会计信息质量在投资中的决策作用对私人信息和监测的影响.doc 19页

会计信息质量在投资中的决策作用对私人信息和监测的影响 安妮比蒂，美国俄亥俄州立大学 瓦特史考特廖，多伦多大学约瑟夫韦伯，美国麻省理工学院 简介 管理者与外部资本的供应商信息是不对称的在这种情况下企业是如何影响金融资本的投资的呢？越来越多的证据表明，会计质量越好，越可以减少信息的不对称和对融资成本的约束。与此相一致的可能性是，减少了具有更高敏感性的会计质量的公司的投资对内部产生的现金流量。威尔第和希拉里发现，对企业投资和与投资相关的会计质量容易不足，是容易引发过度投资的原因。 当投资效率低下时，会计的质量重要性可以减轻外部资本的影响，供应商有可能获得私人信息或可直接监测管理人员。通过访问个人信息与控制管理行为，外部资本的供应商可以直接影响企业的投资，降低了会计质量的重要性。符合这个想法的还有比德尔和希拉里的比较会计对不同国家的投资质量效益的影响。他们发现，会计品质的影响在于美国投资效益，而不是在日本。他们认为，一个可能的解释是不同的是债务和股权的美国版本的资本结构混合了SUS的日本企业。 我们研究如何通过会计质量灵敏度的重要性来延长不同资金来源对企业的投资现金流量的不同影响。直接测试如何影响不同的融资来源会计，通过最近获得了债务融资的公司来投资敏感性现金流的质量的效果，债务融资的比较说明了对那些不能够通过他们的能力获得融资的没有影响。为了缓解这一问题，我们限制我们的样本公司有所有最近获得的债务融资和利用访问的差异信息和监测通过公共私人债务获得连续贷款的建议。我们承认，投资内部现金流敏感性可能较低获得债务融资的可能性。然而，这种可能性偏见拒绝了我们的假设。 具体来说，我们确定的数据样本证券公司有1163个采样公司（议会），通过发行资本公共债务或银团债务。我们限制我们的样本公司最近获得的债务融资持有该公司不断融资与借款。然而，在样本最近获得的债务融资的公司，也有可能是信号，在资本提供进入私人信息差异和约束他们放在管理中的行为。相关理论意味着减少公共债务持有人获取私人信息，因而减少借款有效的监测。在这些参数的基础上，我们预测，会计质量应该有一个对企业的投资现金流比与银行债务的公共债务公司影响较大，。第二，我们预计，对公司的会计质量的投资波利效果，资本投资者入境计划，将取决于是否贷款合同限制投资。我们预测，当企业面临投资合同限制，过度投资问题是部分需要解决的，会计质量变得不那么值钱了，不太可能影响到投资现金流灵敏度。提供关于这些假设的证据，我们估计的投资模型内部现金流分别进行公司抽样调查，只有公共债务问题，有问题的公司，银团贷款的样本。在这些回归中，我们与我们的互动措施的现金流量变量会计质量调查是否影响会计质量的投资现金流敏感性。对于银行的样本，我们进一步互动的现金流和会计质量变量的乘积投资限制的变量，以研究是否资本开支公约影响投资现金流敏感性。我们同时采用了普通最小二乘法（OLS）和内源性转换模型的估计，以解决潜在的样本选择偏差与银行的债务从公众产生融资选择。我们第一阶段的决定因素模型，融资与私人债务市场遵循Bharath。我们估计第二阶段的投资模型回归政权分开对企业具有公共与私人债务。这些制度的回归也在资本开支内生性控制公约的选择使用从决定因素第一阶段回归模型拟合值包括在债务合同的投资限制。 与以前的研究结果一致，我们发现，投资是敏感的内部现金流量和投资现金流敏感性都会较高地降低公司会计质量。这些结果说明了双方的举行与内生转换模型回归。我们还发现，两个估计技术，减少投资限制，投资现金流敏感性，降低了会计质量的重要性。在这总体样本的分析使用中，我们没有发现二者之间的关系类型的贷款人提供资金和会计质量影响投资现金流的敏感性。 在一系列的敏感性分析中，我们也考虑了可能性，投资现金流敏感性获取信息不仅仅是financ-荷兰国际集团的制约，但关于公司的投资机会也是为了解决这个问题的关注，我们研究的资金来源与投资现金的影响流敏感性的财政约束与无约束的企业公司不同。 要确定企业财务约束，我们使用的事前定义金融约束的基础上，吴怀特德2004指数。我们预期该公司的财务约束投资，谁拥有更大的信息不对称问题，谁将更加依赖于内部产生的资金。同样，投资现金流敏感性为财政浓度，紧张的企业将取决于会计信息的品质，私人信息的存在，以及合同投入使用限制。根据这些推测，我们发现，对于经济约束事务所、会计师质量是同样重要的投资在减少，在私人信息存在现金流敏感性。这些结果支持这一假说，即私人信息和会计质量作为替代品。 在第二个敏感性分析，我们采用资产负债表的流动性方法，去解决发展中捕获的我们关注的机会现金流措施影响现金流和协会之间的投资目的。阿尔梅达Campello和魏兹巴赫（2004）认为更换与现金持有量的变化会减少投资的进而影响投资机会集合，如现金余额变动应与融资约束的情况下的投资机会相一致。与以往的结果是相同的，我们发现，会计质量降低现金流量的敏感性，但贷款人获得私人信息减少了这种影响。 我们的研究结果证实了在麻省理工学院会计质量的重要性，信息不对称对投资的负面影响，现金流的敏感度。我们的研究结果也与比德尔和希拉里的一贯2006年会计质量参数一致，应发挥其作用时，以较低的CAPI-塔尔供应商的选择信息解决问题的缓解机制。然而，他们没有通过与会计质量发挥的作用不一致来改善投资在债券市场的效率。 我们的论文也拓展了我们对帐户重要性的认识，其手段是识别环境信息，其中该公司的质量会计信息是可能重要的。当信息-尝试问题可能是最大时，会计信息质量更显得尤为重要。然而，在这些设置中，如果供应商施加外部资本合同投资限制，或有机会获得私人信息，那么会计质量并不那么重要。此外，我们发现，会计质量降低了资产负债表中的现金流动性的敏感性流，而银行债务表明在面对重要会计质量下降时，我们的投资现金流敏感性结果不只是由于现金流量捕捉企业的投资机会集构成的。 第二节是我们的研究背景。第3节中我们讨论我们的会计科技发展。在第4节我们描述了我们的样本。第5节为我们的研究设计。第六节为研究结论。 2 背景 最近的几项研究论文质量的会计事务所效应，投资采用了多种方法。布须曼人，Piotroski，史密斯（2005）公司的重点倾向，及时退出洛杉矶资本荷兰国际集团的项目。当投资机会减少时，他们调查是否与企业所在的国家会计确认的更及时的会计制度的特点经济损失更迅速地减少资本投资。他们争辩说，他们的研究结果支持这一假说而且这种影响是在国家加强同更加分散的所有权。威尔第（2006）不仅关注公司是否在美国过度投资，而且还与他们是否在ING投资不足项目的积极现在的价值工程。他的调查结果，即减轻了会计质量更高过度投资的问题，而影响对企业加强与存款保险计划，与persed、布须曼（2005）等人所有权相似。。不过，他指出，不能断定是由于较低的投资不足，在减少信息不对称会计质量公司和投资者之间的。 比德尔和希拉里（2006年）审查会计质量如何影响公司的投资现金流敏感性。他们发现，更高的质量是会计伴随着较低的投资现金流在美国的敏感性，但不是在日本。他们认为，在这两个不同的结果国家驱动的事实，更在美国资本通过公平交易提供了投资者不获得私人信息的渠道。他们不直接测试此解释。此外，由于他们承认，但目前尚不清楚这私人债务市场属性驱动的结果。一旦资金供应的失去监控管理，他们的测试不区分贷款人的私人信息获取能力与他们的能力。因此，现行的研究审议有关投资的会计质量的影响浓度，会计质量提高投资效率，但主要改进是在公平的企业所有权是可能的资金来源。 鉴于与Bharath结果等。 2008年，弗朗西斯、拉方德、奥尔森和席佩尔2002年，维滕贝格- Moerman2008年，企业相对高等会计质量奖励了在债务成本减少的情况，这种资本成本较低而不会导致提高投资效率。这些文件的基本论点是提高会计质量允许贷款人以降低成本信息不对称，这表明会计质量应是增加对企业债务融资的依赖投资效率。在本文中，我们试图调和的情况下在这些结果中寻找美国债券市场，会计质量很可能是最重要。我们也试图区分贷款人的影响，会计的作用与质量监测信息起着降低融资约束，提高投资效率的作用。 相关研究认为，描述为一个连续获得美国债务市场私人信息和监测。看看其中的一端是公开的债务持有人谁没有直接进入私人的信息，开展非常小的监测。在另一端是私人银行债务，这是贷款人谁获得更多的私人信息和含有大量的贷款契约。有很多理论银行erature，侧重于对银行的能力，以减少信息不对称。在詹姆斯1987年，这些理论模型提供了实证支持的结果银行提供便宜的通知，而不是资金昂贵的提供的资金的公共债务。同样，Krishnaswami和苏巴马廉（1999）发现具有更大程度，企业信息不对称依赖银行债务。 先前的研究也记录了该公约使用不同的市民与银行债务。1995年至1997间样品使用公约和2003年公约的数据可在固定收益证券数据库查到。