翩翩翩翩酷翩攥翩瞄翩瞄翩翩翩翩翩翩跚跚跚跚鹅艘酣黯跚跚嘟嘟跚跚跚跚翩翩蛐时酷翩蛐蛐艘跚跚跚跚翩翩踹2011年9月第13卷第3期，王址，中国会计与财务研7L Ch na Account ng and F nance Rev ew Volume 13, Number 3 -September 2011 翩翩翩翩蹦蹦翩翩翩翩翩翩幽翩翩翩翩蛐翩翩翩酣翩翩翩翩翩翩跚跚跚跚翩翩翩跚跚翩翩翩翩翩翩翩翩翩跚跚跚公司纵向并购财富效应的实证研究一一来自我国上市公司的经验证据1李青原王永超2摘要本文主要检验我国上市公司纵向并购的财富效应。以1998至2008年间发生、上市公司作为收购方公司的国内非关联股权标的并购样本为研究对象，结果发现并购交易双方间纵向关系越高，公司并购财富效应相应较高，而且产业上下游间公司纵向并购财富效应会随着公司并购交易双方所在地区产权保护程度的降低而增加，但是这种产业上下游间公司纵向并购财富效应只存在于民营上市公司作为收购方的并购活动。这表明，在我国不完善的司法体系、较差的契约实施和低效的社会信任条件下，纵向一体化可以作为一种替代机制，保护私人控股收购公司双方事前的关系专用性投资激励和防止交易对方潜在的"敲竹杠"行为，节约高昂的市场交易成本，及规避过量的监管负担和政治干预。关键词:纵向并购、纵向一体化、纵向关系、财富效应中图分类号:F425、F830l 感谢国家自然科学基金项目(70702017)、教育部哲学社会科学研究重大课题攻关项目(1叼ZDOOI9)和武汉大学人文社会科学"70后"学者学术团队建设计划对本研究项目的资助。感谢2010年中国会计与财务研究国际研讨会中香港中文大学王丛教授、第丸届(2010)中国实证会计国际研讨会中浙江大学宋衍衡教授的有益评论，感谢匿名审稿人和执行编辑俞伟峰教授提出的宝贵修改意见。文中的→切错误与疏漏均由作者负责。2 李青原，副教授，武汉大学经济与管理学院会计系，湖北省武汉市武昌区珞功。山(430072), Email: qyli@。王永超，暨南大学管理学院会计系硕士生。
2 李青原王永超一、号|言纵向并购(verticalmerger) ，又称垂直并购，是指上游(下游)公司通过购买下游(上游)公司的股权获得被收购公司的控制权，以便于在内部而非通过市场或契约进行交换，且是公司纵向一体化(verticalintegration)的一种动态过程。正如Williamson(1975)所说，纵向一体化研究是经济理论的一个重要课题，且必将是为新制度经济学所关注的核心问题，是要解释经济活动在厂商、市场及-些中间组织方式间的资漉配置，且至少最近30年内也是微观经济学理论和实证研究领域一个重要的研究对象Ooskow，2005 ; Lafontaine and Slade, 2007) 0 Coase(1937)在其经典论文《企业的性质》指出两个基本问题:(1)企业边界的决定因素;(2)企业边界会影响资源配置吗?后来，Williamson(1975, 1985)、Kleinet al.(1978)、Grossmanand Hart(1986) 和Hartand Moore(1990)等又不断地创新和拓展，试图解释所有的纵向一体化活动，无论他们解释纵向一体化活动是否完全可行，但是他们的研究不仅从理论上对于增进人们对纵向→体化活动的理解有益，而且为实证检验纵向一体化的动因和经济后果提供了理论基础。近20年来，纵向一体化(企业边界)的实证研究一直受到学者们的关注，其主要关注两个基本且相关的问题:(1)何种类型的交易最好在企业内部发生?(2)纵向一体化对产品价格、产品数量、投资和利润等产生何种经济后果?(Lafontaine and Slade, 2007)。因ein(2005)、Lafontaineand Slade(2007)和Macherand Richman(2008)等综述发现3因数据的可获得性使得实证文献通常以某个公司或行业为样本，公司纵向一体化对产品零售价格、产品数量、系统风险和股票评级等方面的影响，而对纵向一体化与企业经营绩效间因果关系的实证研究寥寥无几，至多也是小样本的案例和特定产业研究，这可能归因于现实中难以确定公司的纵向关系，而这恰又是考察公司纵向一体化经济后果的实证研究中非常重要和关键的一步(Fanand Lang, 2000 ; Kedia et al., 2008) ，同时现有的研究对象都处于发达市场经济的美国和英国，因此我们仍然难以清楚地知道纵向一体化对公司经营绩效影响及其相互间因果关系的普适性结论。正如Lafontaineand Slade(2007)所说3公司并购提供了一个难得的机会通过事件研究法来检验纵向一体化对公司经营绩效影响，并且可以较好地解决纵向→体化程度潜在的内生性。Fanand Goyal(2006)和Kediaet al.(2008)以1962至1996年间美国上市公司并购案例为样本，运用Fanand Goyal(2006)的产业间纵向相关性系数界定收购公司与目标公司双方间的纵向关系，结果发现纵向并购宣告期的累计异常收益率显著为正，但后者又发现1996年后公司纵向并购宣告期的累计异常收益率显著为负。3Scheffman and Higgins(2004)指出与横向并购相比，纵向并购的研究既更复杂，又更受某具体制度安排的影响。尽管冯根福和吴林江(2001)、李善民和朱滔(2006)、周小春和李善民(2008)、巫和雄和张晓明(2009)等运用不同的研究方法和样本对我国上市公司并购绩效进行实证研究，得出了许多有益的结论，但是这些实证研究中公司并购类型的界定较为主观，且真正涉及公司纵向并购绩效的专门研究还较少，值得学者们进一步地修正和完善。近年来随着我国资本市场制度基础设施的快速发展和不断Kedia et al. (2008)认为，1996年后公司纵向并购宣告期的累计异常收益率显著为负可能要归因于这些并购不是由行业内支持纵向一体化的基本因素驱动的。
公司纵向并购财富效应的实证研究完善，加之中石油、中石化和中国铝业等纵向收购上市公司的产业链条协调整合战略并购案例将会不断增加，那么这些公司纵向并购绩效的表现如何呢?影响公司纵向并购绩效的制度因素又会是哪些呢?结合我国特定的制度安排以及我国资本市场的特殊性，本文拟以1998至2008年间发生、上市公司作为收购方的国内非关联股权标的并购样本为研究对象，结果发现并购交易双方间纵向关系越高，公司井购财富效应相应较高，而且产业上下游间公司纵向并购财富效应会随着公司并购交易双方所在地区产权保护程度的降低而增加，但是这种产业上下游闻公司纵向并购财富效应只存在于民营上市公司作为收购方的并购活动。这表明，在我国不完善的法律和司法体系、较差的契约实施和低效的社会信任条件下，纵向→体化可以作为一种替代机制，保护私人控股收购公司双方事前的关系专用性投资激励和防止交易对方潜在的敲竹杠行为，节约高昂的市场交易成本，及规避过量的监管负担和政治干预。本文的贡献主要体现在三方面:首先，从公司并购的动态视角，本文通过事件研究法实证检验了纵向→体化对公司经营绩效的影晌，较好地解决了纵向一体化程度潜在的内生性(Lafonraineand Slade, 2007) ，丰富了交易费用经济学类文献，较好地解释了处于转轨加新兴、且社会信任等制度安排低效甚至缺失的中国资本市场中公司纵向并购财富行为的合理性，而这种合理性可能拥于为了保护和促进公司并购交易双方事前的关系专用性投资激励和防止交易对方潜在的敲竹杠行为，节约高昂的市场交易成本，规避过量的监管负担和政治干预。其次，由于公司并购动因的复杂性、多元性和多边性，现实中很难运用一个理论来解释全部绩效的变化，公司并购绩效往往是多种因素相互作用的结果(Bru旧r，2002) ，因此本文从交易双方间纵向关系为解读公司并购绩效之谜提供了新的视角。周小春和李善民(2008)等研究发现员工抵制程度、资源整合程度、支付方式、收购比例等是影响我国上市公司并购绩效的主要因素。本文通过投入产出表中产品流量信息，从纵向关联程度细化了多元化并购的分类，强调了交易双方间纵向关联程度对公司并购绩效的影晌，增进了对收购方并购绩效来源的理解。最后?从产权和制度环境的视角丰富了公司纵向→体化的相关研究。在转型经济国家，产权特'性和制度环境都是理解企业边界的重要因素之一(阻ein，2005 ; Fan et 此，2009 ; Acemoglu et 此，2009)。本文发现在不完善的法律制度、较差的契约实施和低效的社会信任条件下，纵向一体化可以作为→种替代机制，能够较好地保护和促使私人控股的收购公司有价值的关系专用性投资免受机会主义行为的侵害，及规避过量的监管负担和政治干预，这为我国企业边界的调整提供了新的思路。下文结构安排如下·第二部分是理论分析和研究假设;第三部分是研究设计;第四部分实证检验的结果及分析;最后一部分是研究结论。二、理论分析与研究假设自从Coase(l937)发表经典论文《企业的性质》以来'Williamson(l975，1985)、Kleinet al.(l978)、Grossmanand Harr(1986)、Hartand Moore(1990)等不断地在Coase(1937)的基础上进行理论创新和拓展，形成了一种能够有效解释公司边界问题的主流经济学理论。Joskow(2005)指出，解释公司纵向一体化的主流理论主要是Williamson(l975，1985)
李青原王永超4 和Kleinet al.(1978)等发展的交易费用经济学(TCE)与Grossmanand Hart(1986)、Hart扭dMoore(1990)等发展的产权理论(PRT)σWil!iamson(1975，1985)、阳einet al札978)等指出，一旦交易方作出关系专用性投资，契约的不完性就会引出交易主体的机会主义行为(opportunism)，而资产专用性带来的可占用性准租金(appropriab!especialized qu街1-ren ts)使这种机会主义行为由可能转变为现实，产生敲竹杆问题，从而导致事前次优的关系专用性投资和较低的盈利能力。解决敲竹杠问题的一个主要方法便是纵向一体化，这会消除供应商和客户间契约非效率引发未来的敲竹杠行为，及增加双方由于存在关系性专用投资而产生的投资效率(Kleinet al., 1978 ; Joskow, 2005)。总之，交易费用经济学理论认为契约不完全导致的交易费用主要拥于事后的失调，因而强调事后适应性治理，同时认为公司纵向并购主要是解决专用性资产投资不足及规避潜在要挟问题引起的契约非效率，这种非效率使得内部治理结构比市场签约更为有效，那么纵向并购财富效应是这些内部化交易过程中节约的治理成本的函数(Milgromand Roberts, 1992) 0与此同时，Grossman and Hart(1986)、Hartand Moore(1990)等遵循交易费用经济学的传统，从资产专用性和契约不完全性的前提出发而研究纵向一体化的产权理论，把纵向一体化本身视为所有权，运用剩余控制权推导出由于所有权结构的不同而导致双方事前的投资激励扭曲，通过比较不同情况下投资激励的扭曲程度，从激励角度研究纵向一体化的戚本与收益，其中公司纵向并购会因增加收购公司剩余控制权而增加其进行关系专用性投资的激励，但这又会因减少目标公司剩余控制权而降低其进行关系专用性投资的激励，且目标公司事前投资激励的扭曲程度(或激励戚本)随着资产专用性程度的提高而日趋严重(Kval町，2007)。相比于新古典经济学，交易费用经济学范式触发了对公司纵向并购(一体化)更多的研究，这些研究中包括对于某-公司、产业或某种形式合约安排的案例研究及少数基于大样本观察数据的计量分析(Joskow，2005)。Fan and Lang(2000)创造性地通过投入产出表中产品流量构造了以投入产出为基础的方法来反映产业间和部门间的纵向相关性，再以此为基础构造了公司纵向-体化程度，该方法得到了学者们广泛的运用(F皿四dGoyal, 2006 ; Kedia et al., 2008 ; Shenoy, 2009 ; Fan et al., 2009) 0 Fan and Goyal(2006)以1962至1996年间2162个美国上市公司并购样本为研究对象，通过Fanand Goya!(2006)的产业间纵向相关性界定收购公司与目标公司间的纵向关系，将并购类型划分为完全横向并购(purehorizontal mergers)、完全纵向并购(purevertical mergers)、1昆合横向纵向并购(mixedvertical and horizontal mergers) 及棍合并购(conglomeratemergers) ，结果发现纵向并购宣告期的累计异常收益率显著为正。Kediaet al.(2008)通过Fanand Goya!(2006)的并购类型分类方法，以1979至2002年间1692个美国上市公司并购样本为研究对象3结果发现1996年前公司纵向并购宣告期的累计异常收益率显著为正，但之后又显著变为负，即纵向并购是否创造价值随着时间的变化而有所不同。Shenoy(2008)通过Fanand Goyal(2006)的产业间纵向相关性界定公司部门间的纵向关系，以1981至2005年间93，229个美国上市公司年度观测值为样本，结果发现公司部门间纵向关系系数越高，公司总部越可能将投资配置到具有更好投资机会的部门，意味着部门间纵向关系会有助于提高公司内部资本市场的资本配置效率。同时Shenoy(2009)通过Fanand Goyal(2006)的并购类型分类方法，以1981至2004年间453个美国上市公司并购事件为样本，结果发现公司纵向并购宣告期的累计异常收益率显著为正。此外，公司纵向并购还会增加市场势力、生产效率，防止广告
