2002年 9月 第 4卷 第 3期
中 国 会 计 与 财 务 研 究
C h i n 8 A c c O U n t i n g 8 n d F i n 8 n c e R e v i e w
Volume 4,Number 3,September 2002
非流通股释出对股价的影响 :
来 自转配股上市的证据 1
顾勇 吴冲锋 冷大丽 2
摘要
目前 ,不可流通的国有股和法人股逐步上市流通已经成为中国资本市场发展
的必然选择 。本文以 2000年 3月至 2001年4月不可流通的转配股上市流通作为
研究样本 ,运用数理统计方法 ,研究了转配股上市对股票价格的影响 ,并分析了
影响股价变动的相关因素 。实证研究结果表明 :转配股上市这一事件在短期内尉
殴票价格存在着 著的消极影响 :由于中国股票市场相对缺乏效率 ,在转配股上
市后股价继续消极反廊的延续时间要长于国外大宗股权交易后股价反应的延续时
:在转配股上市公告 日与转配股实际上市 日之间 ,股票价格存在明 的反弹现
象 :转配股上市前后股票价格的消极反应程度与市净率指标呈现明 的显著负相
关性 ,而与转配股占流通股的比例无显著关系 ;公司股价的消极反应程度并不随
公司业绩的提高而减小 。
关键词 :股价反应 、转配股 、上市流通 、事件研究 、异常收益率
本研究得到同家杰出青年科学基金 “金融工程与金融复杂性”(项 目编号 :
70025303)和教育部跨世纪人才基金 “金融系统的价格行为”的项目资助 。
作者要特别感谢香港理工大学中国会计与金融研究中心主任陈工孟博士 ,他对
本研究的选题与研究方法提出了许多有益的建议 ;同时还要感谢该中心的梁民
咏小姐为本研究提供了相关国外文献 。感谢上海交通大学管理学院的刘海龙博
士 、秦学志博士 、马春林博士 、李核硕士 ,他们对本研究提出过许多中肯的建
议 。作者感谢 《中国会计与财务研究》杂志的匿名审稿人对本文提出的修改意
见 。本文的一切疏漏和可能的错误均由作者负责 。
顾勇 ,硕士 ,富国基金管理有限公司 ,上海市南京西路 1 565号 (200040) 。
尖冲锋 ,教授 ;冷大丽 ,上海交通大学现代金融研究中心主任助理 ;上海交通
人学管理学 院 ,上海市 法华镇 路 535号 (200052) 。
维普资讯
40
一 、 研究背景
顾 勇 吴冲锋 冷大丽
目前 中国的上市公司股票一般可以分为两种类型 :可流通股份和不可流
通股份 。可流通股一般是指个人公众股 ,而不可流通股则包括国有股和法人股
(2000年 4月以前 ,转配股也不可流通 ) ,至今 国有股和法人股仍然不可流
通 。这种现象的产生是由中国特殊的历史和国情决定的 。按照国际惯例 ,只要
是上市公司 ,其股票就只能在一个统一的市场中流通 ,且在股本结构中只有普
通股和优先股之分 ,而无可流通的个人公众股和不可流通的国有股和法人股之
分 。
随着中国证券市场的逐步规范和与国际接轨 ,转配股 目前已经全部上市流
通 ,不可流通的国有股和法人股逐步上市流通也已经成为中国资本市场发展的
必然选择 ,自1999年起 ,国家有关部门就开始探索不可流通股份逐步上市流
通的有关方案了 。
l999年 l2月 ,中国嘉陵 (600877)和黔轮胎 (000589)率先进行了 “国
有股优先定向配售给其公众股股东”的试点3。从 2000年 3月到 2001年 4
月 ,原先不可流通 的转 配股逐步分批上市 。
所谓 “转配股”是指上市公 司在向全体股东增 资配股时 ,由于部分股东
(国有股和法人股股东)因缺乏资金或其它原因对自身应配的部分或全部股权
有偿转让给其他股东 (主要是社会公众股股东 )认购所形成的一部分股票 (楼
洪豪,1 997) 。在 2000年 4月 以前 ,转配股不能上市流通 。截止 2000年 3
月底 (转配股开始流通以前),具有转 配股历史遗 留问题的上市公司共有 170
家 ,占上市公 司总数的 1 5.6% ;转配股股份总额近 35亿股 ,占流通 A股 比例
, 中国嘉陵按 10配 8.361股的比例将 1亿股国有股优先定向配售给其公众股股
东 ,减持 比例为总股本的 21.10% ;黔轮胎按 10配 1.59097股的比例将 1710
万股国有股定向优先配售给其公众股股东 ,减持比例为总股本的6.75% 。减持
后两公司的国有股股东持有该公司股份的比例分别由原来的 74.76% 和 57.3%