他们发现，只有不到百分之五，这些问题包括资本开支限制，而史密斯和苏菲（2009）文档在银团贷款这些公约的发病率要高得多。三他们发现在32个样品3720个数据处理与公约的资本开支限制。 我们利用在银行获取私人信息的RelA差值tive公共债务持有人的私人信息，探讨如何影响在减少投资现金流量的作用，会计质量起着敏感性。此外，资本开支公约的患病率银行债务的合同，使我们进一步区分的重要性银行的私人信息，并能够监测借款人减轻信息问题。 3 发展 在信息问题的情况下，投资决策不应该依赖内部现金流。然而，存在管理者与外部资本的供应商的信息不对称，企业可能面临融资约束使他们依靠内部产生的资金金融投资（Fazzari，哈伯德和Peterson1988年，2000年）。企业对内部现金流入的依赖程度将增加管理者和资本提供者对信息的不对称性。 例如，弗朗西斯，施舟和Vincent（2002）发现了积极凯联与凯森之间的市场反应分析的报告及季度公司盈利英格斯公告。同样， LO（2006）认为，私人和公共信息来源可以相互补充，如果公共信息包含变得更加复杂的数据信息后可以补充说明和解释。鉴于冲突的预报，从理论与实证研究系统来看，我们期望在确定投资现金流敏感性的情况下，获得私人信息和会计质量可以成为无论是作为替代品或作为补充的依据。我们的第一个假设是： 假设1：如果公共和私人信息是替代品，并如果加强这两个信息来源的补充的话，对于投资现金流影响会计质量灵敏度为减轻企业签发银行债务相对于那些issu-荷兰公共债务。此外，银行往往加强其债务协议，以限制管理者的机会投资的明确条款。这些公约提供了一种替代机制，以减轻银行与信息不对称的相关问题。前提是资本开支科夫- nants提供一个替代解决信息问题的机制的基础上，我们预计，这些公约的存在会缓解会计质量在减少投资，现金流敏感性的重要性。相反，如果这些预测将举行公约和会计质量是互补关系而不是替代品，我们的第二个假设是： 朗读 Cǐwài, yínháng wǎngwǎng zài qí zhàiwù xiéyì, yǐ xiànzhì guǎnlǐ zhě de jīhuì tóuzī de míngquè tiáokuǎn. Zhèxiē gōngyuē tígōng le yī zhǒng tìdài jīzhì, yǐ jiǎnqīng yínháng yǔ Infor - mation bù duìchèn xiāngguān de wèntí. Qiántí shì zīběn kāizhī kē fu- nants tígōng yīgè tìdài jiějué xìnxī wèntí de jīzhì de jīchǔ shàng, wǒmen yùjì, kuàijì zhìliàng zài jiǎnshǎo tóuzī, biāo xiànjīn liú mǐngǎn xìng de zhòngyào xìng, yào zài zhèxiē gōngyuē de cúnzài huǎnjiě. (. Xiāngfǎn, rúguǒ zhèxiē yùcè jiāng jǔxíng gōngyuē hé kuàijì zīgé sài xìng shì hùbǔ ér bùshì tìdài pǐn), wǒmen de dì èr gè jiǎshè shì: 字典 假设2：对投资的影响更高质量的会计，现金流敏感性是缓解了资本昂贵的存在。 4 样品特征与样本选择 我们测试我们的假设使用的是公司的抽样调查数据库中列出的议会已发出的公共债务或将进入一个银团贷款。包括公司年中，这些公共或银行债务未偿还。为了对付在银行贷款投资契约选择偏倚，??我们要求该南区区议会数据库对贷款金额，贷款类型信息，以及是否包含银行贷款投资做出解释。然后，我们融入COMPUSTAT这些数据来获取数据库对公司财务会计信息的输入或到贷款债券。该数据限制导致了最后的样本公司1163个，民政事务总署的数据进行必要的测试提出我们的假设。 在我们最后的样本公司，我们注意到，发行公债，发行的银行债务的意见（银团贷款），和印发的意见。在最终银行债务样本，意见均与使用该契诺限制投资占样本百分之33%银行债务。这些百分比相似。 代理会计质量 给定一个共识的最佳方式缺乏衡量会计资格，我们计算出四种不同的会计质量的措施，并结合他们作出一个综合衡量。我们所有的会计质量多边环境协定，确定在一个坚实的估计基础上根据具体使用的10年数据，包括投资的一年。我们估计至少需要回归八十年的观测。朗读 我们的盈余质量AQ3，第三项措施是衡量的五分之一调整后的R排名从我们的收益角度出发，是持久性模型平方，旨在衡量收入的可预见性。我们的第四项措施，AQ4，Tures的现金流量可预测性。据测定，为五分排名调整后的R现金流量电流回归方形租金收入。然后，我们结合通过采取这些措施，平均四个对我们的四个独立的每项措施的行列，共创compos-伊特措施。 符合现有的研究，我们衡量企业的程度投资活动，作为企业的总资本支出。在一个不稳定敏感性分析中，我们的研究和开发投资的测量还包括对我们的支出，发现如果企业的投资很敏感他们的内部现金流量，那么我们期望有一个积极的CFO变量系数。高等会计质量应减少信息的问题，导致投资现金流敏感性，因此我们预计首席财务官卡迪尔系数为负。我们还包括其他因素的控制变量是可能影响公司的投资选择的。与以前研究相比，我们控制的地方，包括对公司的增长选择TobinsQ，我们期望得到积极的与投资相关的信息。此外，我们控制的大小，与杠杆信用评级。规模较小的公司，信贷评级较低，杠杆公司预计将有较少的投资。最后，我们还包括了有形控制公司的资产，有形性与公司绩效，净资产收益率，但我们不对这些变量对投资的影响作明确的预测。 我们的分析是在企业进行一年的水平年，采用全公司之间的债务发行和到期日的数据。因此，每个一年后，该公司已进入贷款时期，我们衡量投资和相应的自变量，直至贷款到期。对于那些贷款，在我们的样本期末发生，我们人为地设置“因为债务的财务数据2005年以后至2005年到期”不能利用。我们有坚定的固定效应和固定效应的回归年。 为了检验假设1，我们就比较首席财务官的系数，在市民与银行债务的样本中进行。我们推测，该系数对首席财务官*卡迪尔应在较低的银行债务样品中。如果私人会计信息作为替代品，应该更高，作为这些信息来源的补充。为了检验假设2，我们包括对契约的投资限制，使用措施Inv_Cov回归。一个投资契约限制的存在预计将减少信息不对称的重要性，并且前预计将减少会计质量较高的优势，从而首席财务官系数Inv_Cov预期是积极的。 签发银行债务与公共债务可能导致选择在OLS回归自我选择的偏差，我们估计内源性开关模式的选择问题控件相关联的债务financ-ING集团源。同样，使用债务契约限制投资的选择，也可能是内生决定的。为了解决这一问题，我们估计这项决定的因素第一阶段的模型，包括在债务合同的投资限制，包括拟合值在我们的内生转换模型。 内生转换模型与银行债务的公共选择来衡量之间的关系，私人信息影响会计质量和投资现金流的敏感性，我们进行了比较，公众的公司与银行债务。子样本的公共债务包括公司年中无银团贷款目前或预期发行。子样本的银行债务是指企业银团贷款至少一年未偿还。 为了控制该银行的内生性和公共债务的选择，我们遵循Bharath等。 2008年，估计1978年提出的内源性模型。内生性转换模型有一个第一阶段的选择，选择方程和银行债务与公共两个，第二阶段制度方程建模投资的决定因素（子样本的银行债务和公共债务的子样本之一）。西米-摩尔与Bharath等（ 2008年）这一方法使得我们可以控制自我债务的选择和在两个政权比较会计质量的对首席财务官的影响。 第一选择的债务阶段模式，包括从丹尼斯变量和米霍夫2003年，Krishnaswami等。 1999年Bharath等（ 2008年）具体来说，我们认为，选择对公司债务的会计质量（AQ）的依赖，其中质量较好的企业更倾向于选择公共债务。该选择也取决于公司的盈利能力，在更加有利可图的企业将访问的公共债务。基于这个论点，我们包括首席财务官回归。这个选择也可能取决于大小、生长选项、信贷质量并充分利用，所以我们包括我们的规模，TobinsQ，信用评级，与杠杆变量。最后，选择也将依赖于该公司的有形资产，所以我们包括有形的回归。 使用公约模型的投资限制是我们第一阶段的投资决策模型，包括限制决策。 2009年，从我们的模型市民与银行债务的选择。我们还包括在我们的模型中，以便会计质量的可能性和投资契约规定解决债务的代理问题的替代方法。与研究人员的研究结果一致。 2009年，该公约较少使用preva-对企业的较大影响，具有较高的信用评级转借，我们包括大小，信用等级，在我们的模型和评分。他们还发现混合的证据，该公约的用途是相对于现金流量企业普遍具有较高的市场债务。我们包括TobinsQ，杠杆和财务总监在我们的模型来捕捉这些结构。 