公司纵向并购财富效应的实证研究的搭便车行为和阻止研发投资支出的外溢效应，从而也会增加公司纵向并购财富效应(Bishop et al., 2005)。当然，公司纵向一体化会形成更复杂的组织结构，这会造成核心业务的忽视、部门间无效的交叉补贴、部门经理间权力之争和较差的激励薪酬计划，从而损毁公司价值，进而降低公司纵向并购财富效应(Eckbo皿dTho血1m，2008)。2010年6月底召开的国务院常务会议提出，切实推进企业兼并重组，促进产业结构优化升级，充分发挥资本市场推动企业重组的作用，并于2010年8月28日发布了重要的政策性文件一《国务院关于促进企业兼并重组的意见>(国发[2010)27号)。随着我国资本市场制度基础设施的快速发展和不断完善，同时为了应对我国经济发展到现阶段所面临的资源约束、产能过剩、需求降低的现状，及提升我国企业在国际上的竞争力，我国企业需要通过纵向并购来重塑产业链、进行产业链内企业的整合，从而形成完整的产业和价值链，以充分发挥协同效应，降低总体戚本，增强公司的核心竞争力。公司纵向并购能够增加收购公司进行关系专用性投资的激励及规避潜在要挟问题引起的契约非效率，但是它也会增加组织内部的官僚主义成本和降低目标公司进行关系专用性投资的激励(Williamson，1975, 1985 ; Klein et al., 1978 ; Grossman and Hart, 1986; Hart and Moore, 1990)及复杂组织结构造成的价值损毁效应(Eckboand Thorburn, 2008) ，这种纵向并购产生收益的增加和引起成本的增加间权衡会对公司边界的调整和公司并购绩效产生决定性的影晌(李青原，2006)。与西方发达的市场经济国家相比，处于转轨加新兴经济中的我国缺乏完善的产权保护、契约实施以及发达的市场要素等。在经济转型的过程中，由于缺乏良好的价格体系和完善的法律体系，从定价和法律执行的角度看，交易成本将十分昂贵(罗党论和唐清泉，2009) ，同时各级地方政府出于对经济发展的引导、稳定经济以及干预经济发展的战略目的，普遍存在对要素资源的配制、管制权和定价权的内在要求(黄益平，2009) ，那么纵向一体化作为一种非正式的制度则能减少产业上下游间的交易成本、规避过量的监管负担、政治干预和保证可靠的要素供应等(Fanet α1. , 2009) ，从而会使得产业上下游间公司并购效益跑过其相应增加的戚本。并且本文认为这种纵向并购的财富效应可能主要集中于民营上市公司:首先，由于我国现阶段地方官员任期的不稳定性和政治升迁中对绩效考核的过度关注，因而地方政府往往会出于政策性负担或者政治晋升目标而非基于公司效率最大化目标，主导其控制的公司的并购决策行为(张军和高远，2007) ，所以这些公司的纵向并购较多体现为政府干预下"拉郎配"式并购(陈信元和黄俊，2006) ，因此这种由政府控制的产业上下游间的"拉郎配"式并购会导致并购的低效率、甚至产生更大的交易成本。其次，面对国家产业政策和国资委考核等的巨大压力，中央政府控制的公司可以利用其权威和天然的"血缘关系"较顺利地收购企业做强做大，实现资源的有效整合，形成完整的产业链和价值链，以充分发挥协同效应(方军雄，2008) ，但是"所有者"缺位的治理结构对公司重大投资缺乏有效的监督和约束，极易导致公司过度投资，同时还会因公司规模的增加而导致组织内部官僚主义成本相应的增加(Williamson，1985)。最后，私人控股的公司是一个相对更加以利润最大化或股东价值最大化为经营目标的公司，其出于公司战略需要而进行的并购可能更符合市场规律导向，同时在转轨经济中，由于法律对私有财产权保护不明确，私营企业的发展存在不确定性，面临更大的风险(比如被勒索、遇到纠纷时受到不公平待遇等)，而且与国有企业相比还面临着较大的政策劣势(Cullet 此，2005) ，因此它们更可能会通过产业上下游间的公司
李青原王永超6 纵向并购活动来保护和促使交易双方事前的关系专用性投资激励和防止交易对方潜在的敲竹杠行为(Fanet al., 2009 ; Acemoglu et al., 2009 ;李青原，2011)。因此，本文提出第一个可检验的研究假设:Hl :某它条件不变时，并购交易双方间纵向关系与公司并购财富效应正相关，且这种纵向并购财富效应可能主要集中于民营上市公司。尽管交易费用经济学(Kleinet al., 1978 ; Williamson, 1975, 1985)和产权理论(Grossman and Hart, 1986; Hart and Moore, 1990)都强调了契约不完全性，但现有的实证研究都隐含假设即使契约是不完全的，但契约都会得到良好的实施(Duet al., 2009) ，从而忽视了法律环境和契约实施对企业边界的影响(Klein，2005)。然而，由于企业面临的外部环境的复杂性以及不确定性，信息的不对称和不完全性，以及契约当事人或契约的仲裁者无法证实和观察一切因素(Grossmanand Hart, 1986) ，这就会造成契约条款的不完全性，加之契约实施作用于法庭正式程序之外，即所谓的"法律阴影"之下(inthe shadow of the law) ，况且处于转轨加新兴经济中的中国还处于交易不发达、市场制度不成熟的发展阶段，不健全的法律与司法体系、不充分的信息披露、较弱的市场监管力量及社会信任度使得交易双方更难以实施正式契约(Khannaet al., 2000 ;张维迎和柯荣住，2002 ; Allen et αI叮2005;蔡洪滨等，2006)。一旦交易方做出关系专用性投资，那么无论交易契约是否完全，较低程度的产权保护和契约实施都会引发交易主体的机会主义行为和事前次优的关系专用性投资，从而导致高昂的市场交易成本和较低的盈利能力，因此地区产权保护和契约实施越弱，该地区的公司越可能会通过纵向一体化来保护和促使交易双方事前的关系专用性投资激励、防止交易对方潜在的敲竹杠行为，节约高昂的市场交易成本、同时还能规避过量的监管负担和政治干预(Fanet al., 2009 ; Acemoglu et al., 2009)。现如今，我国社会主义市场经济体系才刚刚建立，各项法律、法规都存在不完善的地方?对于法律的保护以及执行相对较为薄弱，因而通过纵向一体化作为一种替代机制来保障交易顺利实现则显得至关重要。Fan et al.(2009)以2001至2003年我国深沪两市的上市企业为样本，结果发现地区产权保护越弱，该地区的公司纵向一体化程度越高。Duet al.(2009)以2003年初世界银行与中国国家统计局合作的投资气氛调查项目提供的我国制造业企业为样本，结果发现越弱的契约实施自我感知会导致企业纵向一体化程度越高，而且地区契约实施程度越低，该地区内越依赖于外部环境的企业，其纵向一体化程度相应越高。李青原和唐建新(2010)运用相同的样本也发现契约实施强度越弱，公司纵向一体化程度相应越高。李青原(2011)以1998至2008年间发生的、上市公司作为收购方的国内非关联股权标的纵向并购为研究对象发现，随着公司并购交易双方所在地区产权保护程度的降低，资产专用性与公司纵向并购财富效应间的正相关性越明显。唐建新和陈冬(2010)以2003至2008年间发生的、上市公司作为收购方的异地非关联并购为研究对象，地区投资者保护是影响公司并购协同效应的重要因素，这种影响主要源于交易成本的节约和市场摩擦的下降。与以有效的法律和司法体系、发达的金融体系等为特征的发达市场国家相比，这种以市场失灵和弱契约实施等为特征的我国新兴加转轨市场环境会影响到纵向并购产生收益的增加与引起成本增加间的权衡，从而会相应地影响到公司边界的调整及公司并购绩效。世界银行(2006)对中国120个城市的投资环境调查显示，各个城市
公司纵向并购财富效应的实证研究之间的产权保护存在着相当的差异，那么这种地区间产权保护的程度差异必然会影晌到产业上下游间公司纵向并购财富效应。特别是，如果产业上下游交易双方都处于产权保护和契约实施较弱的地区，它们都可能更易引发机会主义行为、事前次优的关系专用性投资，因此它们越愿意通过纵向一体化来保护和促使交易双方事前的关系专用性投资激励和防止交易对方潜在的敲竹杠行为，节约高昂的市场交易成本，及规避过量的监管负担和政治干预等。本文提出第二个可检验的研究假设:H2:其它条件不变时，产业上下游阎公司纵向并购财富效应会随着公司并购交易双方所在地区产权保护程度的降低而增加。4三、研究设计(一)样本选择我们以CSMAR中国上市公司并购重组研究数据库中的1998至2008年发生的、上市公司作为收购方公司的国内非关联股权标的并购事件作为初选样本，然后按照以下的标准进行筛选:第一3由于财务指标不可比，剔除收购方公司或目标公司为金融行业的并购事件，第二，前后两次并购至少间隔6个月;第二，当同一家上市公司在同一天宣告两笔或两笔以上的并购交易时，如果目标公司不是同一家公司时，为避免目标公司异质性对并购事件的累积超常收益率产生噪音，对这样的并购事件给予剔除，如果是同一家上市公司与同一目标公司的不同股东进行并购交易，则将这些交易合并为)个事件;第四，剔除上市当年发生的并购事件;第五，为避免同时披露季报、中报或年报等可能会对公司并购的累积异常收益率产生噪音，剔除在并购首次公告日当天同时披露季报、中报或年报等其它重大信息披露的样本公司;第六，剔除无法识别目标公司行业的并购事件;第七，剔除财务指标存在缺失的并购事件;第八，剔除交易金额小于当00万元，并购前持有股权比例超过50%或收购的股份占目标公司股权比例低于10%的并购事件。最终得到有效的公司并购总样本为377个。(二)纵向关系度最Fan and Lang(2000)创造性通过投入产出表中产品流量构造了以投入产出为基础的方法来反映产业间和部门间的纵向相关性，再以此为基础构造了公司纵向一体化程度，该方法得到了学者们广泛的运用(Fanand Goya!, 2006 ; Kedia et 砾，2008; Shenoy, 2009 ; Fan et al., 2009)。首先，对于收购方公司，通过CSMAR数据库中的分行业主营业务收入数据找到第一大主营业务收入的行业信息，而对于目标公司，逐→查阅井购公告中披露的目标公司行业信息;其次，构造收购公司与目标公司行业间的纵向相关性系数yij。中国国家统计局提供的2002年部门投入产出表包括了122个行业中每一对行业间的投入产出系数，其主要说明了对于每一对行业i和Ij'生产1元的行业j的产出所需要行业i的投入值(A)和生产1元行业i的产出所需要行业ij的投入值(A)，那么我们通过两个投入产出系数的最大值计算收购公司与目标公司行业i和jl目的纵向相关系数(飞)，即飞=Max(A俨A)，从而能有效地解释它们间潜4 事实上，该假说可以较好地排除产业组织文献中市场封闭、共谋假设对公司纵向并购财富效应的解释。
李青原王永超表1样本按照《上市公司行业分类指寻1>的大类行业分布收购公司目标公司农、林、牧、渔业% % 采掘业% % 制造业% % 电力、煤气及水的生产和供应业% % 建筑业% % 交通运输、仓储业% % 信息技术业% % 批发和零售贸易% % 房地产业% % 社会服务业% % 传播与文化产业% % 综合类% % 资料来源·作者整理。在的纵向-体化机会(Fanand Goyal, 2006)。以2007年北京京能热电股份有限公司(代码:600578)收购内蒙古伊泰京粤酸刺沟矿业有限责任公司为例，收购公司的第一大行业电力、热力的生产和供应业的为每1元的目标公司煤炭开采和洗选业(j)产出提供元的投入值，即Aυ= '而煤炭开采和洗选业(j)为每1元的电力、热力的生产和供应业的产出提供元的投入值，即AP=，则收购公。司与目标公司行业i和/间的纵向相关系数几为图l显示了公司并购双方间纵向相关系数的累计概率分布。按照Fanand Goyal(2006)的公司并购类型分类，若将收购公司与目标公司间纵向相关系数飞超过临界值1%的样本界定为纵向并购，那么%的公司并购样本的纵向相关系数小于1% '相应地若将收购公司与目标公司间纵向相关系数超过临界值5%的样本界定VSJ为纵向并购，贝。另.7%的公司并购样本的纵向相关系数小于5%'这意味着与Fanand Goyal(2006)的研究样本相比，具有较高纵向关系的并购样本所占比重较高，这可能与以市场失灵和弱契约实施等为特征的我国新兴加转轨市场环境有关。.8 持台.6 M何时再十.4 O o .2 .4 .6 纵向相关系数图1公司并购双方间纵向相关系数的累计概率分布