下降为 53.66% 和 51% 。
t 关于转配股问题 ,国家国有资产管理局[1 994]1 2号文件 《关于在上市公司送配
股时维护国家权益的紧急通知》规定 :“国家股股东单位只能在确有必要的情
况下 ,才可能在股东会上投票赞成配股 ,如果国家股股东没有能力认购配股 ,
则不应投票赞成配股”。任何情况下国家股持股单位不得随意放弃配股权 ,若
转让国家股配股权 ,宜采取协议转让的方式 。而中国证监会 1 994年 10月 27
日颁发的[1 994】161号文件 《上市公司办理配股申请和信息披露的具体规定》指
出 ,“配股权出让后 ,受让者由此增加的股份暂不上市流通 。1 994年已实施过
配股方案的上市公司 ,也按本规定执行 。”在这之后 ,国家有关部门又多次重
申有关转配股暂不上市流通的政策要求 。这样 ,面对着强烈的资金需求且符合
配股条件的上市公司 ,无力认购配股的国家股和法人股股东便采取有偿转让配
股权的方式将配股权转让给社会公众 (许圣道和耿志民 ,1 997) 。1998年 ,
国家停止了转配股的做法 ,以防止这一历史遗留问题不断扩大 。
维普资讯
{I:流通股释出对股价的影响 :来 自转 配股上市的证据 4
的 4% 。从转配股发生的}耐间上看 ,1 994年发生 1 5家 ,1 995年至 1 997年
问发生 132家 ,1998年发生 23家 。这些上市公 司发生转配股时配股价普遍
较低 ,配股价最高的是东大阿派 (600718),每股 14元 ;最低的是 sT粤海
发 一每股 1.50元 。其中配股价 10元以上的有 3家 ,6—1 0元的有 27家 ,4—
6元的有 76家 ,其余配股 价均在 4元以下 ,转让费一般每股 0.2元以下 f易
晓明,2000) 。
2001年 6月 ,国务院发布了 《减持国有股筹集社会保障基金管理暂行办
法》s,规定 “减持国有股 (包括 国家股和国有法人股)是指 向社会公众及证
券投资基金等公共投资者转让上市公司 (包括拟上市公司)国有股的行为”,
并且规定以后“国有股减持主要采用存量发行的方式”。至此 ,不叮流通的国
有股的减持全面启功 。
不可流通股份的上市流通 ,必将增加原先流通股市场的股票供给 ,影响市
场信心 , 股票价格带来冲击 。正是出于不可流通股份“释出 刈二级市场冲
击方面的考虑 ,监管部门对于国有股减持方案非常慎重 。投资者也十分关注国
有股 、法人股的流通问题 ,并以此作为投资决策的重要依据 。本文将以转配股
j-市的数据为研究样本 , 不可流通股份的上市流通jc寸股票市场价格的影响程
度进行初步的实证研究 ,以期为今后对国有股上市流通的研究及政策制定提供
部分参 考资料 。
之所以选择转 配股作为研究样本 ,不仅 因为转配股原本是 非流通股的一
种 ,其 市流通与国有股和法人股上市流通十分类似 ,还 因为截止 2001年 5
月 ,所有转配股已经全部上市 ,其样本容量适合作实证研究 。
本文的逻辑结构如下 :第一部分介绍研究背景 ,简要叙述 了非流通释 出的
制度背景 ;第二部分对国内外相关研究成果作了文献回顾 ;第三部分给出了本
文的研究思路与方法 ;第四部分是实证研究的数据说明及其描述性统计 ;第五
部分给出了实证研究的结果 ,并X,l~g进行了分析 ;第六部分是研究结论与研究
局 限性 。
二 、文献回顾
由于国外证券市场不存在个人公众股 、国有股和法人股之分 ,因此不存在
“固有股减持 ”这种说法 。而且 ·西方国家国有企业的私有化 (Privati ation)
一 般指直接出售国有企业或者让企业作首次公开发行 (IPO)·因此不存在对
股票价格 的冲击 。
s 但是 ·由于该 《办法》实施后 3个 月『大J,沪深股市 大幅下挫 30% 多 ,证临会 不
得不在 2001年 10月暂停了该 《办法》的实施 。
维普资讯
42 顾 勇 吴冲锋 冷大丽
从不可流通股权上市流通的角度来看 ,国有股减持类似于国外的 “大宗
股权交 易 (Large BIock Trading)” 。在解释何为大宗股权交易时 ,Kraus和
Stoll(1972)认为当一次交易的股票数量大于市场所能消化的股票时 ,即可认
为该 次股票交易是大宗股权交易 。而 Aitken和 Frino(1996)则更加明确地指