我们也允许决定包括在投资限制公司的债务合同，取决于企业的盈利能力，有形性，以及股息支付。因此，我们包括净资产收益率，有形性，并在我们的模型中体现。最后，我们允许投资的决定取决于公约对合同作为工具变量的特点，如大小和贷款。具体来说，我们包括债务金额，calcu-lated作为银行债务金额的自然对数，其中是一个指标变量编码为1，如果该贷款周转，否则0。因为该决定将在债务合同成立时决定债务投资限制合同，所有变量测量前向银行发行债券的。 我们建构了一个预测Inv_Cov等于1，如果变量的预测使用Inv_Cov概率估计，从下面的一阶段Probit模型比百分之50%更大，否则为0。 融资约束指标其实是灵敏度捕捉缺乏获得外部融资（Fazzari等，虽然有相当大的争论是否投资现金流。 2000年Kaplan和津加莱斯（2000年），有一些共识，这测量不同的行为约束和财政之间的不整合公司。怀特德和吴（2004年）规定，财政约束作为企业，如果他们有较高的折扣率，这很可能导致这些公司依靠内部现金流。如果存在融资限制建议信息不对称问题的重要性的增加，那么我们所期望的会计质量，获取私人信息，投资限制，有一个投资现金流量较大的影响力灵敏度作为财政约束的企业。 对资金来源的决定因素为重点，我们发现公司更可能使用银行债务时，规模较小，有降低信贷质量，有形资产少，且没有支付股利。更多重要的是，双方（评级）工具变量不能负荷。为了测试是否这两个变量的有效变数，我们在一个总投资中验证这两个变量模型。结果似乎表明，这两个工具是有效的标识符，即这两个变量没有显著的相关性与投资。 类似的审查决定因素融资方面，我们还发现，盈利能力（如首席财务官衡量），规模，信用质量，有形资产和分红的位置都与该决定将在该公司的债权投资契约限制合同有关。我们还发现，我们的外生变量，一个是相关的决定，其中循环贷款是不太可能有投资限制的。由于循环贷款可以提取，而且偿还了该合同有效期内多次，这并不奇怪，他们不太可能有投资的限制。 从OLS的结果与内源性的投资估算开关模型报告我们对我们的投资模型分析的结果。该报告前两列从我们的OLS回归，结果占了内生性的融资或投资契约选择。第二个报告，我们的两列是内源性的结果转换模型。 结果从现金现金流量模型中我们报告估计我们的资产负债表的流动性模型的结果得到。结果看起来非常相似，从投资模式约束企业来看。具体来说，我们发现，现金之间的关联流量和现金余额为公司更高的公共债务，这个问题对企业使用私人银行贷款有重要影响。研究结果还表明，与现金流量及现金结余减少的关联帐户，荷兰国际集团对企业质量问题的公共债务（而不是企业发卡银行债务）。从内生性转换模型的结果表明，会计质量的影响是私人银行对企业贷款比公共债务使用低对企业发行的。朗读5 结论 以前的研究表明，会计质量灵敏度较高会降低公司对内部现金流投资的影响。比德尔和希拉里（2006）发现这种关系在美国存在，但不是在日本，他们得出结论，会计质量在发挥作用时，以较低资本为目的，有备用的信息来源。 我们的文件扩展了这一研究，在降低融资约束的情况下，研究范围内的债务因素增强或减弱市场对会计信息的重要性的影响。我们研究了美国样本公司最近提出的债务融资和调查私人的数据和监测敏感性的投资内部现金流动性。我们发现，提高会计质量降低投资现金对企业流动敏感性这一问题是公共债务和银行债务共同存在的问题。我们还发现，投资限制，消除了对会计的质量、投资现金流敏感性的影响。此外，我们还发现，对于高的财政限制，贷款人的私人信息也对减少投资敏感性、缓解现金的会计素质起着重要作用。 我们发现，我们的结果持有我们允许的可能性，现金流动不仅搜集了企业内部产生的依赖资金，也可能捕获有关公司的投资机会信息联系。具体来说，我们发现，对于企业面临的融资约束，其中信息不对称问题可能是最大的，占会计质量最重要的部分。此外，我们发现，我们的结果验证了我们研究资产负债表流动性现金流的敏感度。 我们的论文扩展了金融经济学文献讨论财政拮据的关系和投资决策问题。我们发现资金来源是影响投资决策的主要原因。我们的文件进一步区分了信息和监测银行发挥的降低融资约束作用。这些结果是很有趣的，因为它们帮助我们了解了不同形式的债务融资是如何影响投资决策与会计信息间的相互作用关系的。 附录B The Effect of Private Information and Monitoring on he Role of Accounting Quality in Investment ANNE BEATTY, The Ohio State University W. SCOTT LIAO, University of Toronto JOSEPH WEBER, Massachusetts Institute of Technology 1 Introduction Information asymmetry between managers and outside capital suppliers can affect how ?rms ?nance capital investments. A growing body of evidence indicates that better accounting quality can reduce information asymmetry costs and reduce ?nancing constraints. Consistent with this possibility, Bid- dle and Hilary (2006) document that higher accounting quality reduces the sensitivity of ?rms’ investment to their internally generated cash ?ows. Verdi (2006) and Biddle, Hillary, and Verdi (2009) ?nd that accounting quality is positively correlated with investment for ?rms prone to under- invest and is negatively correlated with investment for ?rms prone tooverinvest. The importance of accounting quality on investment inef?ciency may be mitigated when outside capital suppliers have private information or can directly monitor managers. By accessing private information and controlling managerial actions, outside capital suppliers can directly affect a ?rm’s investments, reducing the importance of accounting quality. Consistent with this idea, Biddle and Hilary (2006) compare the in?uence of accounting quality on investment ef?ciency across countries. They ?nd that accounting quality in?uences investment ef?ciency in the United States, but not in Japan. They suggest that one potential explanation for this cross-country difference is the mix of debt and equity in the capital structures of U.S. ver-sus Japanese ?rms. We extend this research by examining how different sources of ?nancing affect the importance of accounting quality on ?rms’ investment–cash ?ow sensitivity. Directly testing how different sources of ?nancing in?uence the effect of accounting quality on the investment–cash ?ow sensitivity is chal-lenging