公司纵向并购财富效应的实证研究9 (三)模型设计为了检验纵向关系对公司并购财富效应的影响程度，根据Fanand Goyal(2006)、Kediaet al.(2008)和Shenoy(2009)等的研究，本文选取收购公司规模、交易规模、资产负债率、成长性、收购公司市场集中度、行业管制等作为检验模型的控制变量，回归模型如下:VH1αR42=λrλlV~，，+λFHCj，，+λ3λ4MBW川乒lZEj勺，t_l手λVOLλjÆ川ERFλζ/DIOSλJ6，，+.J飞I/f户;，/町μ手λCGOLGO尺，fλ1211陀sfλI~OAjt_l+λ14REGM，川PROTECTφλ16血4RJZE(l)l乒，式(1)中CAR是被解释变量，表示收购方公司首次并购公告日前后若干个交易日的累积异常收益率，以度量公司纵向井购的财富效应(Lafontaineand Slade, 、2007)。遵循Fanand Goyal(2006)Kediaet al.(2008)和Shenoy(2009)等的主流文献，我们将公司纵向并购财富效应分别界定为运用市场模型法计算出的并购首次公告日、、前后125和10个交易日收购方公司的累积异常收益率，其中β参数的估计区间为并购首次公告日前180个交易日至公告前个交易日。30、、按照McGuckinet al.(1991)Fanand Goyal(2006)Kediaet al.(2008)和Shenoy(2009)等的公司并购类型分类方法，同时又参考了图l中公司并购双方间纵向相关系数的累计概率分布特征，将收购公司与目标公司行业I和j间的纵向相关系数飞超过1%或5%的并购样本界定为纵向并购(V。基于此分类标准，将收购公司与C)目标公司属于不同行业，但纵向相关系数飞超过临界值1%或5%的样本界定为"完全纵向并购(purevertical mergers)" ，将其定义为PVC'其中PVC为完全纵向井购，取值为1'否则为o，将收购公司与目标公司属于同一行业，但纵向相关系数几小于临界值1%或5%的样本界定为"完全横向并购(purehorizontal mergers)" ，将其定义为PHC'其中PHC为完全横向并购，取值为10;'否则为将收购公司与目标公司属于同一行业，同时纵向相关系数超过11自界值1%或5%的样本界定为"泪合纵向并购Vtl，其中(mixed horizontal and vertical mergers)" HVC为混合并购，取值为1'否则为o将界定为"完全纵向并购"、"完全横向并购"和"泪合纵向并购"之外的并购样本界定。为"t昆合并购(conglomeratemergers)"根据已有文献，公司特征和并购交易特征会影响并购收益，同时结合我国上市公司并购活动的实际情况，我们使用如下控制变量:公司特征变量包括公司规模、负债水平、成长性SIZE LEVGROWTH、产品市场集中度AHI、投资者异质性预期IDIOS、股权制衡度HERFS、资产收益率ROA、地方政府控股LGOV、中央政府控股CGOV;交易特征变量包括交易规模VOL、行业管制REG和并购年度阻仪等。公司规模越大，收购方公司越易存在过度自信(Roll1986) ，越易进行溢价收购，(Moeller et )。根据自由现金流量假说，收购方公司自由现金流大且负债率, 低时，更易进行无效并购(J司成长性影晌并购收益，追ensen1986)。公求公司成长, 驱动公司实施战略性并购，但盲目扩张会导致公司价值下降;收购方公司的治理状况越好，越会选择进行价值创造型的并购(Servaes，1991)。不同的并购规模下收购方公司对目标公司产生影响的能力不同，进而影响到收购方公司能够获取的并购收益
10 李青原王永超(Fuller et此，2002)。如果公司并购双方属于纵向井购，则该并购活动可能改变其市场势力和市场份额，而且还会在非竞争性市场创造价值(Kediaet 此，2008 ; Shenoy, 2009) ，故本文加入收购公司并购行业的赫芬达尔指数(AHI)，以反映它们所在行业的产品市场竞争程度，而且预期符号为正。此外，公司最终控制人的产权性质可能会影响到收购公司进行并购的动机和能力，进而影响到公司并购财富效应，因此本文又控制了地方政府控股和中央政府控股这两种最终控制人的产权性质对公司并购财富效应的影响。收购公司的股价表现不仅与公司的决策行为相关，同时也与投资者的信心和对市场的预期有重要关联。在做空套利机制受限的情况下，由于存在事前的信息不对称，投资者的异质预期会导致对收购公司股价的严重高估，市场对错误股价的反应有滞后性，股价回归基本面需要时间，投资者长期持有的损失可能是先前的错误定价导致的，故投资者的异质性预期σDIOS)可能会负向影响收购公司公告日前后若干个交易日的累积异常收益率(赖步连等，2006)。政府管制改变了市场交易成本和公司内部组织成本的对比关系，从而使得公司通过并购来绕过政府管制，进而影响了公司的经营边界，且经验证据表明政府规制政策是影响厂商边界的重要因素之一(陈信元和黄俊，2006 ;李青原和唐建新，2010) ，故将目标公司是否属于政府规制行业(煤炭、钢铁、石油、天然气、水供应、金属、航空、电力、铁路、邮政、电信等)(Fan et 此，2009)作为政府规制政策的替代变量(REG)0 由于新会计准则的实施，股权分置改革、《公司法》、《证券法》及《上市公司收购管理办法》的重新修订与实施，故2006年是我国资本市场的转折点(中国证券业监督管理委员会，2008) ，本文将YEAR定义为并购事件发生于2007和2008年，且取值为1'否则为O。变量定义如表2所示。四、检验结果及分析(一)描述性统计表3示了描述性统计。在公司并购首次公告的窗口期刊，+1]、[-2，+2]、[弓，+5]、[-10，+10]内，收购方公司均获得的累积异常收益率(CAR)均值分别为、、、'这表明我国上市公司对非上市公司的收购会带来显著正的累积异常收益率，与李善民和陈玉里(2002)等的研究结论相一致，也与Fulleret al.(2002)、Moelleret al.(2004)、Faccioet al.(2006)对以美国或非美国上市公司收购非上市公司事件为样本的研究结论相一致。以收购公司与目标公司行业I和j间的纵向相关系数几超过1%(5%) 的并购为分组标准，纵向并购占总样本比例为%(%) ;完全横向并购占总样本比例为%(%) ，完全纵向并购占总样本比例为%(%) ，混合纵向并购占总样本比例为%(%) ，这意味着与以有效的法律和司法体系、发达的金融体系等为特征的发达市场国家相比，在以市场失灵和弱契约实施等为特征的我国新兴加转轨市场环境下，纵向并购更可能会因节约产业上下游间的交易成本而成为一种较普遍的现象。与此同时，公司并购财富效应的分组检验显示，纵向并购样本的平均累积异常收益率(CAR)超过了总样本的平均累积异常收益率，而且进一步细分并购类型发现混合纵向并购财富效应超过其它类型并购的财富效应。表3Panel C中主要变
公司纵向并购财富效应的实证研究11 表2变量定义与符号预期变量名变量定义预期CAR 收购方公司首次并购公告日前后若干个交易日的累积异常收益率VC 收购公司与目标公司间纵向相关系数飞超过临界值1%或5%的并购样本，取值为1'否则为O+ 收购公司与目标公司属于不同行业，但纵向相关系数vPVC 超过临界值1%或当%的并购样本，取值为1'否则为o+ PHC 收购公司与目标公司属于相同行业，但纵向相关系数V小于临界值1%或5%的并购样本，取值为1'否则为O+ VHC 收购公司与目标公司属于相同行业，同时纵向相关系数γ超过临界值1%或5%的并购样本，取值为1'否则为O+ LPROTECT表示较低的地区产权保护指数，若收购公司和目标公司所在地区间的产权保护指数最小值都位于40分位数以下()，其中产权保护指数来自于世界银行对我国120个城市投资环境调查数据，则将其定义为1'否则为O+ 收购方公司在并购前一年总资产的自然对数SIZE 收购方公司在并购前一年的资产负债率LEV + GROWTH 收购方公司在并购前一年市帐比率+ AHI 按照《上市公司行业分类指引》的工级分类计算收购公司所处行业的赫芬达尔指数+ IDIOS 运用市场模型法计算出的并购首次公告日前180个交易日至公告前30个交易日的回归残差平方之和HERFS 公司前5位大股东持股比例的平方和+ ROA 收购方公司在并购前一年的资产收益率+ VOL 本次公司购买目标公司的股权比例+ LGOV 若最终控制人为地方政府，取值为1'否则为OCGOV 若最终控制人为中央政府，取值为1'否则为O目标公司是否属于政府管制行业(煤炭、钢铁、石油、REG 天然气、水供应、金属、航空、电力和铁路等)+ E宜ÆR如果并购事件发生于2007和2008年，则取值为1'否则为O+ 量的Spearman相关系数矩阵显示，收购方公司在并购首次公告窗口期内累积异常收益率CAR[斗，+1]、CAR[-2，+2]、CAR凹，而和CAR[-1O，+10]相互间都在1%的显著性水平下正相关，且相关系数至少为'而且其与1昆合纵向并购(VHC)至少在5%的水平下显著正相关。此外，收购公司为地方政府控股的企业占%'收购公司为中央控股的企业占18%'中央控股样本少于地方控股样本;交易规模VOL的均值(中位数)56%(51%)表明公司纵向并购多为控股收购。
12 李青原王永超表3分组检验和Spearman相关系数7兰'G1主、完全混合总体纵向并购纵向并购横向并购纵向并购混合并购Panel A : 1%的临界点CAR[件,+l] **东*** * *** CAR[-2,+2] *** *** * *** 件*CAR[-5,+5] ** *** 材*CAR[-10,+10] * * * * V N 377 259 148 8 111 110 Panel B : 5%的临界点CAR[林*,+l] *** *** *** CAR[-2,+2] *** *** ** *** CAR[-雪，+5]** *** ** *** CAR[-10,+10] * 树 * *** * V N 377 167 77 29 90 181 Panel C :主要变量的Spearman相关系数CARH,+l] CAR[-2,+2] CAR[-5，+当]CAR[-10,+10] PVC PHC CAR[-l,+l] *** CAR[-2,+2] *** CAR[-5，+当]*** *** CAR[-10,+10] *** *** *树 PVC PHC 树 * VHC ** *** 树**** *** * 其中，变量描述见表2时、材、*分别表示显著性水平为1%、5%、10%。此外，Panel C中的1%的并购类型分组标准为1%。(二)检验结果与分析以CAR[-l，+l]为因变量，本文分别运用模型(1)对公司并购类型的财富效应进行普通最小二乘法回归分析，得到各变量的回归系数。首先，本文对所有回归模型进行White(1980)异方差检验，均拒绝了存在异方差的零假设，表明检验模型没有异方差问题;其次，考察了模型中自变量的VIF值，发现所有自变量的VIF值均小于5'表明模型没有多重共线性问题，因此检验模型的建立是合理的o
公司纵向并购财富效应的实证研究13 1、假设Hl的检验表4Panel A中模型(1)和(2)显示，对全部样本进行的回归中，以收购公司与目标公司行业i和j间的纵向相关系数飞超过1%的并购为纵向并购时，纵向并购(VC)的回归系数边际显著为正，而且再按照收购公司与目标公司是否属于相同行业进一步细分井购类型，结果发现泪合纵向并购(VHC)的回归系数显著为正，同时表4Panel B中模型(1)和(2)也表明，若以收购公司与目标公司行业i和/间的纵向相关系数yzj超过5%的并购为纵向并购时，纵向并购(VC)、混合纵向并购(VHC)和完全纵向并购(PVC)的回归系数都显著为正，意味着相对于其它并购类型，纵向并购将会获得至少%的更高累积异常收益率，同时具有较强纵向关系的横向并购将会获得至少%的更高累积异常收益率，意味着并购交易双方间纵向关系越高，公司并购财富效应越明显。本文又按公司最终控制人的产权性质进一步分组检验，结果发现仅当最终控制人为私人企业控制时，公司纵向并购(VC)、1昆合纵向并购(VHC)和完全纵向并购(PVC)的回归系数都显著为正，而其它产权性质的公司并购类型回归系数不显著，同时较低地区产权保护组分别与纵向并购(VC)、混合纵向并购(VHC)和完全纵向并购(PVC)间交叉项的回归系数也都为正，且该系数至少在10%的显著性水平下显著异于零，意味着产业上下踌间公司纵向并购财富效应可能主要集中于民营上市公司，支持了本文的研究假设Hl。这表明我国不完善的法律和司法体系、较差的契约实施和低效的社会信任条件下，纵向关系可以作为一种替代机制，保护和促使交易双方事前的关系专用性投资激励和防止交易对方潜在的敲竹杠行为，节约高昂的市场交易成本，还能稳定产业链上下游公司间的产品供销关系及一定程度上缓减地区保护主义(Fanet al., 2009 ; Acemoglu et al., 2009) 0 5同时相对地方和中央政府控制的公司而言，私人控股的公司更可能面临着弱的产权保护和契约实施，从而使得它们更有动力和意愿通过产业上下游间的公司纵向并购活动来保护和促使交易双方事前的关系专用性投资激励和防止交易对方潜在的敲竹杠行为，节约高昂的市场交易成本，及规避过量的监管负担和政治干预等(Fanet al., 2009 ; Acemoglu et 此，2009) ，进而会提高这些具有较高纵向关系的私人控股型公司并购绩效，2、假设H2的检验由于各地区产权保护程度在不同年度间相对稳定，本文采用了世界银行报告的中国120个城市的产权保护指数作为地区产权保护的替代变量，从而能在城市而非省区的视角度量产权保护程度。将收购公司和目标公司所在地区间的产权保护指数9 当然，公司纵向并购的样本具有选择性偏见。由于Heckman(1979)选择模型依赖于特定的函数间接估计公司选择纵向并购的处理效应，而且还需要一个有效的工具变量，否则估计是有偏的，但现实中难以找到有效的工具变量，同时以属性为基础的匹配模型并不依赖于特定的函数，且能直接估计公司选择纵向并购的处理效应(Liand Prabhala. 2007) ，因此本文又运用倾向评分模型(propensity-score matching models)找到与纵向并购样本相匹配的样本进行回归分析，此外又以纵向并购前后两年间的总资产报酬率(税前利润/总资产)平均数为因变量，通过倍数差分法进行回归分析，这些结果仍然支持本文的假设，即并购交易双方间纵向关系与公司并购财富效应正相关。