出 ,当一次股票交易的数量大于该股票在市场上已经流通股份的1%时 ,则该
股权交易为大宗股权交易 。Kraus和 Stoll(1972) 、Holthausen ,Leftwich
和 Mayers(1987,1990)对纽约股票交易所 (NYSE)发生的大宗股权交易对
股价的消极影响作 了检验 ,均发现这种影响在统计上是非常显著的 。Keim和
Madhavan(1996)也对纽约股票交易所 (NYSE)、美国股票交易所 (AMEX)
和纳斯达克的全国市场 (NASDAQ NMS)发生的大宗股权交易对股价的影响
进行了检验 ,发现影 响显著 。Aitken和 Frino(1996)在对澳大利亚股票交易
所 (ASE)发生的大宗股权交易对股价影响的实证研究中也发现了类似的结
论 。
另外 ,二次分销 (Secondary Distribution)与大宗股权交易也十分类似 。
所谓“二次分销”是指销售某些已经发行在外的股票 ,这些股票由于金额太大
而不在股票交易所交易 。二次分销通常在收盘以后进行 ,销售价格略低于或等
于当日的收盘价 。Mikkelson和Partch(1985)检验了二次分销对股票价格的
影响 ,发现二次分销对股票价格的影 响与大宗股权交易的影响十分类似 ,即二
次分销对个股收益率存在十分显著的消极影响 。
那 么为什 么大宗股权交 易和二次分销会 对股票价格产生如此显著的影响
呢 ?Kraus和 Stoll(1 972)认为主要有三方面的原因 : (1)信息因素 ,即大
宗股权交易的发生向投资者传递 了某种信息 ,而这种信息会改变投资者对股票
的期望收益 ; (2)投资者偏好因素 ,即对特定证券 ,投资者会有不同的偏
好 ; (3)短期流动性成本因素 。Holthausen ,Leftwich和 Mayers(1987)
认为流动性成本 、股票需求曲线缺乏弹性以及信息影响是导致大宗股权交易影
响股票价格的因素 。
对于大宗股权 交易对股价影响的显著性 检验一般采 用 “事件研 究方法
(Event Study)”。Peterson(1989)对于事件研究的方法作 了详细的阐述 。事
件研究的步骤大致如下 : (1)确定事件发生日 ; (2)选取事件区间 (Event
Period) ;(3)选取估计区间 (Estimation Period)用来估计股票的正常收益
率 ;(4)计算事件区间的异常收益率 (Abnormal Return)来检验事件对股票
价格的影响是否显著 。进行显著性检验可以用参数检验 (参见 Mikkelson和
Partch,1985;Peterson,1 989)和非参数检验 (参见 Corrado,1 989;Arnold,
1992) 。
国外学者对影响股价变动程度的相关因素进行分析时一般采用多元线性回
归分析法 。Mikkelson和 Partch(1985)认为二次分销所造成 的信息 因素和供
维普资讯
非流通股释出 股价的影响 :来 自转配股上市的 、J下据 43
求 素导致 了股票价格的异常变动 其 中信息因素用出售者的类 (是否是公
司管理层 )、二次分销的股票持有人在分销前后的持股 比例和出售股票 的市值
米度量 ;供求 因素用二次分销股票的数量 占发行在外总股数的比例来度量 。他
们的研究发现异常收益率与二次分销的类型 显著负相关 ,但 与二次分销的股
票的价值和数量的相关程度并不显著 。因此 Mikkelson和 Partch(1985)认为
在二次分销中 ,信息 因素是影响股票价格的主要因素 。Kraus和 Sto11(1972)
则发现股价变动效应与大宗股权交易释出股份的总量呈显著的相关关系 。
由于中国国『大】国有股减持才刚刚起步 ,大量的研究主要从定性分析入手 ,
非流通股上市流通对股 票价格的影响进行 实证 研究的并不多 。唐国正等
(1998)以 1998年 4月沪 、深两市 626只股票的平均价格为被解释变量 ,以
上年 (即 1997年 )的每股收益 、流通股 比例为解释变量 ,进行了回归分析 ,
以此考察股价对国有股 比例的反应 。随后 ,朱武祥和郭志江 (1999)在此基础
冉加入行业平均市盈率作为解释变量进行回归分析 ,上述研究均发现股票价
格与流通股 比例呈负相关关系 。唐国正等 (1998)和朱武祥等 (1999)的上述
定量研究均是静态分析 ,只检验了股票价格与流通股 比例之间的关系 ,而没有