because a comparison of ?rms that recently obtained debt ?nancing to those that did not could be affected by their ability to obtain ?nancing.To alleviate this problem, we restrict our sample to ?rms that have all recently obtained debt ?nancing and exploit the differences in access to pri-vate information and monitoring that exist across the public debt to private lending continuum as suggested by Diamond 1991.1 We acknowledge that the sensitivity of investment to internal cash ?ows may be lower immedi-ately after obtaining debt ?nancing. However, this possibility would bias against rejecting our hypotheses. Speci?cally, we identify a sample of 1,163 ?rms on the Securities Data-corp (SDC) database that have recently raised capital through the issuance of either public debt or syndicated bank debt. We restrict our sample to ?rms that have recently obtained debt ?nancing to hold constant the ?rm characteristics associated with borrowing. However, within the sample of ?rms that have recently obtained debt ?nancing, there are likely to be sig-ni?cant differences in the capital provider’s access to private information and the constraints they place on managerial actions. Diamond’s (1991) theory implies that public debt holders have less access to private information and are thus less effective in monitoring bor-rowers than banks. Based on these arguments, we predict that accounting quality should have a larger in?uence on ?rms’ investment–cash ?ow sensi-tivity for ?rms with public debt than for those with bank debt. Second, we expect that the effect of accounting quality on ?rms’ investment poli-cies will depend on whether lenders contractually restrict investment. We predict that, when ?rms face contractual restrictions on investments, over-investment problems are partially addressed, and accounting quality becomes less valuable and less likely to affect the investment–cash ?ow sensitivity. To provide evidence on these hypotheses, we estimate a model of invest-ments on internal cash ?ows separately for the sample of ?rms that only issue public debt and for the sample of ?rms that issue syndicated loans. In these regressions, we interact the cash ?ow variable with our measure of accounting quality to investigate whether accounting quality in?uences investment–cash ?ow sensitivities. For the bank sample, we further interact the product of the cash ?ow and accounting quality variables with the investment restrictions variables to examine whether capital expenditure covenants affect the investment–cash ?ow sensitivity. We use both ordinary least squares (OLS) and endogenous switching model estimation to address potential sample selection bias arising from the public versus bank debt ?nancing choice. Our ?rst-stage model of the determinants to enter the pub-lic versus private debt market follows Bharath, Sunder, and Sunder 2008.We estimate second-stage ‘‘regime’’ investment model regressions separately for ?rms that have public versus private debt. These regime regressions also control for endogeneity in the capital expenditure covenant choice using the ?tted values from a ?rst-stage regression model of the determinants of including an investment restriction in the debt contract. Consistent with prior research, we ?nd that investment is sensitive to internal cash ?ows and the investment–cash ?ow sensitivity is lower for ?rms with higher accounting quality. These results hold in both our OLS and endogenous switching model regressions. We also ?nd that, for both estimation techniques, investment restrictions reduce the investment–cash ?ow sensitivity and reduce the importance of accounting quality. In this analysis using the overall sample, we do not ?nd a relationship between the type of lender