14 李青原王永超最小值都位于40分位数以下()的样本定义为地区产权保护较低组，6其中产权保护指数来自于世界银行对我国120个城市投资环境调查数据，并将其定义为1'否则为0，并通过其与公司并购类型的替代变量(VC、VHC、PVC和PHC)交叉相乘项来检验上述理论推测。表5Panel A中模型(1)和(2)显示，较低地区产权保护组分别与纵向并购(VC)、混合纵向井购(VHC)和完全纵向并购(PVC)间交叉项的回归系数都为正，且该系数至少在10%的显著性水平下显著异于零，意味着公司并购交易双方所在地区产权保护程度越弱，公司通过产业上下游间纵向并购获得的财富效应越强，支持了本文的研究假设H2'即产业上下游间公司纵向并购财富效应会随着公司并购交易双方所在地区产权保护程度的降低而增加。7这意味着在不完善的法律制度、较差的契约实施和低效的社会信任条件下，纵向一体化可以作为一种替代机制，保护交易双方在产业上下游间有价值的关系专用性投资免受机会主义行为的侵害，节约交易戚本，及规避过量的监管负担和政治干预等。此外，表4和当中控制变量的回归结果表明，年度哑变量YEAR的回归系数为正，且基本显著性异于零，意味着随着会计准则、资本市场监管法律法规等资本市场基础设施的完善，上市公司收购时面临的外部治理环境得到改善和加强，从而增加了并购财富效应。此外，收购公司的产品市场集中度AHI不显著为负，意味着收购公司的财富效应并非来自于其市场势力的加强。尽管本文加入了文献中常见的控制变量，但是这些控制变量都不显著，原因还有待进一步探讨。(三)稳健性检验为了检验结果的稳健性，我们对表4和乡的结果进行了敏感性测试。首先，将收购方公司在并购首次公告窗口期内累积异常收益率CAR[-2，+2]、CAR[号，+5]和CAR[ -10，+10]作为因变量重新进行普通最小二乘法(OLS)估计;(2)若公司管理层拥有公司并购潜在协同效应的私人信息，那么普通最小二乘法(OLS)估计是有偏的(Eckboe t a 1., 1990) ，因此本文又进行了一致性最大似然估计;(3)由于本文是通过收购公司第一大主营业务收入行业界定公司并购双方的纵向关系系数，事实上收购公司的第二大主营业务收入行业与目标公司也可能具有重要的纵向关联性(Fanand Goyal, 2006) ，因此首先结合收购公司第一大和第二大主营业务收入行业界定公司并购双方的纵向关系系数，并将收购公司与目标公司行业i和j间的纵向相关系数yij1%或5%作为公司是否具有纵向关系的临界点，然后进行普通最小二乘法(OLS)估计。稳健性检验结果都表明这些检验都在所有的重要结论上基本与前面保持一致。6 约15%的公司并购交易样本所在地区不属于世界银行(2006)调查的120个城市范围内，为了增加样本量，本文通过本省所在具有数据的城市平均值来替代其它城市，同时本文存在一个既无地区调查数据，又无省平均值的样本，即拉萨地区样本(600773)，但由于拉萨地区属于中西部地区，假定其属于产权保护较低的地区样本是合理的。事实上，若临界值采用30%或50%时，结果依然保持不变。当然，当地区产权保护程度来自《中国市场化指数一一-各地区市场化相对进程2006年报告》中"市场中介组织发育和法律制度环境"时，结论基本保持不变3仅是显著性系数略有下降。7 值得注意的是，以纵向并购前后两年间的总资产报酬率(税前利润/总资产)平均数之差为因变量，结果发现研究假设H2中测试变量的回归系数虽符合预期，但它们基本不显著。
公司纵向并购财富效应的实证研究15 表4纵向关系与公司并购绩效Panel A:纵向相关系数V的临界点为1%全部样本样本分组(2) 地方政府控股(3)中央控股(4)私人控股(当)(1) 树VC * () () () () PVC * () () () () PHC () () () () VHC 柑 ** () () () ( 1) LPROTECT 。当 () () () () () () () () IDIOS * () () () () (.) () () () HERFS () () () () () () () () VOL () () () () () () () () SIZE () () () () () () () () LEV * * () () () () () () () () ROA () () () () () () () () 来MB () () () () () () () () LGOV () () CGOV * * () () AHI * * () () () () () () () () , * 忡REG () () () () () () () () }包树** () () () () () () () () cons () () () () () () () () A句ustedR2 柿件** ** F N 377 377 176 176 68 68 133 133
16 李青原王永超Panel B:纵向相关系数V的临界点为5%全部样本样本分组(2) 地方政府控股(3)中央控股份私人控股(5)(1) VC *** *** () () () () PVC ** ** () () () () PHC () () () () *** VHC *** () () () () LPROTECT () () () () () () () () ID/OS () () () () () () () () HERFS () () () () () () () () VOL () () () () () () () () SIZE 国 () () () () () () () () LEV * * () () () () () () () () ROA () () () () () () () () MB () () () () () () () () LGOV () () CGOV * * () () AHI * * () () ( 0) (当)() () () () REG () () (国)() () () () () }包时树 () () () () () () () () CONS () () () () () () () () 2 A句 F *** *** * *** *** N 377 377 176 176 68 68 133 133 注:变量描述见表2;T值运用White(1980)进行方差修正;忡*、材、*分另IJ表示显著'性水平为1%、5%、10%。
公司纵向并购财富效应的实证研究17 表5产权保护、纵向关系与公司并购绩效Panel A:纵向相关系数v的临界点为1%全部样本样本分组(2) 地方政府控股(3)中央控股(4)私人控股份(1) VC * () () () () VC求LPROTECT 柑* ** () (1.%) () () PVC * () () () () PVC吃*** ** ** () () () () PHC () () () () PHC乞 () () VHC * * () () () () VHC*LPROTECT * 本 () () () () LPROTECT () () () () () () () () Control Yes Yes Yes Yes Yes Yes Yes Yes 2 A句u旧 件*F *** * * ** 忡N 377 377 176 176 68 68 133 133 Panel B:纵向相关系数V的临界点为5%全部样本样本分组(2) 地方政府控股(3)中央控股(4)私人控股(5)(1) VC * *** () () () () VC欢LPROTECT ** * () () () () PVC () () () () PVC干* 林 () () () ()
18 李青原王永超Panel B:纵向相关系数v的临界点为5%全部样本样本分组(2) 地方政府控股(3)中央控股(4)私人控股(当)(1) PHC () () () () PHC*LPROTECT () () () () VHC * 树() () () () VHC*LPROTECT * () () () () LPROTECT () () () () () () () () Control Yes Yes Yes Yes Yes Yes Yes Yes 2 Adjusred F *** *** * *** *** N 377 377 176 176 68 68 133 133 注变量描述见表2;T值运用White(1980)进行方差修正;忡七材J分别表示显著'性水平为1%、5%、10%。五、结论与以有效的法律和司法体系、发达的金融体系等为特征的发达市场国家相比，在以市场失灵和弱契约实施等为特征的我国新兴加转轨市场环境下，纵向一体化应该更能保护公司交易双方事前的关系专用性投资激励及防止交易对方潜在的敲竹杠行为。本文以1998至2008年间发生的、上市公司作为收购方公司的国内非关联股权标的并购样本为研究对象，结果发现并购交易双方间纵向关系较高，公司并购财富效应相应较高，而且产业上下游间公司纵向并购财富效应会随着公司并购交易双方所在地区产权保护程度的降低而增加，但是这种产业上下游间公司纵向并购财富效应只存在于民营上市公司作为收购方的并购活动。这表明，在我国不完善的法律和司法体系、较差的契约实施和低效的社会信任条件下，纵向一体化可以作为一种替代机制，保护私人控股收购公司交易双方事前的关系专用性投资激励和防止交易对方潜在的敲竹杠行为，节约高昂的市场交易成本，及规避过量的监管负担和政治干预等。本文研究发现产业上下游间公司纵向并购是一种有效率的资源配置行为，那么究竟什么动因驱动了我国上市公司纵向并购财富效应呢?同时我国新兴市场和转轨经济的特征使得我国上市公司纵向并购动因显得非常复杂和更具政策含义，因此今后的研究将运用投入产出表中产品流量找到公司纵向井购双方的竞争对手、上游供应商和下游客户，将效率假设、共谋假设、市场封闭假设等国外相关理论同中国实际相结合，构建适合我国上市公司纵向并购的动因理论体系。
公司纵向并购财富效应的实证研究15 参考文献蔡洪滨、张玻、严旭阳.2006."中国企业信誉缺失的理论分析".{经济研究》第9期，8予102。陈信元、黄俊.2006."政府管制与企业垂直整合刘永行炼铝的案例分析".{管理世界》第2期，134-138。方军雄.2008."政府干预、所有权性质与企业并购".{管理世界》第9期，118-123。冯根福、吴林江.2001."我国上市公司井购绩效的实证研究".{经济研究》第1期，抖，61。黄益平.2009."要素市场需引入自由市场机制".{财经报道}'7月26日o赖步连、杨继东、周业安.2006."异质波动与并购绩效:基于中国上市公司的实证研究".{金融研究》第12期，126-139。李善民、陈玉皇.2002."上市公司兼并与收购的财富效应".{经济研究》第11期，27-35。李青原.2006."公司并购绩效与公司边界·交易费用的视角".{南开管理评论》第1期，38-44。李青原、唐建新2010."企业纵向一体化的决定因素与生产效率一来自我国制造业企业的经验证据".{南开管理评论》第3期，60-69。李青原.2011."资产专用性与公司纵向并购财富效应一来自我国上市公司的经验证据武汉大学经济与管理学院工作论文。李善民、朱滔.2006."多元化并购能给股东创造价值吗?".{管理世界》第3期，129-137 0 罗党论、唐清泉.2009."中国民营上市公司制度环境与绩效问题研究".{经济研究》第2期，106-118。世界银行.2006."中国政府治理、投资环境与和谐社会:中国120个城市竞争力的提高报告编号-CN。唐建新、陈冬.2010."地区投资者保护、企业性质与异地并购的协同效应".{管理世界》第8期。0张军、高远.2007."官员任期、异地交流与经济增长".{经济研究》第11期'91-113张维迎、柯荣住.2002."信任及其解释:来自中国的跨省调查分析".{经济研究》第10期，59-70。中国证券业监督管理委员会.2∞008."中国资本市场发展报告周小春、李善民.2008."并购价值创造的影响因素研究".{管理世界》第5期，134-143。巫和愚、张晓明.2009."中国上市公司收购方损益之探究北京大学国家发展研究院工作论文。Acemoglu, D., Johnson，鼠，and Mitton T. (2009), Determinants of Vertical Integration: Financial Development and Contracting Costs’, Journal 0/ Finance 64 (3): 1251-1290. Allen, F叮Qian，]., and Qian, M. (2005), ’Law, Finance, and Economic Growth in China’ , Journal 0/ Financial Economics 77 (1):叨-116.