柃验释放非流通股对股票价格的影响 。
三 、研究设计
本文的研究主要采取事件研究方法 ,对非流通的转配股“释出”对股票价
格的影响进行实证研究 ,同时还将采用回归方法柃验有哪些因素会影响 “释
出”对股价的影响程度 。
(一)股票价格被影响程度的度量
Kraus和 Stoll(1972)将股票的被影响程度 (或股票对事件的反应程度
定义为
- )一-n(砉) ㈩
其中 、D
,
、 分别定义为第 日的股票价格 、股票股利和市场指数 。
从 20世纪 80年代起 ,随着事件研究方法的 日益成熟 ,度量股票价格的
被影响程度也逐步有 _r一套比较完善的方法 。
首先利用 CAPM 模型来估 计股票 的正常收益率 (N。rmaI R。tu ) ,即
R ,=&,+卢 R ,+£,, (2)
:次分销的股票出售者是否为公司的管理层 ?如果是管理层 则该哑变量取
维普资讯
44 顾 勇 吴冲锋 冷大丽
其中 ,为 f日 股票的 日收益率 ,R ,为 f日市场指数的 日收益率 ,£ 为随
机误差项 ,服从标准正态分布 。利用估计区间的个股与市场指数的数据 ,采用
最小二乘法估计出参数 和 ;然后利用市场指数在事件区间的 日收益率与
估计出的参数计算出每个股票的正常收益率 。在事件 区间的个股异常收益率
(Abnormal Return ,简称 AR)为实际 日收益率减去正常收益率 ,即
AR = ,一(&,+声,R ) (3)
在事件区间第 t日的 Ⅳ 个股票 的平均异常收益率 (Average Abnormal
Return ,简称 AAR)为
AAR =吉∑ , (4) r^●一 、 ,
” ,=1
除了考察一个交易 日股价被影响的程度 ,我们还将考察在事件区间一段时
间内股价的累计被影响程度 ,这种累计股价被影响程度可以用累计异常收益率
(Cumulative Abnormal Return ,简称 CAR)来度量 ,即
D。是计算累计异常收益率 的起始 日 ,D:是终止 日 。在事件区间一段时间
I 的Ⅳ个股票的平均 累计异 常收益率 (Average Cumulative Abnormal Return,
简称 ACAR)为
,
=
1备N 132
(二)显著性检验
为了考察事件区间的股价被影响程度是否显著 ,必须对上述异常收益率和
累计异常收益率进行统计检验 ,在这里我们将同时采用参数检验和非参数检验
的方法进行显著性检验 。
1.参数检验
要对异 常收益率进行显著性参数检验 ,必须对有关统计量进行归一化处
理 ,在这里我们将采用 Mikkelson和 Partch (1985)的处理方法 。
首先对 日异常收益率进行归一化处理 ,即
SARi,=ARi?~Sh
露
∑
=
D
D
露
维普资讯
JI:流通收释H。对股价的影响 :米 自转配月殳h市的证据
其中
45
(8)
在 (8)式中 ,V÷为公司 在凹归模型 (2)中的回归残差的方差 ,ED
为估计 问的长度 (交易 日)、R ,为在事件区间第t日的市场指数 日收益率 ,
页 为估 计区问内的平均 市场指数 日收益率 。平均标 准异常收益率 (Ave rage
S candardized Abnormal Return ,简称 ASAR)为
尺,= 1 N
(9)
一 般可以假设异常收益率 (即回归残差)AR 服从正念分布 (Mikke Json
和 PartCh ,1985) ,那 么 SAR 服从 t分布 ,方差 为 ED/(ED一2)。根据 中心
极 限定理 ,在 AR ( =1 ,⋯ Ⅳ)相互独立的假设下 ,ASAR 服从方差为
『ED/(ED一2)(M 1的正态分布 。由于一般 ED 取值较大 , ED/(ED一2)接近于
1 ,因此 ASAR 的方差接近于 1/Ⅳ ,即 ASAR~N (0 ,1/Ⅳ) 。柃验异常收
益率是否显著的标 准统 计量可以设为 :
ZfA =√Ⅳ(ASAR,) (10)
在这里z ~N(O,1)。同样地 对于检验累计异常收益率是否显著 ,我们
也可以构建标准统计量 :
z 一 。。。。。。。。。。。。。。。。。。。。。。。。。。。。。。。一 4D
2 一 Dl+1
∑z (11)
,=D
D2
由_丁: ~N(0,1) ,贝IJ∑z ~N(0,D 一D。+1),因此一Z (.AR ~N(0,1)。
,=DI