providing capital and the effect of accounting quality on the investment–cash ?ow sensitivity. In a series of sensitivity analyses, we also consider the possibility that the investment–cash ?ow sensitivity captures information not only about ?nanc-ing constraints but also about ?rms’ investment opportunities. To address this concern, we examine the effect of sources of ?nancing on investment–cash ?ow sensitivities for ?nancially constrained ?rms versus unconstrained ?rms. To identify ?nancially constrained ?rms we use an ex ante de?nition of ?nancial constraint based on the Whited and Wu 2004 index. We expect that the investments of ?nancially constrained ?rms, who have greater information asymmetry problems, will depend more on internally generated funds. Similarly, investment–cash ?ow sensitivities for ?nancially con-strained ?rms will depend more on the quality of accounting information, the existence of private information, and the use of contractual investment restrictions. Consistent with these conjectures, we ?nd that, for ?nancially constrained ?rms, accounting quality is less important in reducing invest-ment–cash ?ow sensitivities in the presence of private information. These results support the hypothesis that private information and accounting qual-ity serve as substitutes. This result is also consistent with the notion that access to private information is more important when information asymme-try problems are likely to be high.. In a second sensitivity analysis, we adopt a balance sheet liquidity approach to address the concern that growth opportunities captured in our cash ?ow measures affect the association between cash ?ows and invest-ments. Almeida, Campello, and Weisbach (2004) suggest that replacing investment with changes in cash holdings will reduce the in?uence of the investment opportunity set, as changes in cash balances should be uncorre-lated with investment opportunities in the absence of ?nancing constraints. Consistent with previous results, we ?nd that accounting quality reduces the cash–cash ?ow sensitivity, but that lenders’ access to private information decreases this effect. Jointly, our results con?rm the importance of accounting quality in mit-igating the negative effects of information asymmetry on the investment–cash ?ow sensitivity. Our results are also consistent with Biddle and Hilary’s 2006 argument that accounting quality should play a lower role when capi-tal suppliers have alternative information-problem-mitigating mechanisms. However, they are inconsistent with accounting quality playing no role in improving investment ef?ciency in the debt market. Our paper also expands our understanding of the importance of account-ing information by identifying the contexts in which the quality of the ?rm’s accounting information is likely to be important. When information asymme-try problems are likely to be the largest, accounting quality is more impor-tant. However, in these settings, if outside capital suppliers impose contractual investment restrictions, or have access to private information, then accounting quality is less important. Furthermore, our ?nding that accounting quality reduces the sensitivity of balance sheet liquidity to cash ?ows and that the importance of accounting quality decreases in the face of bank debt suggests that our investment–cash ?ow sensitivity results are not merely due to cash ?ows capturing ?rms’ investment opportunity sets. Section 2 provides background for our study. We discuss our hypothe-sis development in section 3. We describe our sample in section 4 and our research design in section 5. We discuss our empirical results in section 6 and conclude in section 7. 