20 李青原王永超Bishop 鼠，Lofaro, A., Rosati，丑，and Young, J. (2005), The Eff ciency-Enhancing Effects ofr、Jon-HorizontalMergers’ , European Communities Report. Bruner, R. F. (2002), ’Does M and A Pay? A Survey of Evidence for the Decision-Maker’, Journal 01 App/ied Finance 12 (1):钮, R. (1937), The Nature of the Firm’ , Economica 4 (16): 386-405. Cull, R. and Xu, L. C. (2005), Institutions, Ownership, and Finance: The Determinants of Profit Reinvestment among Chinese Firms’ , Journal 01 Financial Economics 77 (1): 117-146. Du J., Lu, Y., and Tao, Z. (2009), Contract Enforcement and Vertical Integration: Evidence from China’ s Manufacruring Firms’ , Chinese University of Hong Kong, W˙lrking Paper. Eckbo, B. E., Maksimovic, V., and Williams, J. (1990), Consistent Estimation of Cross›Sectional Models in Event Studies’, The Review 01 Financial Studies 3 (3): 343-365. Eckbo, B. E. and Thorburn, K. S. (2008), Corporate Restructuring: Breakups矶dLBOs’, in Eckbo, B. E. (ed.), Handbook 01 Corporate Finance: Empirical C。中orateFinance, North-Holland: Elsevier. Faccio, M., McConnell, J., and Stolin, D. (2006), Returns to Acquirers of Listed and Unlisted Targets’, Journal 01 Financial and Quantitative Analysis 41 (1): 197-220. Fan, J. P. H. and La吨，L. H. P. (2000), ’The Measurement of Relatedness: An Application to Corporate Diversification’ , Journal 01 Business 73 (4): 629耐, J. and Goyal, V. (2006), On the Patterns and Wealth Effects of Vertical Mergers’, Journal 01 Business 79 (2): 877-902. Fan, J., Huang, J., Mord王，氏，and Yeung, B. Y. (2009), Vertical Integration, Institutional Determinants and Impact: Evidence from China’, Chinese University of Hong Kong, Working Paper. Fuller, K., Netter, J., and Stegemoller, M. (2002), What do returns to acquiring firms tell us? Evidence from firms that make many acquisitions’ , Journal 01 Finance 57 (4): 1763-1793. Grossman, S. J. and Hart, O. (1986), The Costs and Benefits of Ownership: A Theory of Vertical and Lateral Integration’, Journal 01 Po/itical Economy 94 (4): 691-719. Hart, o. and Moore, J. (1990), Property Rights and the Nature of the Firm’ , Journal 01 Po/itical Economy 98 (6): 1119-1158. Heckman, J. J. (1979) , Sample Selection Bias as a Specification Error’ , Econometrica 47 (1): 153-161. Jensen, M. C. (1986), Agency Costs of Free Cash Flow, Corporate Finance, and Takeovers’ , American Economic Review 76 (2): 323-329. Joskow, P. L. (2005), Vertical Integr
公司纵向并购财富效应的实证研究2 Khanna,T. and Palepu, K. G. (2000), Is Group Affiliation Profitable in Emerging Markets? An Analysis of Diversified Indian Business Groups’ , Journal 01 Finance 55 (2): 867-891. 阳ein，且，Crawford, R. G., and Alchian, A. A. (1978), Vertical integration, appropriable rents, and the competitive contracting process’ , Journal of Law and Economics 21 (2): 297-336. Klein, P. G. (2005), The Make-or-Buy Decision: Lessons from Empirical Studies’ , University of Missouri, Working Paper. Kvalły, O. (2007), Asset Specificity and Vertical Integration’ , Scandinavian Journal of Economics 109 (3): 551-572. Lafontaine, F. and Slade, M. (2007), Vertical Integration and Firm Boundaries: the Evidence’ , Journal of Economic Literature 45 (3): 629-685. Li, K. and Prabhala, N. (2007)，‘Self二selectionmodels in corporate finance’, in Eckso, B. E. (ed.), Handbook of Corporate Finance: Empirical Corporate Finance,. Elsevier Science ., 37-86. Macher, J. T. and Richman, B. D. (2008), Transaction Cost Economics: An Assessment of Empirical Research in the Social Sciences’ , Business and Politics 10 (1): 1-63. McGuckin, R. H., Nguyen, S. V., and Andrews, S. H. (1991), The relationships among acquiring and acquired firms’ product lines’ , Journal of Lαw and Economics 34 (2): 477-502. Milgrom, P. and Roberts, J. (1992), Economics, Organization, and Management, New Jersey: Prentice Hall International. 扎1oeller，S.丘，Schl ngemann, F.丑，and Stulz, . (2004), Firm Size and the Gains from Acquisitions’ , Journal 01 Financial Economics 73 (2): 201-228. Roll, R. (1988), ’R2’ , Journal of Finance 43 (3): 541-566. Scheffman, D. T. and Higgins, R. S. (2004), Vertical Mergers: Theory and Policy’ , George Mason Law Review 12 (4): 967-977. Servaes, H. (1991), ’Tobin’s Q, Agency Costs, and Corporate Control: An Empirical Analysis of Firm Specific Parameters’ , Journal of Finance 46 (1): 409-419. Shenoy, J. (2008), Determinants of Firm Vertical Boundaries and Implications for Internal Capital Markets’ , Georgia State University, Working Paper. Shenoy, J. (2009), An Examination of the Efficiency, Foreclosure, and Collusion Rationales for Vertical Takeovers’ , Georgia State U niversity, Working Paper. White, H. (1980), A Heteroskedasticity-Consistent Covariance Matrix Estimator and a Direct Test for Heteroskedasticity’ , Econometrica 48 (4): 817-838. Williamson, O. E. (1975), Markets and Hierarchies: An吵"Sisand Antitrust Implications, New York
l2 2011年9月第13卷第3期中计句是，义、国z;: 与财务研71二Ch na Account n9 and F nance Rev ew Volume 13, Number 3 -September 2011 The Wealth Effects ofVertical Mergers: Empirical 1 Evidence from Listed Corporations in China2 Qingyuan Li and Y ongchao WangAbstract This paper tests primarily the wealth effects of vertical mergers of Chin巳selisted firms. Studying a sample of domestic mergers with non-related-party acquirers and private target firms from 1998 to 2008, w巳缸ldthat the clos巳rthe vertical relation betwe巳nthe transaction parties of mergers and acquisitions (M&As), the higher the wealth effects of vertical mergers. Moreover, these wealth effects increase as local legal protection of property rights decreases, but only for those M&As whose acquirers are privately controlled listed corporations. The results indicate that presently in China, where laws and regulations are imperfect, law enforcement is poor, and social trust inefficient, vertical integration serves as a substitute mechanism to protect valuable relationship›sp巳cificinvestments from opportunism, pr巳ventpotential holdups for each party to the transaction, save on high transaction costs, and avoid excessive regulatory burdens and political intervention, especially for privately controlled listed corporations. Keywords: Vertical Merger, Vertical Integration, Vertical Relation Coefficient, Wealth Effects CLC codes: F425, F830 We would like to exprωs our gratitude to the National Natural Science Foundation of China (70672066 and 70702017), Key Projects of Philosophy and Social Sciences Research, the Ministry of Education (1 OJZDOO 19), and the Post-70s Scholars Academic Development Program of Wuhan University. We are deeply grateful to Dr Cong Wang, an associate professor of the Chinese University of Hong Kong, for his valuable comments provided at the China Accounting and Finance Review Intemational Symposium 2010, and to Dr Yanheng Song, an associate professor of Zhejiang University, for his useful review given at the 9th Intemational Symposium on Empirical Accounting Research in China We also greatly appreciate the suggestions for amendments of Prof Wayne W. Yu, the executive editor, and the anonymous reviewers. All eηors remam our own Qing严lanLi, Associate Professor, Economics and Management School, Wuhan University. Mailing Address: Department of Accounting, Economics and Management School, Wuhan University, 430072, Wuhan, China. E-mail: qyli@. Yongchao Wang, Master Candidate, Department of Accounting, Management School, Jinan University.
The WeaIth Effects of Vertical Mergers 2 1. Introduction A vertical merger is defined as a dynamic process of vertical integration in which upstream (downstream) companies acquire the control riglits of downstream (upstream) companies through purchasing shares of target firms so that transactions wi11 take place within the company rather than through the market or by contract. WiI1iamson (1975) predicted that vertical integration, which is used to explain the economic activities of resource allocation among firms, markets, and intermediate organisations, would become a core issue in what he termed the "new institutional economics". And indeed it has been an important subject in theoretical and empirical studies of microeconomics for at least the past 30 years (Joskow, 2005; Lafontaine and Slade, 2007). In his seminal paper "The Nature of the Firr口气Coase (1937) raises two basic questions: (1) what are the determinants of firm boundaries, and (2) do firm boundaries affect resource allocation? Since then, Williamson (1975, 1985), Klein et al. (1978), Grossman and Hart (1986), and Hart and Moore (1990) have continuously provided new insights and extended these studies to try to explain all types of vertical integration activities. Whether or not their explanations work, their studies have not only promoted theoretically the comprehension of vertical integration, but have also rendered empirically a theoretical basis for its motivation and economic consequences. Over the past near1y 20 years, scholars have paid much attention to empirical studies on vertical integration (且口nboundaries), focusing on two basic and interrelated questions: (1) what types of transactions are best made within the firm, and (2) what are the economic consequences of vertical integration in terms of prices and quantities of products, investment, and profits (Lafontaine and Slade, 2007)? Klein (200匀，Lafontaine and Slade (2007), and Macher and Richman (2008) reveal that traditional empirical studies usually take a single company or industry as their sample because of the accessibility of data, and examine the impact of vertical integration only on retail prices, quantities of products, systematic risk, and stock ratings. Thus, empirical studies on the causality of vertical mergers and co叩orateperformance are rare, and at most are only case studies of a small sample and a specific indus位 consequence may be attributed to the undeterminable intensity of vertical integration at tl
24 Li and Wang both Fan and Goyal (2006) and Kedia et al. (2008) def ne a vertical relation between the acquirer and target company by adopting the longitudina1 correlation coefficient used by Fan and Goyal (2006). Although they find the cumu1ative abnormal returns (CARs) to be signif cantly positive, Kedia et al. (2008) find them to be signif cantly negative after 1996. 3 Schef企nanand Higgins (2004), moreover, point out that compared with horizonta1 mergers, vertical mergers are more complicated and tend to be affected by certain institutiona1 arrangements. And although Feng and Wu (2001), Zhou and Li (2008), and Wu and Zhang (2009) use different methods and samples to test the M&A performance of Chinese 1isted companies and draw many usefu1 conclusions, their various definitions of merger types are so subjective that few of them really focus on the performance of vertica1 mergers. Thus, it remains worthwhi1e for scho1ars to conduct further research to improve on these results. With the rapid deve10pment and continuous improvement of the institutiona1 infrastructure in the Chinese capita1 market, vertica1 acquisitions such as the cases of China National Petro1eum Corporation, China National Offshore Oil Corporation, and Aluminium Corporation of China are continuing to increase, thereby obtaining synergy in the industry chain. So how do these companies perform? What are the institutional factors influencing the performance of vertical mergers? Considering the particu1arity of China’s institutiona1 arrangements and its unique capita1 market, in this paper we use a samp1e of domestic mergers with non-related-party acquirers and private target f rms from 1998 to 2008. We f nd that the closer the vertica1 re1ation coeff cient between the M&A transaction parties, the higher the wealth effects of vertica1 mergers; such effects a1so increase as 10ca1 protection of property rights decreases. These f ndings, however, exist on1y for those M&As whose acquirers are private1y controlled 1isted corporations. The results indicate that under the condition of imperfect 1aws and regulations, poor 1aw enforcement, and ineff cient socia1 trust, vertica1 integration serves as a substitute mechanism in present-day China to protect va1uab1e relationship-specif c investment from opportunism, prevent potentia1 ho1dups for each party to the transaction, save on high transaction costs, and avoid excessive regulatory burdens and politic Kedia et al. (2008) believe that the negative CARs of finns after 1996 may result from the fact that these vertical transactions were not motivated by fundamental forces in the industry favourable to vertical integration
The Wealth Effects of VerticaI Mergers 2’: Second, since the motives for corporate M&As are complicated, diverse, and multilateral, and since M&A performance tends to result from interactions between multiple factors, it is quite diff cult to use a particular theory to illustrate changes in overall performance (Bruner, 2002). Consequently, this paper provides a new perspective from which to explain the mystery of corporate M&A performance by looking at vertical relationships. Zhou and Li (2008) discover that the main factors affecting the performance of M&As between listed corporations in China are employee res stance, resource integration, payment method, and acquisition percentage. Us ng commodity flow informatìon仕oman nput-output (10) table, this paper further c1ass f es the types of d vers f ed M&As accord ng to the r degree of vert cal relatedness, and emphasises the in uence of th s degree of relatedness between the two parties on corporate M&A performance, thereby enhancing our understanding of the performance of acquiring f rms. F nally, th s paper enriches the relevant studies on vertical integration from the perspective of property rights and institutional background. In countries with trans tional economl白，ownersh p and inst tutional background are both mportant factors for understanding f rm boundaries (Klein, 2005; Fan et al., 2009; Acemoglu et al., 2009). S nce presently in Ch na the laws and regulations are imperfect, law enforcement s poor, and social trust ineff cient, vertical integration serves as a substitute mechanism to protect valuable relationship-specif c investments from opportunism as well as avoid excessive regulatory burdens and political intervention, and thus it provides new ideas for adjusting Chinese f rm boundaries. The remainder of this paper is organised as follows: Section 11 provides the background and hypothesis development, Section III describes the research design, Section IV presents the results of the empirical investigation, and the f nal section conc1udes. 11. Background and Hypothesis Development Since Coase (1937) f rst published his seminal paper ’The Nature of the Firm", Williamson (1975, 1985), Klein et al. (1978), Grossman and Hart (1986), and Hart and Moore (1990) have made theoretical improvements and innovations, forming the mainstream economic theory that effectively explains f rm boundaries. Joskow (2005) points out that the main theories explaining vertical integration fall into two categories: transaction cost economics (TCE), generally identified by Klein et al. (1978) and Williamson (1975, 1985), and property rights theory (PRT), generally identif ed by Grossman and Hart (1986) and Hart and Moore (1990). Williamson (1975, 1985) and Klein et al. (1978) indicate that once the counterparties make relationship-specific investments, contractuaI incompleteness Ieads to opportunistic behaviour; at the same time, appropriable specialised quasi-rents from asset specif city turn this probability of opportunism into reality and generate the holdup problem, thus leading to ex ante sub›optimal, relationship-specif c investments and low prof tability. One main solution to