构建完成上述 日异常收益率和累计异常收益率的标准统计量以后 ,在确定
显著性水平 (本文的显著性水平设为0.05)以后 ,就可以检验上述异常收益率
是否显著了 。
2.非参数检验
为了进一步验证事件 间异常收益率的显著性
秩榆验 (Wilcoxon signed—rank ces c)(参见吴喜之 ,
秩检验 的步骤如下 :
, 我们采取 Wilcox0n符号
1999) 。Wilcoxon符 号
嚣
一l肋
维普资讯
46 顾勇 吴冲锋 冷大丽
(1)对于 =1,⋯ ,Ⅳ ,计算 I 月.,I。
(2)对集合 ={IAR川 月 I≠0}中的元素排序 ,并找出它们各自的秩;
如果有相 同个样本点 ,则每个样本点取平均秩 。
(3)g-u,/『等于 月,,>0 03 I 月,,I的R03和 , _等于 月 <0的I 月 I
的秩的和 。
(4)对于双边检验 ,在零假设下 , +与 应该相差不多 ,因而当其中
之一很小时 ,应怀疑零假设 ;在此 ,取检验统 计量 W =min( +, )。类似
地 ,对于单边检验 ,可以取 = 或 W=W-。
(5)当样本容量较小时 (N<30) ,可以查 Wilcoxon符 号秩检验的分布
表进行检验 ,当 Ⅳ 较大时 ,则要用正态分 布近似 ,构建大样本渐近正态统 计
量 ,其公式为
Z 。 W—N(N+1)/4
N【N+1)(2N+1)/24 (12)
在这里Znp—N (0,1)。在确定显著性水平 (本文的显著性水平设为0.05)
以后 ,就可以检验上述异常收益率是否显著了 。
(三)影响个股价格变动程度的潜在相关因素分析
非流通股的释出会对股票价格的波动产生影响 ,影响的程度用累计异常收
益率来度量 。不同股票的累计异常收益率各不相同 ,那么会有哪些因素影响个
股的累计异常收益率呢 ?考虑在本文第二部分 “文献回顾”中所提到 的影响异
常收益率的各个影响因素 ,以及鉴于国内的非流通股释出与国外的大宗股权交
易或者二次分销在成因和操作过程 中的差异性 ,从中国实际国情和为非流通股
释出提供政策依据的角度出发 ,我们提出了下面三个影响非流通股释出股价反
应的因素 。
1.股票供给因素
如果投资者进行投资决策时某一公司的股票不能被其它证券完全替代 的
话 ,那 么投资者对股票的需求就不具有完全的弹性 (Perlectly Elastic),也就
是股票的需求曲线向右下方倾斜 ,因此股票供给的增加将导致供给曲线沿着需
求曲线 向右移动 ,股票的均衡价格将下降 (Mikkelson和 Partch ,1985) 。
由于转配股的上市导致股票的供给增加 ,供给曲线向右下方移动 ,同时由于股
票的需求并不具有完全弹性 ,因此导致股票价格对转配股上市的消极反应 。释
出的非流通股的相对数量越多 ,股票供给曲线向右移动的幅度越大 ,股票价格
的变动也就越大 。因此 ,要检验股票供给因素对非流通股释出时股价变动效应
是否显著 ,需要检验下列命题 1是否成立 。
命题 1:待释 出的转配股 占该股票已经流通在外的总股份的比例越 高 ,那
么对于转配股释 出 ,该股票价格的消极反应程度越 高 。
维普资讯
非流通股释 出对股价的影响 :米 自转 配IJ殳l 市的证据 47
如果上述命题 1成立 ,则说明股票的供求关系是引起非流通股释 出的价
格消极反廊的一个重要 因素 。
2.公司盈利能力因素
同前许多探讨国有股流通的方案认为国有股流通首先应从绩优公司开始 ,
即昕谓“靓女先嫁”。他们认为如果公司的业绩越好 ,那 么投资者对公司的股
票越欢迎 ,越乐下接受这些释出的股票 ,因此非流通股释出对股价的消极影响
越小 ;反之 ,如果公司业绩较糟 ,那 么非流通股释出对股价的消极影响就越
夫 。要检验转配股释出的实际情况是否如此 ,我们需要检验下列命题2是否成
。
命题2:待释出转配股的上市公司业绩越好 ,那么对 于转配股释 出 ,该股
票价格 的消极反应程度越小 。
在以后的实证检验 中 ,我f『J将 用 1999年的净资产收益率 (ROE)来衡量
上市公司的业绩 。之所以选择 1999年的净资产收益率 ,主要是考虑转配股主
要 发生在 2000年 4月以后和 2001年 3月以前 ,在这一段时间内大多数转配
股公 司的 2000年年报还没有公布 ,因此 1 999年年报 中披露 的公 司收益信息
是市场对上市公 司业绩度量的主要依据 。