2 Background Several recent papers examine the effect of accounting quality on ?rms’investments using a variety of approaches. Bushman, Piotroski, and Smith (2005) focus on ?rms’ propensities to promptly withdraw capital from los-ing projects. They investigate whether ?rms located in countries with accounting regimes characterized by more timely accounting recognition of economic losses reduce capital investments more quickly when investment opportunities decline. They argue that their results support this hypothesis and that this effect is stronger in countries with more diffuse ownership. Verdi (2006) is concerned not only with whether ?rms overinvest in los-ing projects, but also with whether they underinvest in positive net-present-value projects. His ?ndings that higher accounting quality mitigates the overinvestment problem and that the effect is greater for ?rms with dis-persed ownership are similar to those in Bushman et al. 2005. However, he states that he ‘‘cannot conclude that accounting quality is associated with lower underinvestment due to the reduction in information asymmetry between the ?rm and investors’’. Biddle and Hilary (2006) examine how accounting quality affects ?rms’investment–cash ?ow sensitivity. They ?nd that higher accounting quality is associated with lower investment–cash ?ow sensitivity in the United States, but not in Japan. They argue that the difference in results across these two countries is driven by the fact that more capital in the United States is provided through arm’s-length transactions with investors who do not have access to private information channels. They do not directly test this interpretation. Furthermore, as they acknowledge, it is not clear which attributes of the private debt markets drive their results. Their tests do not distinguish between lenders’ ability to obtain private information versus their ability to monitor managers once capital is supplied. Thus, the existing research examining the effects of accounting quality on investments con-cludes that accounting quality improves investment ef?ciency, but the improvements are predominantly in ?rms with diffuse ownership where equity is likely to be the source of capital. Given the results in Bharath et al. 2008, Francis, LaFond, Olsson, and Schipper 2002, and Wittenberg-Moerman 2008 that ?rms with relatively higher accounting quality are rewarded with a reduction in the cost of debt, it is somewhat surprising that this lower cost of capital would not lead to improved investment ef?ciency. The central argument underlying these papers is that improved accounting quality allows lenders to reduce the costs associ-ated with information asymmetry, suggesting that accounting quality should increase investment ef?ciency even for ?rms that rely on debt ?nancing. In this paper we attempt to reconcile these results by looking at the contexts in the U.S. debt markets in which accounting quality is likely to be the most important. We also attempt to distinguish between the effects of lenders’ pri-vate information versus monitoring on the role accounting quality plays in reducing ?nancing constraints and improving investment ef?ciency. Features of U.S. debt markets Diamond (1991) describes the U.S. debt market as a continuum of access to private information and monitoring. At one end of the spectrum are public debt holders who do not have direct access to private information and engage in very little monitoring. At the other end is private bank debt, which is char-acterized by lenders who have greater access to private information and by loans that contain numerous covenants. There is much theoretical banking lit-erature that focuses on banks’ ability to reduce information asymmetries. Empirical support for these theoretical models