Li and Wang 26 holdup problems is vertical integration, which eliminates such potential problems caused by ineff cient contracts between buyer and seller, thus improving investment eff ciency resulting仕omrelationship-specif c investments (Klein et al., 1978; Joskow, 2005). In a word, according to TCE, transaction costs created by contractual incompleteness are due mainly to臼postmaladaptation problems. Accordingly, TCE emphasises ex post adaptive govemance and maintains that vertical integration resolves the contract ineff ciency caused by under-investment of relat onship-spec f c assets and avoids potential holdup problems. This ineff ciency makes intemal govemance more effective than arms-length market transactions, and thus the wealth effect of vertical mergers becomes a function of govemance costs saved in the process of intemalising trading relationships (Milgrom and Roberts, 1992). Furthermore, Grossman and Hart (1986), and Hart and Moore (1990), follow ng the tradition of TCE, take asset specif city and contract incompleteness as a premise for understanding vertical integration, which they regard as the ownership of physical assets. Using residual rights of control over physical assets, they deduce that the distortion of the ex ante investment incentive is th巳resultof different ownership s位uctures;they further study the costs and benef ts of vertical integration by comparing the degree of investment distortions田ldervarious circumstances仕omthe perspective of incentives. That is to say, the incentive of the acquiring缸mto make a re1ationship-specif c investment will increase because the f rm will receive a bigger仕actionof the surplus from the transaction. This will, however, decrease the fraction of the acquired且rm，thus diminishing the latter’s incentive to make relationship-specif c investments. The incentive costs (distortion) of integration are moreover positively related to asset specif city (Kval町，2007). Compared with neoclassical economics, TCE paradigmatically has inspired more studies on vertical mergers (vertical integration), among which are numerical analysis studies of certain compan es, industr es, and cases of typ caI contracts, as well as a few econometr cs analyses based on large samples (Joskow, 2005). Fan and Lang (2000) innovatively use the dollar value of commodity flows between different industries provided by the benchmark IO accounts
The Wea1th Effects of Vertical Mergers 27 period. Shenoy (2008) defines vertical relations between business segments according to the inter-industry vertical relatedness created by Fan and Lang (2000), and selects 93,229 annual observations in the US between 1981 and 2005 as the sample. Her study finds that as vertical relatedness between segments increases, corporate headquarters allocates more investment to those segments with better investment opportunities; this indicates that vertical integration improves the capital allocation efficiency of intemal capital markets. Meanwhile, following Fan and Goyal (200哟，Shenoy (2009) identifies 453 vertical takeovers in the US between 1981 and 2004 and finds that CARs are significantly positive in the announcement period of a vertical merger. Additionally, vertical integration enhances market power and productive efficiency, prevents fr町-ridingon pre-sale services, and avoids the spill-over effects of investment in research and deve1opment, thereby increasing the wealth effects of the vertical merger (Bishop et al., 2005). Certainly, vertical integration might also lead to a more complicated organisational structure, which could result in ignorance of the core business, the inefficiency of cross›subsidisation, power struggles between department managers, and inferior incentive plans. Those factors might final1y destroy firm value as well as reduce the wea1th effects of vertical mergers (Eckbo and Thorbum, 2008). In an executive meeting in June 2010, the State Council of China pointed out the necessity of effectively promoting corporate M&As, upgrading industrial structure, and taking full advantage of the capital market in promoting corporate M&As. On 28 August 2010, it also published the important policy document, Opinions of the State Council on Promoting Corporate M&As. In view of the rapid development and increasing improvement of the nst tut onal infrastructure of the Ch nese cap tal market, and to address such present issues as resource constraints, over-capacity, reduced demand, and the need to upgrade the intemational competitiveness of domestic corporations, Chinese corporations need to rebuild industry chains and make inter-industry integrations with complete industry value chains using vertical mergers; these in tum could maximise synergies, reduce overall cos钮，and enhance core competitiveness. But although vertical integration can increase the incent
28 Li and Wang common for local govemments to regulate the distribution, confinement, and pricing of factor resources for the strategic pu叩oseof guiding economic development and stabilising and intervening in the national巳conomy(Huang, 2009)国Consequently，as an informal institution, vertical integration can decrease the transaction costs of upstream and downstream industries, avoid heavy-handed regulation and political intervention, and ensure a stable factor supply (Fan et al., 2009), and thus the benefits of vertical mergers outweigh the corresponding increased costs. At the same time, such wealth effects of vertical mergers may exist primarily in private listed companies. Firstly, considering the unc巳rtaintyof the tenure of local off cials and the excessive focus on performance evaluations during political promotion in China, local govemments may direct corporate M&As for reasons of policy burdens or political promotions, rather than maximisation of co甲orateefficiency (Zhang and Gao, 2008). A large majority of such merger activities are accordingly referred to as "forced marriages" under local govemmental intervention (Chen and Huang, 2006), resulting in corporate M&A ineff ciency and even higher transaction costs. Secondly, facing pressure 仕omnational industrial policies and performance evaluations by the State-owned Assets Supervision and Administration Commission (SASAC), c巳ntralgovemment-controlled firms are able to take advantage of their privileges and inherent connections to effectively integrate resources in order to make themselves bigg巳rand strong巳r，as well as construct integrated value chains and industry chains to fully achieve their synergies (Fang, 2008). But a co甲orategoverηance structure in which the "owner" is absent may lack eff cient monitoring and constraints on important investments, making it easier for the company to over-invest. Meanwhile, intemal bureaucratic costs increase with company expansion (Williamson, 1985). Finally, privately controlled firms are more dedicated to maximising profits or shareholder value, which is considered to be their operational objective, and so M&A decisions based on strategic needs may be more market-oriented. On the other hand, in a transitional economy lik巳China's，property protection is unclearly defined in law, and the development of privately controlled firms faces greater uncertainty and risk, such as blackmail, unfa Bl: All other conditions held constant, the vertical relation between the acquirer and the target firm will have a positive correlation with the wealth effects of merger activities, which will exist mainly in private listed firms. Although both TCE (Klein et al., 1978; Williamson, 1975, 1985) and PRT (Grossman and Hart, 1986; Hart and Moore, 1990) emphasise contractual incompleteness,
The Wealth Effects of Vertical Mergers 29 existing empirical studies hold an implicit assumption that contracts, albeit incomplete, can be well enforced (Du et al., 2009); thus, they ignore the impacts of legal environment and contract enforcement on firm boundaries (Klein, 2005). But because of the uncertainty and complexity of extemal environments, as well as information asymmetry and incompleten巳ss，contracting parties or arbitrators are unable to confirm and observe all factors (Grossman and Hart, 1986), resulting in contractual incompleteness. In addition, contract enforcement is conducted according to formal legal procedures, or in the so-called "shadow of the law". Under the transitioning and emerging Chinese economy, transactions are also still in the stage of under-development, and the market remains immature. An unsound legal and judicial system, inadequate information disc1osure, and weak market regulation and social trust thus make it harder for both parties involved to conc1ude formal contracts (Khanna et al., 2000; Zhang and Ke, 2002; Allen etα1.， 2005; Cai et al., 2006). Consequently, onc巳th巳transactionparties mak巳relationship-specific investments-whether the contract is complete or not-weak property protection and contract enforcement will lead to opportunism and ex ante sub-optimal, relationship-specific investments, resulting in ex post higher transaction costs and lower profitability. Therefore, the weaker the prope此yprotection and contract enforcement of a district, the more likely vertical integration wi1l be used to protect and promote valuable, relationship-specific investrnents仕omoppo此unism，thereby preventing potential holdups, saving high transaction costs, and avoiding excessive regulatory burdens and po1itical intervention (Fan et al., 2009; Acemoglu et al., 2009). Moreover, because the socialist market economy of China has been newly established with imperfect laws and regulations, and both legal protection and enforcement are relatively weak, vertical mergers are of vital importance as a substitute mechanism to guarantee smooth transaction enforcement. For instance, Fan et al. (2009) use samples of companies listed in the Shanghai Stock Exchange and Shenzhen Stock Exchange between 2001 and 2003 and find that vertical integration is more common when regional property prot巳ctionis weak. Du et al. (2009) use samples of Chinese manufacturing companies provided by the Surve
30 Li and Wang costs and alleviating market frictions. Thus, compared with developed countries that have efficient legal and judicial systems and advanced financial systems, the emerging and transitional market environment in China, characterised by market failure and imperfect contract enforcement, will influence the trade-off between the increment of benefits and the increase in costs induced by vertical mergers; th s wi1l accordingly affect the呻ustmentof firm boundaries as well as the performance of corporate M&As. A survey on the investment environments of 120 Ch nese cit es conducted by the W orld Bank (2006) further reveals significant differences in property protection among different cities which have a definite influence on the wealth effects of vertical mergers between upstream and downstream ndustries. In particular, f both the upstream and downstream transaction parties come from areas w th weak property protection and contract enforcement, t will be easier for them to exhibit opportunistic behaviour and make ex ante sub-optimal, relationship-specific investments. Thus, they will be more w lling to use vert cal integration to protect and promote ex ante relationship-specific investments, prevent potent al holdups, save high transaction costs, and avo d excess ve regulatory burdens and political ntervention. In this way, we put forward the second testable hypothesis: H2: All other conditions held constant, the wealth effects of vertical integration among upstream and downstream industries will increase as the degree of local 4 property protection . Research Design Sample Selection We choose domest c mergers with non-related-equity acquirers and private target firms between 1998 and 2008 from the Chinese listed-firm M&A and restructuring database in the CSMAR database as our primary sample, which we then screen according to the following cr ter a. First, we exc1ude merger cases in which financial firms are involved as the acquiring or target companies owing to the incomparability of their financial data. Second, the time interval between two consecutive merger cases should be at least six months for a specific firm. Third, when the same listed firm announces two or more than two merger events on the same day, (1) if the target companies in those events are different, we eliminate those cases to avoid affecting CARs owing to the target companies’ heterogeneity; and (2) if the mergers take place between the same listed firm and different shareholders of the same target company, we consolidate these transactions into a single event. Fourth, we eliminate those merger events taking place in the same year that the acquiring firm goes public. Fifth, we exc1ude cases where In fact, the hypothesis effectively excludes the explanations of the wealth effects of vertical integration in terms of market foreclosure and the collusion hypothe时，as described in the industrial organisation literature.