3.价值因素
一 般而言 ,上市公 司相对其股价的 内在投 资价值是吸引投资者的重要 因
素 。非流通股释出时 ,如果释出非流通股的股价高于公司内在投资价值的程度
不是很大的话 ,那 么股价对此的消极反应程度也就相对较小 ;反之 ,如果释 出
的股价远远高于公司的内在投资价值 ,则投资者就会“用脚投票”,股价的消
极反应程度就会很大 。在 中国.L市公司的资产评估 中 ,主要用公司净资产反映
公司的内在价值 ;同时 ,投资者对每股净资产这一指标也具有相 的认同度 ,
因 每股净资产相对于净资产收益率 ( 0E)等盈利性指标 而言 、具有较好的
稳定性 。因此当股价和每股净资产的接近程度越高 ,投资者越愿意接受释出的
非流通股 ,从而 非流通股释出导致的股价的消极反应程度也就越小 。
为了进一步研究价值因素 股价的影响 ,我f『J引入市净率 P/BV (股价 /
每股净资产)指标 ,考察股价的累计异常收益率与市净率之问的关系 。如果
市净率越小 ,即股票价格与每股净资产的差距就越小 ,则投资者越愿意接受该
’ 在美罔 ,市净率的高低是被用米判断一个股票是否是价值型股票的重要标志 。
比如美同的投资评级公司 —— 晨星 (MorningStar)公司就把共同基金的股票
投资组合的平均市净率作为判断该基金是价值型基金还是成长型基金的重要
杯准 。
维普资讯
48 顾 勇 吴冲锋 冷大丽
释出股票 ,因此股价对非流通股释出的消极反应程度就越低 。对此 ,可以进一
步检验下列命题 3。
命题 3:如果股票的市净率越小 ,则股价的消极反应程度就越小 。
为了计算市净率 ,我们取转 配股上市公告 日前 20日的平均收盘股价作为
分子 ,用 1999年年报披露的每股净资产作为分母 。选择 1999年年报 中的净
资产作为分母 ,也是 因为在转 配股上市这一段时间内 ,大多数转配股公 司的
2000年年报还没有公布 ,1999年年报中披露的每股净资产数值是市场对上市
公司价值度量的主要依据 。
在以后的分析中 ,我们将采用多元线性回归技术 ,将累计异常收益率
CAR⋯ 、作为被解释变量 ,度量股价对转配股释出的消极反应程度 ;上述三个
命题中的分析指标——转配股占已上市流通股的比例(RATIO)、待上市转配
股的 1999年的净资产收益率 (ROE)和市净率 的倒数—— 作为解释变量 。据
此 ,我们将构造下面的回归方程 :
1
尺 =al+y】】 +y12ROE+71 +£1 ( 3)
通过对各个解释变量进行显著性检验 ,可以发现有哪些变量对个股的累计
异常收益率会有显著的影 响 。
四 、数据说明和样本选取
1.数据说明
本文数据来源于路透信息系统 、《上海证券报》 、《中国证券报》和证券
之星网站 。其中 ,股票日交易数据取自上海交通大学路透金融工作室的路透信
息系统 ;转 配股上市股数 、公告 日期和上市 日期取 自上市公 司在 《上海证券
报》和 《中国证券报 》的转配股上市公告 ;净资产收益率和流通股数量直接
取 自 1999年年报 (对于在 2000年年初至转配股上市公告 日之间有送股 、配
股及增发新股的 ,实际股数根据相关公告进行调整 )。有关变量的定义及其取
值如表 1所示 。
表 1中 口值以转配股公告日前第 1 50日至公告 日前第 30个交易日之间
共 120个交 易 日的数据为基础 (计算期 间发生送股 、配股的股票均进行 了复权
处理 ) ,根据市场模型进行计算 。在计算的过程 中 ,根据股票挂牌交易的市
场 ,分别选用 了上海 A股指数和深圳 A股指数作为市场基准组合 。在选择平
均股票价格时 ,为了排除 日期的不确定对分析结果可能产生的影响 ,本文采用
了转配股上市公告 日前 20日的平均收盘股价 。
维普资讯
非流通股释 出对股价 的影响 :来 自转 配股上市的证据
表 1 变量及取值说明表
49
2.样本选取
2000年 3月到 2001年 4月 ,沪深两地发 布转 配股上市 公告 的 A股上市
公 司共 有 167家 s,其 中沪市 99家 ,深市 68家 。 167家上市公 司的转配股
总数 为 34.67亿 股 ,占这些公 司流通股总数 的 20.7% 。
从表 2可 以看出 ,转配股 占流通股 比例小于 1% 的上市公司有 1 8家 ,共
计有转 配股 706.9万股 ,仅 占总转配股股数 的 0.20% 。同时 ,目前 中国股 票