is provided by the ?ndings in James 1987 that banks provide cheap ‘‘informed’’ funds as opposed to the costly ‘‘uninformed’’ funds provided by public debt. Similarly, Krishnaswami, Spindt, and Subramaniam (1999) ?nd that ?rms with a greater degree of information asymmetry rely more on bank debt. Previous research has also documented that covenant use differs for public versus bank debt. For example, Billet, King, and Mauer (2007) docu-ment covenant use for a sample of 7,567 public bonds issued between 1995 and 2003 with covenant data available on the Fixed Income Securities Database. They ?nd that less than 5 percent of these issues include capital expenditure restrictions, while Nini, Smith, and Su? (2009) document that the incidence of these covenants in syndicated bank loans is much higher. They ?nd capital expenditure restrictions in 32 percent of their sample of 3,720 deals with covenant data. We exploit the difference in banks’ access to private information rela-tive to public debt holders to explore how private information affects the role that accounting quality plays in reducing investment–cash ?ow sensitivities. In addition, the prevalence of capital expenditure covenants in bank debt contracts allows us to further distinguish between the importance of banks’ private information and their ability to monitor the borrower in mitigating information problems. 3 Hypotheses development In the absence of information problems, investment decisions should not depend on internal cash ?ows. However, in the presence of information asymmetry between managers and outside capital suppliers, ?rms may face ?nancing constraints causing them to rely on internally generated funds to ?nance their investments (Fazzari, Hubbard, and Peterson 1988, 2000). Firms’ dependence on internal cash ?ows will increase with the extent of information asymmetry between managers and capital providers. Capital providers with greater access to private information (like banks) are likely to have superior information about ?rms compared to lenders that obtain their information from public sources (like public debt hold-ers). Biddle and Hilary (2006) argue that banks’ superior private informa-tion should serve as a substitute for accounting quality in determining the sensitivity of investment to internal cash ?ows. While this prediction is con- sistent with theoretical models such as Holthausen and Verrecchia 1988, recent empirical research suggests that, rather than acting as substitutes, pri-vate information and public information may act as complements. For example, Francis, Schipper, and Vincent (2002) ?nd a positive asso-ciation between the market reaction to analysts’ reports and quarterly earn-ings announcement. Similarly, Chen, Cheng, and Lo (2006) argue that private and public information sources can complement one another if the public information contains complex data that is made more informative after additional explanation and interpretation. Given the con?icting predic-tions from the theoretical versus empirical work, we expect that access to private information and accounting quality could serve either as substitutes or as complements in determining the investment–cash-?ow sensitivity. Our ?rst hypothesis is: Hypothesis 1. The effect of accounting quality on the investment–cash ?ow sensitivity is mitigated for ?rms issuing bank debt relative to those issu-ing public debt if public and private information are substitutes and is enhanced if these two sources of information are complements. In addition, banks often include explicit covenants in their debt agreements to limit managers’ opportunistic investments. These covenants provide banks with an alternative mechanism to mitigate the problems