The Wealth Effects of Vertical Mergers 31 quarterly reports, interim reports, and annual reports are announced on the same day as the first day of the M&A announcement to avoid the impact on CARs. Sixth, we exclude cases where we are unable to identify industries in which the target companies operate. Seventh, we exclude cases with inadequate f nancial data. Eighth, we eliminate those merger events of less than 5 million renminbi, or in which the acquiring f rm already holds more than 50 per cent of the target company’s shares before the merger, or less than 10 per cent after the merger. Our f nal valid sample thus consists of 377 f rms. Table 1 M&A Activities by Industry Acquiring company Target company Agriculture, herding, f shery % % Mining % % Manufacturing % % Power, gas, water production and supply % % Construction % % Transportation and storage % % IT % % Wholesale and retail trade % % Real estate % % Social service % % Communication and cu1ture % % Integrated industry % % Note: Percentag巳srepresent the proportion of all M&A transactions in a particular industry classification. The Measurement 01 Vertical Relation Fan and Lang (2000) innovatively use a methodology based on commodity flows from 10 tables to capture the vertical re1ation between industries, and then obtain a proxy for the degree of vertical integration; many scholars have widely followed this approach (Fan and Goyal, 2006; Kedia et al., 2008; Shenoy, 2009; Fan et al., 2009). First, for the acquiring companies, we look to the segment information in the CSMAR database for a f rm’s primary industry in which it has the largest sales; for the target companies we look for their industry information as disc10sed in the merger announcement. Afterwards, we construct飞asthe vertical relatedness coeff cient between the acquiring and target companies. The 2002 Chinese industry 10 table provided by the National Bureau of Statistics of China inc1udes each pair of 10 coeff cients among 122 industries; it explains mainly the input of industry i (A) needed to produce every one renminbi output of indus町j，and the input of industry j (A) n巳巳dedto produce every one renminbi output of industry i, for each pair of industries i and j. We may then employ the larger coeff cient to ca1culate the vertical relatedness coeffìcient飞betweenthe acquiring company and
32 Li and Wang industries i and j, wh ch s defined as V;; = Max例，AJ, so as to effect vely explain υυJI the potential vertical integration opportun ty between them (Fan and Goyal, 2006). To take the acqu s t on by Beijing Jingneng Thermal Power (stock code: 600578) of Inner Mongolia Yitai Coal in 2007 as an example, the acquiring company’s largest bus ness s power and thermal production and supply (i), which provides the target company with an input of renminbi for every 1 renminbi output of coal mining and dressing business (j)-that is, Aij = -and in retum, j provides an input of renminbi for every 1 renminbi of output of i一thatis, Aji = . We then obtain the coeff cient between the acquir ng enterpr se ndustry (i) and the target company industry (j)飞，wh ch s . Figure 1 presents the cumulative distribution of the vertical relatedness coeff cients of our M&A samples. We class fy the sample mergers accord ng to the M&A c1assif cat on of Fan and Goyal (2006). If we def ne the merger as vertical when the vert cal relatedness coefficient气betweenthe acquiring company and the target company exceeds 1 per cent, then the vert cal re1atedness coeffic ents of per cent of the sample are less than 1 per cent. Likewise, if we def ne the merger as vertical when the ve巾alrelatedness coefficient飞betweenthe acquiring company and the target company exceeds 5 per cent, then the vertical relatedness coeff cients of per cent of the sample are less than 5 per cent. Compared with the sample of Fan and Goyal (2006), a Iarger proportion of the merger sample has a higher vertical relatedness, which may be related to the Chinese emerging and transitional market environment characterised by market failure and weak contract enforcement. coz=GEm-0028642 苟言E20。.2 .4 .6 Vert cal Relatedness Coeff c ent Figure 1 Cumulative Probability Distribution of Vertical Relatedness Coefficients between Merging Companies Research Design To test the impact of vertical relations on the wea1th effects of mergers, this study foIIows the designs by Fan and Goyal (2006), Kedia et al. (200的，and Shenoy (2009) and selects f rm size, transact on size, Ieverage, growth, market concentration of products,
The Wealth Effects of Vertica1 Mergers 33 and industry regu1ation as the contro1 variab1es of the regression mode1, which is as follows: CARs;., =¯o+ ¯1PVC+ ¯PHC;.,+ ¯VHC;.,+ ¯4MB;.’_1+ ¯sSIZE, U23H1+ ¯VOL;.,+ Åj-E几1+ ¯gHERFS;,’_l + ¯/DIOSu+ ¯laA_l 6u+ ÅllLGO几+¯CGOV;.,+ ÅROA;.，_1十ÅREG121314+ Ål~PROTECT+ ¯YEAR+ ¯;., (1) 16In Mode1 (1), CARs are the abnorma1 returns of the acquirers during the announcement period, measured as the wealth effects of mergers (Lafontaine and Slade, 2007). Following Fan and Goya1 (200哟，Kedia et al. (2008), and Shenoy (2009), we def ne the wealth effects of mergers as the CARs of acquirers for the windows of [-1, +1], [-2, +2], [-5, +坷，and [-10，十10]trading days around the丑rstannouncement day, which are measured by the market mode1, where the卢parametersare estimated over the [-180,-30] event window re1ative to the f rst announcement day. Consistent with the c1assif cation scheme used by McGuckin et al. (1991), Fan and Goya1 (200哟，Kedia et al. (2008), and Shenoy (2009), we def ne a merger as vertically re1ated if the vertica1 re1atedness coeff cient (V;) exceeds 1 per cent or 5 per cent, after referring to the cumu1ative probability distribution of the vertica1 re1atedness between the acquiring and target companies in Figure 1. Following this c1assif cation, we identif)r p旧evertica1 mergers (PVC) as those mergers between f rms be10nging to different 10 industries but exhibiting a vertica1 relatedness over the 1 per cent or 5 per cent cut›off point. The variab1e PVC will take a value of 1 if the merged f rms are vertically related but belong to different 10 industries. We identif)r pure horizontal mergers (PHC) as those mergers that take place between f rms in the same 10 industry but exhibiting a vertical relatedness less than the 1 per cent or 5 per cent cut-off point. The variable PHC will take a value of 1 if the merged f rms are vertically unrelated and in the same 10 industry. We c1assify mergers between f rms in the same 10 industry that also have a vertical relatedness over the 1 per cent or 5 per cent cut-off point as mixed vertical›and-horizontal mergers (VHC). The variable VHC will take a value of 1 if the merged f rms are vertically related and belong to the same 10 industry. Finally, we c1assify merger samp1es other than PVC, PHC, or VHC as conglomerate mergers According to existing literature, the characteristics of f rms and M&A transactions will influence the wealth effects of mergers. Thus, in view of the actual conditions of the merger activities of Chinese listed f rms, we use control variables as follows: the characteristics of f rms inc1ude f rm size (SIZE), debt levels (LEη， growth (GROWT.町，market concentration of products (AHl), heterogeneous expectations (IDIOS), ownership balance (HERFS), return on assets (ROA), local govem
34 Li and Wang confident (Roll, 1986), and the easier it wi11 be to pay an M&A premium (Moeller et al., 2004). According to the free cash flow hypothesis, if the acquirer has a large sum of cash flow and a low leverage level, it will be able to pay a premium more easily (Jensen, 1986). Corporate growth will influence the performance of mergers, and the pursuit of growth will drive firms to implement strategic mergers; aimless expansion, however, w 11 destroy firm value. Thus, the better the corporate govemance of the acquirer, the more value-creating mergers it will choose (Servaes, 1991). The ability of the acquirer to influence its target will also vary with different merger sizes, which in tum will influence the wea1th effects gained by the acquirer (Fuller et al., 2002). If the M&A transaction between the acquirer and the target is considered a vertical merger, then this merger may change the acquirer’s market power and market share and will create value in an uncompetitive market (Kedia et al., 2008; Shenoy, 2009). Hence, we add the Herfindahl index (AHI) of the acquirer’s industry to re丑ectthe competition level of its product market, and we expect the sign to be positive. In addition, the nature of the property rights of the u1timate controller will influence the motive and the ability of the acquirer, and w 11 further influence the wealth effects of the merger. In this paper we thus introduce two types of property rights nature -local govemment-controlled and central govemment-controlled -to test their impact on the wealth effects of mergers. At the same time, the acquirer’s stock p巳rformanceis not only related to the firm’s decision-making behaviour, but also has an important linkage to investor confidence and expectations. Under the condition that the arbitrage mechanism is restricted, the heterogeneous expectations of investors will result n over-estimating the acquirer’s stock price owing to ex ante information asymmetry. Because there is a lag in market reaction to a wrong stock price, it takes time for the price to resume reflecting the fundamentals. And because losses of long-term investors may be caused by previous wrong pricing, their heterogeneous expectations (IDIOS) may negative1y affect the CARs of the acquirer within a certain number of trading days before and after the announcement day (Lai et al., 2006). Also, since govemmental regulation changes the difference between market transaction costs and intemal organisational co
The Wealth Effects of Vertical Mergers 3~ Table 2 Variable Definitions Variable Variable definition Expectation CARs Cumulative abnormal retums of the acquirer for windows around the first announcement day VC Takes the value of 1 if the merged firms are vertically related at the 1 per cent or 5 per cent cut-off point, and otherwise 0 + PVC Takes the value of 1 if the merged firms are vertically related with飞higherthan the 1 per cent or 5 per cent cut-off point but belong to different 10 industries, and otherwise 0 + PHC Takes the value of 1 if the merged firms are vertically unrelated with飞lowerthan the 1 per cent or 5 per cent cut-off point and belong to the same 10 industry, and otherwise 0 + VHC Takes the value of 1 if the merged firms are V町ticallyrelated with ~; higher than the 1 per cent or 5 per cent cut-off point and belong to the same 10 industry, and otherwise 0 + LPROTECT Index for lower regional property protection, which takes the value of 1 if the minimums of both the property protection indices of the acquirer and the target are under the 40th percentile (匀，and otherwise o. The property protection indices are taken from the investment environment data of 120 cities in China investigated by the W orld Bank. + SIZE The natural logarithm of the acquirer’s total assets one year before the merger LEV The leverage ratio of the acquirer one year before the merger + GROWTH The market-book ratio of the acquirer one year before the merger + AHI The industry Herfindahl index of the acquirer computed according to the three-tier c1assification provided in the Guidelines on Industry Classificαtion 01 Listed Companies issued by the China Securities Regulatory Commission + IDIOS The square of regr巳ssionresiduals from 180 trading days before to 30 trading days before the first announcement of the merger computed by the market model HERFS The sum of squares of prop rtions of 十shares held by the five largest shareholders
36 Li and Wang Variable Variable definition Expectation ROA Retum on assets of the acquirer one year before the merger + VOL The shareholding proportion acquired 企omthe target company + LGOV Takes the value of 1 when the acquiring firm is control1ed by the local govemment, and otherwise 0 CGOV Takes the value of 1 when the acquiring firm is control1ed by the central govemment, and otherwise 0 REG Takes the value of 1 when the target firm belongs to an industry regulated by the govemment (coal, steel, oil, gas, water supply, metals, aviation, power, railway, etc.), and otherwise 0 + YEAR Takes the value of 1 when the merger takes place in 2007 or 2008, and otherwise 0 + IV. Empirical Results and Analysis Descriptive Statistics Table 3 presents the descriptive statistics. For the first announcement window of [-1, +1], [-2, +2], [-5, +坷，and [-10, + 10] trading days around the first announcement day published by the successful acquirer, the mean values of the CARs gained by the acquirer are , , , and , respective1y. This indicates that a listed firm’s acquisition of a non-1isted firm leads to positive CARs in China, which is consistent with the findings of Li and Chen (2002), and is the same as the conclusions of Ful1er et al. (2002)，如foel1eret al. (2004), and Faccio et αl. (2006), who study listed firms’ acquisition of non-listed firms in the US. Taking mergers with a vertical relatedness coef丑cientexceeding 1 per cent (5 per cent) between the acquiring and target companies as the classifying standard, we find that vertical mergers account for per cent ( per cent) of the ful1 sample, ful1y horizontal mergers per cent ( per cent), ful1y vertical mergers per cent ( per cent), and mixed vertical-and›horizontal mergers per cent ( per cent). This means that compared with a developed market with effective legal and financial systems, vertical mergers in China tend to be pe凹asivein its emerging and transitional market characterised by market failure and weak contract enforcement since this kind of merger can save transaction costs for upstream and downstream industries. At the same time, the tests by group on the acquirers’ wealth effects show that the mean CARs of vertical mergers are higher than those of the ful1 sample. Moreover, after subdividing the samples by merger type, we find that the wealth effects of mixed vertical-and-horizontal mergers are larger than those of the other merger types. Panel C of Table 3 provides the Spearman correlation coefficient matrix, which shows that the acquirers experience, on average, a positive and
The Wealth Effects of Vertical Mergers 3 significant abnorrnal retum at least at the 1 per cent level in the CAR[斗，+1], CAR[-2, +2], CAR[-5, +坷，and CAR[ -10, +1 0] windows of the first merger announcement, and their correlation coefficient is at least . This is positively related to mixed vertical mergers at least at the 5 per cent significance leveI. In addition, per cent of merging firrns are controIled by local govemments, and 18 per cent by the central governn随时，so the latter number of samples are fewer than the forrner number of samples. The average (median) of transaction size (VOL) is 56 per cent (51 per cent), meaning that acquirers are Iikely to become controIling shareholders through vertical mergers. Table 3 Wealth Effects of the Sample Mergers by Type and Spearrnan Correlation Coefl且cientsMixed Pure Pure vertical-Full Vertical vertical horizontal horizontal Conglomera能sample mergers mergers mergers mergers mergers Panel A: 1% critical point CAR[-I,+1] *** *** ** * *** CAR[-2,+2] *** *** * *** CAR[-5,+5] ** *** *** CAR[-IO,+10] * * * *** * y N 377 259 148 8 111 110 Pane1 B: 5% critical point CAR[-I,+I] *** *** *** *** CAR[-2,+2] *** *** ** *** CAR[-5,+5] ** *** ** *** CAR[-IO,+10] * ** * *** * V N 377 167 77 29 90 181 Pane1 C: Spearman coefficients of main variables CAR[-I,+I] CAR[-2,+2] CAR[-5,+5] CAR[-lO,+lO] PVC PHC CAR[-I,+I] *** CAR[-2,+2] *** 材*CAR[-5,+5] *** CAR[-IO,+10] *** *** *** PVC PHC * ** VHC ** *** *** *** 四***帽*Note:The table reports the wealth effects of the sample mergers by type and Spearman correlation coefficients. Variables are defined in Table 2. * *气and*树representsignificance at the 10%, 5%, and 1% levels, respectively.
