖股市较长时间的高频数据集进行了检验。该检验在大多数情况下都拒绝接受 null。这表明相关过程中的昼夜变化在实践中具有非同小可的影响。
{"title":"A nonparametric test for diurnal variation in spot correlation processes","authors":"Kim Christensen, Ulrich Hounyo, Zhi Liu","doi":"arxiv-2408.02757","DOIUrl":"https://doi.org/arxiv-2408.02757","url":null,"abstract":"The association between log-price increments of exchange-traded equities, as\u0000measured by their spot correlation estimated from high-frequency data, exhibits\u0000a pronounced upward-sloping and almost piecewise linear relationship at the\u0000intraday horizon. There is notably lower-on average less positive-correlation\u0000in the morning than in the afternoon. We develop a nonparametric testing\u0000procedure to detect such deterministic variation in a correlation process. The\u0000test statistic has a known distribution under the null hypothesis, whereas it\u0000diverges under the alternative. It is robust against stochastic correlation. We\u0000run a Monte Carlo simulation to discover the finite sample properties of the\u0000test statistic, which are close to the large sample predictions, even for small\u0000sample sizes and realistic levels of diurnal variation. In an application, we\u0000implement the test on a monthly basis for a high-frequency dataset covering the\u0000stock market over an extended period. The test leads to rejection of the null\u0000most of the time. This suggests diurnal variation in the correlation process is\u0000a nontrivial effect in practice.","PeriodicalId":501293,"journal":{"name":"arXiv - ECON - Econometrics","volume":"90 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-08-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141947448","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
引用次数: 0
期刊
arXiv - ECON - Econometrics
全部 Acc. Chem. Res. ACS Applied Bio Materials ACS Appl. Electron. Mater. ACS Appl. Energy Mater. ACS Appl. Mater. Interfaces ACS Appl. Nano Mater. ACS Appl. Polym. Mater. ACS BIOMATER-SCI ENG ACS Catal. ACS Cent. Sci. ACS Chem. Biol. ACS Chemical Health & Safety ACS Chem. Neurosci. ACS Comb. Sci. ACS Earth Space Chem. ACS Energy Lett. ACS Infect. Dis. ACS Macro Lett. ACS Mater. Lett. ACS Med. Chem. Lett. ACS Nano ACS Omega ACS Photonics ACS Sens. ACS Sustainable Chem. Eng. ACS Synth. Biol. Anal. Chem. BIOCHEMISTRY-US Bioconjugate Chem. BIOMACROMOLECULES Chem. Res. Toxicol. Chem. Rev. Chem. Mater. CRYST GROWTH DES ENERG FUEL Environ. Sci. Technol. Environ. Sci. Technol. Lett. Eur. J. Inorg. Chem. IND ENG CHEM RES Inorg. Chem. J. Agric. Food. Chem. J. Chem. Eng. Data J. Chem. Educ. J. Chem. Inf. Model. J. Chem. Theory Comput. J. Med. Chem. J. Nat. Prod. J PROTEOME RES J. Am. Chem. Soc. LANGMUIR MACROMOLECULES Mol. Pharmaceutics Nano Lett. Org. Lett. ORG PROCESS RES DEV ORGANOMETALLICS J. Org. Chem. J. Phys. Chem. J. Phys. Chem. A J. Phys. Chem. B J. Phys. Chem. C J. Phys. Chem. Lett. Analyst Anal. Methods Biomater. Sci. Catal. Sci. Technol. Chem. Commun. Chem. Soc. Rev. CHEM EDUC RES PRACT CRYSTENGCOMM Dalton Trans. Energy Environ. Sci. ENVIRON SCI-NANO ENVIRON SCI-PROC IMP ENVIRON SCI-WAT RES Faraday Discuss. Food Funct. Green Chem. Inorg. Chem. Front. Integr. Biol. J. Anal. At. Spectrom. J. Mater. Chem. A J. Mater. Chem. B J. Mater. Chem. C Lab Chip Mater. Chem. Front. Mater. Horiz. MEDCHEMCOMM Metallomics Mol. Biosyst. Mol. Syst. Des. Eng. Nanoscale Nanoscale Horiz. Nat. Prod. Rep. New J. Chem. Org. Biomol. Chem. Org. Chem. Front. PHOTOCH PHOTOBIO SCI PCCP Polym. Chem.
×
引用
GB/T 7714-2015
复制
MLA
复制
APA
复制
导出至
BibTeX EndNote RefMan NoteFirst NoteExpress
×
0
微信
客服QQ
Book学术公众号 扫码关注我们
反馈
×
意见反馈
请填写您的意见或建议
请填写您的手机或邮箱
×
提示
您的信息不完整,为了账户安全,请先补充。
现在去补充
×
提示
您因"违规操作"
具体请查看互助需知
我知道了
×
提示
现在去查看 取消
×
提示
确定
Book学术官方微信
Book学术文献互助
Book学术文献互助群
群 号:481959085
Book学术
文献互助 智能选刊 最新文献 互助须知 联系我们:info@booksci.cn
Book学术提供免费学术资源搜索服务,方便国内外学者检索中英文文献。致力于提供最便捷和优质的服务体验。
Copyright © 2023 Book学术 All rights reserved.
ghs 京公网安备 11010802042870号 京ICP备2023020795号-1