associated with infor-mation asymmetry. Based on the assumption that capital expenditure cove-nants provide a substitute mechanism for addressing information problems, we expect that the importance of accounting quality in reducing the invest-ment–cash ?ow sensitivity to be mitigated in the presence of these covenants. (The opposite prediction would hold if these covenants and accounting qual-ity are complements rather than substitutes.) Our second hypothesis is: Hypothesis 2. The effect of higher quality accounting on the investment–cash ?ow sensitivity is mitigated by the existence of a capital expen-diture covenant. 4 Sample characteristics and Sample selection We test our hypotheses using a sample of ?rms listed on the SDC database that have issued public debt or entered into a syndicated loan. We include ?rm years in which these public or bank debts are outstanding. To deal with the selection bias of investment covenants in bank loans, we require that the SDC database have information on the size of the loan, the type of loan (revolver or term), and whether the bank loan contains an investment cove-nant. We then merge this data with the COMPUSTAT database to obtain ?nancial accounting information for the ?rm that entered into the loan or debenture. This data restriction resulted in a ?nal sample of 1,163 ?rms (4,475 ?rm years) that had all of the data necessary to test our hypotheses. Of the 4,475 ?rm years in our ?nal sample, we noted that 1,083 obser-vations issued public debt, 3,392 observations issued bank debt (syndicated loans), and 1,026 observations issued both. Within the ?nal bank debt sample, 1,124 observations are associated with the use of covenants that restrict investments representing 33 percent of the bank debt sample. These percentages are similar to those reported in Nini et al. 2009. Accounting quality proxy Given a lack of consensus about the best way to measure accounting qual-ity, we calculate four different measures of accounting quality and combine them to make one composite measure.All of our accounting quality mea-sures are estimated on a ?rm-speci?c basis using 10 years of data, including the investment year.When estimating the regression we require at least eight years of observations. Our ?rst accounting quality measure, AQ1, is based on Dechow and Di-chev 2002. This measure captures the extent to which accruals map into cash ?ows. For each ?rm in the sample, we run a time-series regression of total current accruals on cash ?ows from operations measured concurrently and with a one-period lead and one-period lag. The variable AQ1 is mea-sured as the quintile ranking of Adjusted R-squared from these regressions (this measure is consistent with the measure used in Biddle and Hilary 2006). Following Wysocki (2005), our second measure of earnings quality is a measure of earnings persistence. For each sample ?rm, we run a time-series regression of one-period-ahead earnings on current earnings. The variable AQ2 is measured as the quintile ranking of coef?cients on current earnings from these regressions. Our third measure of earnings quality, AQ3, is measured as the quintile ranking of Adjusted R-squared from our earnings persistence model and is designed to measure earnings predictability. Our fourth measure, AQ4, cap-tures cash ?ow predictability. It is measured as the quintile ranking of Adjusted R-squared from a regression of one-year-ahead cash ?ows on cur-rent earnings. We then combine these four measures by taking the average of the ranks of each of our four individual measures to create our compos-ite measure, AQ. Consistent with existing research, we measure the extent of the ?rm’s investing activities, Investment, as the total capital expenditures of