38 Li and Wang Regression Analysis w巳regardCAR[斗，+1] as也etabulated depend巳ntvariable, and examine the wealth effects of different M&A types by using Model (1) with common least-square regression analysis. Then we obtain the regression coeff cient of every variable. First, we run the heteroscedastic tests of White (1980) on all regression models and f nd that all results reject the null hypothesis that heteroscedasticity exists, showing that the test model has no heteroscedastic problem. We then look at the variance inflation factor (VIF) values of all independent variables in the model and f nd that these are less than 5, indicating that the model has no substantial multicollinearity problems. In sum, the specif cation of the test model is reasonable. Analysis of Hypothesis 1 Models (1) and (2) in Panel A of Table 4 show that in the full sample regression, when a vertical merger is def ned as a merger with the vertical relatedness coeff cients of the acquiring and target f rms exceeding 1 per cent, the regression coefficients of the vertical mergers (VC) are signif cantly positive. Moreover, when we further subdivide merger types according to whether the acquiring and target companies belong to the same industry, the regression coeff cients of mixed vertical-and-horizontal mergers (VHC) are signif cantly positive. Meanwhile, Models (1) and (2) in Panel B of Table 4 also show that when a vertical merger is def ned as a merger with the vertical relatedness coeff cients of the acquiring and target f rms exceeding 5 per cent, the regression coeff cients of the vertical mergers (VC), mixed vertical-and-horizontal mergers (VHC), and pure vertical mergers (PVC) are all signif cantly positive, meaning that compared with other merger types, vertical mergers will obtain CARs of at least per cent. Moreover, horizontal mergers with strong vertical relations will obtain CARs of at least p巳rcent, meaning that the closer the vertical relation between the transaction parties, the more signif cant the merger’s wealth effects. We also divide samples according to the ownership type of the ultimate controller, and f nd that only when the acquiring f rms are privately controlled are all regression coeff cients of the above three types of mergers signif cantly positive, while those of other types are not. Additionally, the regression coeff cients of the interaction terms between the group of low regional property protection and the three types of mergers are all positive and differ signif cantly from zero at least at the 10 per cent level, meaning that the wealth effects of vertical mergers between upstream and downstream industries may exist mainly in private listed firms, thereby supporting Hl. The results thus show that in present-day China, where laws and regulations are imperfect, law enforcement is poor, and social trust ineff cient, vertical integration serves as a substitute mechanism for protecting valuable relationship-specif c investment from opportunism, preventing potential holdups for each party to the transaction, saving high transaction costs, stabilising production and supply relations between upstream and downstream industries, and somewhat allevia
The Wealth Effects of Vert cal Mergers 3 Acemoglu et al., 2009).5 Meanwhile, compared with acquirers under the control of local and central govemments, privately controlled firms may face weaker property rights protection and contract enforcement, and thus will be more willing to protect and promote ex ante incentives for relationship-specific investments, prevent potential holdups, save high transaction costs, and avoid excessive regulatory burdens and political intervention through vertical mergers between upstream and downs位eamindustries (Fan et al., 2009; Acemoglu et al., 2009). This could accordingly improve the performance of privately controlled firms with higher vertical relations. Analysis 01 Hypothesis 2 Because prope向rprotection in different regions is relative1y stable in different years, we also use the property protection index of 120 Chinese cities reported by the World Bank as a proxy for local property protection; this allows us to measure the degree of protection with respect to a city rather than a province. The World Bank computes this index using investigative data on the investment environment in these 120 cities. We define samples where the minimum property protection indices of the locations of the acquiring and target firms are below the 40th percentile () as the group of 6 low regional property merger samples belong to this group, the index takes the value of 1, and otherwise O. We then use the interaction terms between the property protection index and the proxy variables (VC, VHC, PVC, and PHC) of merger types to test the above theoretical hypothesis. Models (1) and (2) in Panel A of Table 5 show that the regression coefficients of interaction terms between the group of lower regional property protection and vertical mergers (VC), mixed vertical-and-horizontal mergers (VHC), and pure vertical mergers (PVC) are all positive and significantly distant from zero at least at the 10 per cent level. This means that the wealth effects of 5 Of course, the merger samples may be selectively biased. Because the choice model of Heckman (1979) depends on a speci直cfunction to estimate indirectly the treatment effect for the choice of a vertical merger, it also requires an efficient instrumental variable; otherwise, the estimate will be biased. But it is hard to丑ndsuch a variable in reality, and at the same time f nd a match model based on properties but not dependent on a specific function, in order to estimate directly the treatrnent effect for choice of a vertical merger (Li and Prabhala, 2007). Therefore, in this paper we use propensity›score matching models to find samples matched with vertical merger samples in order to regress and analyse our model specifications. In addition, we also use the difference-in-difference model to regress and analyse these same specifications, taking the mean return on total assets (pretax pro且ts/ total assets) two years before and after the vertical merger as a dependent variable. These results support the conclusion that the vertical relatedness between transaction parties is positively related to the wealth effects of the merger. 6 About 15 per cent of the regions where merger samples are located were not included in the Wo rld Bank (2006) survey of 120 cities. Thus, to incr侃侃thesample size, we use the mean value of cities that have data in the same province to replace the other cities. One sample has neither regional survey data nor the provincial average, namely the Lhasa area samples (600773). But because Lhasa city belongs to the Central Western region of China, it is reasonable to consider it to be within the lower local property protection group. In fact, if we de且nethis group with a critical value of 30 per cent or SO per cent, the result remains qualitatively unchanged. When the degree of local prope町protectionis defined according to the development of market interrnediary organisations and the legal system environment as found in the China Marketisation lndex -A Report 01 Marketisation of China for 2006, our conclusion remains unchanged in general, with only the significance coefficient declining slightly
iO Li and Wang vertica1 mergers between upstream and downstream industries will increase in re1ation to the degree of decrease in 10ca1 property protection where both transaction parties 7 are 10cated, thus supporting , under an environment with imperfect 1aws and regu1ations, poor 1aw enforcement, and inefficient socia1 trust, vertica1 integration serves as a substitute mechanism to protect va1uab1e re1ationship-specific investments from opportunism, save high transaction costs, and avoid excessive regu1atory burdens and politica1 intervention. In addition, the regression results of the contro1 variab1es in Tab1es 4 and 5 show that the regression coefficients of the yearly dummy variab1e YEAR are positive and significantly distant from zero. This means that as the infrastructure of capita1 markets improves, such as accounting standards and 1aws and regu1ations in capita1 market supervision, the extema1 govemance environment of the merger activities of listed firms a1so improves and strengthens, thus increasing the wealth effect of mergers. Moreover, AHI, which represents the product market concentration of the acquiring firms, is not significantly negative, meaning that the acquirers’ wealth effects do not come from the enhancement of market power. But although we have added con仕01variab1es common1y used in the literature in our mode1 specification, these variab1es are insignificant, which needs further examination. Robustness Test To make our results more robust, we conduct sensitivity tests on the results in Tab1es 4 and 5. First, we use the CARs of acquirers for windows of [-2, +2], [-5, +坷，and [-10, +10] trading days around the first announcement day as dependent variab1es to again estimate Mode1 (1) using ordinary least squares (OLS). Second, if managers have private information about potentia1 synergies of the merger activity, the OLS estimate will be biased (Eckbo et al., 1990); thus, we use the maximum-like1ihood estimate (MLE) procedure, which is based on the methodo1ogy in Eckbo et al. (1990). And third, since we define the vertical re1atedness coefficient by the acquirer’s prime industry, but in fact there may a1so be a vertica1 re1ation between the acquirer’s secondary industry and the target company (Fan and Goya1, 2006), we combine the primary industry with the secondary industry of the acquirer to define the vertica1 re1atedness between the two transaction parties. To do so, we take the vertica1 re1atedness coefficient of 1 per cent or 5 per cent as the critica1 point to judge whether a vertica1 re1ation exists, and then run the estimation using OLS once again. In a word, the results of these sensitivity tests suggest that all important conclusions are qua1itative1y unchanged. 7 When we use the difference in the average retums on total assets (pre阳pro且ts/ total assets) two years before and after the vertical merger announcement as the dependent variable, we find that although the regression coefficients of the tested variables of Hypothesis 2 confo口nto our expectations, they are not significant
1 The Wealth Effects of Vertical Mergers 4Table 4 Vertical Relations and M&A Perforrnance Panel A: 1% critical point of vertical coeff cients FuJl sample Sub-samples Local Central govemment govemment Privately (1) (2) controJled (3) controJled (4) con位。lled(5) VC * ** () () () () PVC * () () () () 晤。.021PHC () () () () VHC ** ** () () () () LPROTECT () () () () () () () () IDIOS * () () () () () () () () HERFS 唰() () () () () () () () VOL () () () () () () () () SIZE () () () () () () () () LEV * * () () () () () () () () ROA () () () () () () () () MB * () () () () () () () () LGOV () () CGOV * * () ( ) AHI * * () () () () () () () (幽)REG 国 * ** () () () () () () () () YEAR *丰** () () () () () () () () cons 。.041() () () () () () () () Adjusted R2 F ** ** ** ** N 377 377 176 176 68 68 133 133
42 Li and Wang Panel B: 5% critical point of vertical coefficients Full sample Sub-samples Local Central government government Private\y (2) con柱。lled(3) controlled (4) controlled (5) (1) \5材*VC 忡* \ () () () () PVC ** \6 ** () () () () PHC () () () () VHC *** 。.*** () () () () LPROTECT () ( ) () () () () () () IDIOS () (-1川)() () () () () () HERFS 。. () () () () () () () () VOL () () () () () () () () SIZE () () () () () () () () LEV * * () () () () () () () () ROA (0旦)() () () () () () () MB 。. () () () () () () () () LGOV () () CGOV * * (1刀)() AHI * * () () ( 0) () () () () () REG 。. () () () () () () () () YEAR ** ** 。. 。. (2.\5) (2. \3) () () () () () ()
The Wealth Effects of Vertica1 Mergers 43 Full samp1e Sub-samp1es Loca1 Centra1 govemment govemment Private1y (2) controlled (3) controlled (4) controlled (5) (1) CONS () () () () () () () () 2 Adjusted 。. 。. F *** *** * *** 忡*N 377 377 176 176 68 68 133 133 Note:The table reports OLS regression results. Variables are defined in Table 2飞*气and*** represent significance at the 10%, 5%, and 1% levels, respectively, using two-tailed tests. Huber-White robust standard errors are used to control for heteroscedasticity. Table 5 Property Protection, Vertica1 Re1ations, and M&A Performance Panel A: 1% critica1 point of vertica1 coefficients Full samp1e Sub-samp1es Loca1 Centra1 govemment govemment Private1y (2) controlled (3) controlled (4) controlled (5) (1) VC * 。. () () () () VC*LPROTECT ** * ** () () () () PVC * () () () () PVC*LPROTECT *** ** 忡() () () () PHC () () () () PHC*LPROTECT () () VHC * * () () () () VHC*LPROTECT * * () () () () LPROTECT 回 () () () () () () () () Contro1 Yes Yes Yes Yes Yes Yes Yes Yes A司justedR2 。.054。. F *** *** * * ** ** N 377 377 176 176 68 68 133 133
44 Li and Wang Panel B: 5% critical point of vertical coefficients Full sample Sub-samples Local Central government govemment Privately (2) controlled (3) controlled (4) controlled (5) (1) VC * *料() () (1肘)() VC*LPROTECT ** * () () () () PVC 。. () () () () PVC*LPROTECT * ** () () () () PHC 。.016() () () () PHC*LPROTECT () () () () VHC * ** ( ) () () () VHC*LPROTECT 。. * () () (1且)() LPROTECT () () () () () () () () Control Yes Yes Yes Yes Yes Yes Yes Yes 2 Adjusted F *** * ** * *** *** N 377 377 176 176 68 68 133 133 Note:The table reports OLS regression results. Variables are defined in Table 2飞**，and *** represent significance at the 10%, 5%, and 1% levels, respectively, using two-tailed tests. Huber-White robust standard errors are used to control for heteroscedasticity v. Conclusion Compared with developed markets that e叮oyeffective legal and financial systems, China’s and transitional market environment is characterised by market failure and weak contract enforcement; therefore, vertical integration should be better able to protect the ex ante investment incentives of relationship-specific assets and prevent potential holdups for the transaction parties. Using a sample of domestic mergers with non-related-equity acquirers and private target firms from 1998 to 2008, we find that the c10ser the vertical relation between M&A transaction parties, the higher the
The Wealth Effects of Vertical Mergers 45 wealth effects of a merger. Additionally, the wealth effects of vertical mergers increase as regional protection of property rights decreases, but these findings exist only for privately controlled listed acquirers. The results thus indicate that in present-day China, with imperfect laws and regulations, poor law enforcement, and inefficient social trust, vertical integration serves as a substitute mechanism to protect valuable, relationship›specific investments from opportunism, prevent potential holdups for each party of the transaction, save high transaction costs, and avoid excessive regulatory burdens and political intervention, especially for privately controlled listed companies. In this paper, we discover that vertical mergers between upstream and downstream industries are an efficient way to al10cate resources. What exactly, then, is the intrinsic motive driving the wealth effects of vertical mergers of listed firms in China? China’s emerging and transitional economic features make the motives for vertical mergers by listed firms more complicated, and they also have more policy implications. Therefore, in future studies we wi11 continue to use commodity flow in 10 tables to find competitors to the vertical merging parties, upstream suppliers, and downstream c1ients, as well as utilise relevant theories such as ef自ciency，forec1osure, and collusion rationales in reference to the Chinese background, in order to establish systematic motives for the vertical mergers of listed firms in China. References Please refer to pp. 19-21.