市场 中大部分个股 的平均 日换手率 (日成交量 /流通股 总股 数 )高 于 1% 。由
于 Kraus和 Sto11(1972)认为大宗股权交 易的股票数量要 大于市场所能 消化
的股 票数量 ,本文剔 除了转 配股 占流通股 比例小于 1% 的 1 8个样 本 ,从上述
167只股 票 中仅选取 149只股票样 本 。另外 ,在这 149个股 票 中 ,PT 网点
(600833) 由于在转配股上市期 间 已经实行 了特 别股票转让 (PT)9,股 票 日
交易数据非常少 ,也将其剔除出样本 。因此 ,本文进行实证研究的样本容量共
计 148个股票 ,其中深市 52个 ,沪市 96个 。
上市公司在其转配股上市前要在 《中国证券报》或 《上海证券报》上发布
转 配股 上 市 公告 ,公 告 说 明转 配股 的上 市 数 量 和上 市 时 间 。公 告 日
s 另有 3家在 《中国证券报》和 《上海证券报》上没有查到相关公告 。
, PT是英文 “Particular Transfer”(意 “特别转让”)的缩写 ,是指当上市公司因
连续三年亏损等原因被暂停上市期间 ,交易所及相关会 员在每周五 (法定假 日
除外)为持有或欲购该公司原上市流通股的投资者提供的 “特别转让服务” ,
PT公司的股票只在每周五 (法定节假 日除外)开市时间内申报转让委托 ,收市
后对有效 申报按集合竞价方法进行撮合成交 。也就是说 PT公 司每周只撮合成
交一次 ,产生一个交易价格 。
维普资讯
50
表 2 l67家上市公司转配股占流通股比例统计
顾 勇 吴冲锋 冷大丽
转配股家数 18 29 18 32 22 15 10 17 6 167
转 股股 数 706 9 7377 1 1 5600 0 551 35.8 65764.5 47776.4 27736 1 93344 3 33250.4 346691.4
(力 股 )
ll』总的转 配 0 20 2.1 3 4.50 1 5 90 1 8.97 13.78 8 00 26.92 9 59 100
H殳比例(%)
表 3 样本中股价影响因素的描述性统计
(Announcement Date(AD))和上市 日 (Distribution Date(DD))之间间隔
的交易 日不等 ,间隔时间从 0到 4个交易 日不等 。在上述 148个样本 中 ,公
告 日和上市 日的间隔为 0天的有 26家 (占样本总数的 1 7.6%) ,间隔为 1天
的有 29家上市公司 (占样本总数的 19.6%) ,间隔为 2天的有 78家上市公
司 (占样本总数的 52.7%) ,其余的间隔为 3天或 3天以上 。
表 3 中的数据 为股 价影 响 因素的描述 性 统 计 。本 文的数 据处 理应 用
EXCEL工作表软件和 SPSS统计分析软件 。
五 、结果及分析
(一)异常收益率及其显著性检验
1.转配股上市公告日前和上市日后的统计结果及分析
由于每个转配股的公告 日 (AD)和上市 日 (DD)之间的问隔从 0日到 4
日不等 ,因此在对样本总体作统 计分析时 ,只能分别计算公告 日前25日以及
上市 日后 25日的股票异常收益率 ,并对其作显著性检验 。
对公告 日前25日的异常收益率及其参数和非参数显著性检验的结果列示
于表 4,对上市 日后 25日的异常收益率及其参 数和非参数显著性检验 的结果
列示于表 5 。
在表 4和表 5中 ,第 (1)列表示交易 日序列 , (AD—m)表示公告 日前
第 m 日 , (DD+m)表示上市 日后第 m 日 ;第 (2)列表示股票的平均异常
维普资讯
I 流通股释出对股价的影响 :来 自转配股 市的 据 5 l
a 零假设是半均 常收益率 0 。
, 假设是 一个交易 同中股票异常收益率 负的 比例等丁 O.5O 。
。 O.05的水 r_显著 。 在 0.01的水平娃著 。
收益率 ,由公式 (4)计算得到 :第 (3)列是 第 (2)列数据的显著性参数
柃验 ,检验值 Z由公式 (1 0)计算得到 ;第 (4)列表示样本股票中异常收益
率为负的比例 ;第 (5)列是对第 (4)列数据的显著性非参数检验 '检验值由
公式 (12)计算得到 。
从表4中可以看出 ·在公告 口 (AD)样本的平均异常收益率力一0.38% ,
维普资讯
52
表 5 上市日后 25日异常收益率及检验结果
颐 勇 吴冲锋 冷大丽
a 零假设是平均异常收益率为 0 。
b零假设是一个交易 日中股票异常收益率为负 的比例等于 0.50
c 住 0.05的水平显著 。 d在 0.01的水平显著 。
且在 0.05的水平上显著为负 ,同时在该 日 ,有 68%的股票的异常收益率小于
0 ,并也在 0.05的水平上显著成立 ,这说明个股对转配股上市这一信息作出
了明显的消极反应 。在转配股上市公告前 10个交易 日(AD一10,AD一1)内 ,
平均异常收益率均为负值 ,尽管只有在 (AD一10)日的参数检验在 0.05的水
平上显著为负 ,但是 非参数检验显示 ,在第 (AD一1)日 、(AD一2)日 、(AD—
维普资讯
{F流通股释出对股价的影响 :米 自转配f]殳』一市的证据 53
4)日 、 (AD一9)日和 (AD一10)日 ,异常收益率为负且在 0.05的水平上显
著 。在公告 日前第 11日到公告 日前第 25日 (AD一25 ,AD一11)内 ,有 9天
的平均异常收益率为负 ,且在 0.05的水平上均不显著为负 。上述这些现象说
明股票价格对转配股 t:市这一信息 的提前反应主要集 中在 (AD一10 ,AD一1)
这 10天中 ,而在此以前 的反应则并不明显(参见图 1)。这 种在公告 日前一段
时 I大I股 票价格 对转 配股上 市信 息的提 前 反应现 象 ,与 K r a u s和 S t o II
(1972) 、MikkeIson和 Partch(1985)对美国市场的统计结果一致 。对于这
种现象 ,一般用信息因素来解释 一即投资者对转配股上市这一事件十分关注 ,
他f『J会千方百计地对相关信息进行收集 、分析与传播 ,从而在转 配股上市公告
以前 一市场价格 已经部分反映了转配股上市这一信息 。
再来看转配股上市后的情 况 。在表 5中 ,转配股上市 日的平均异 常收益
率 一0.26% ,且在 0.05的水平上显著为负 。尽管在该 日有 5 5% 的股票的异
常收益率为负 ,但在 0.05的水平 上不显著 。这些仍然在一定程度上说 明了在
转配股上市当日 ,非流通股转 流通股而导致流通股股数增加对股票价格存在
一 定的消极影响 。在转配股上市后 7个交易 日 (DD+1 ,DD+7)内 ,在第
(DD+2)日和第 (DD+7)日 ,异常收益率为负且在 0.05的水平 上显著 ;而
非参数检验显示 ,在第 (DD+2)日 、第 (DD+4)日和第 (DD+7)日 ,异 常
收益率为负且在 0.05的水平上显著 。而在转 配股上市 日后第 8天到第 25天
(DD+9 ,DD+25)内 ,仅有 11天的平均异常收益率为负 ,且在进行参数检
验}i1f并不显著 。上述这些说 明在转配股实际上市以后股票价格的消极反应主要
集中在『DD+1,DD+71这 7天 中 、而在此以后的反应并不明显 (参见图 2)。
中国证券市场这种在转配股上市后较长一段 时间内股价的异常收益率仍然
负的情况与国外有关 “大宗股权交易”和 “二次分销”在上市交易 日后的情
况不一样 。Kraus和 Stoll(1972) 、Mikkelson和 Partch (1985) 、Aitken
和 Frino(1 996)的研究发现仅在事件发生 日后的 2到 3天 内 ,异常收益率存
在显著性 ,可见在中国的反应时间显然延长了 。这主要是 由于与美国等国外成
熟资本市场相比 ,中国的资本市场仍然相对缺乏效率 ,股价 某一事件的反应
没有象美国资本市场那 么迅速 。尽管转配股上市 日前的股价已经吸收了转配股
上 市公告的信息 ,但在上市 日后相当一段时间内 ,较大程度的股价消极 反应仍
是 公告信息的进一步加深 。
关于这一点也可以从上 市日后的累计异常收益率与公告 日前的累计异常收
益率存在显著的正相关关系看出 。以公告 日前 11日 (包括公告 日)的累计异
常收益率 CAR 为解释变量 ,以上市 日后 11日 (包括上市 日)的累计异
常收益率 CAR ⋯⋯、为被解释变量 一进行一元线性回归 ,得到
CARDD DD ∞=-0.0106+o.329CARAD_11_AD I14
(5.401)
维普资讯
54
图 1 公告Et前的Et平均异常收益率和累计异常收益率
(%)
o.5
o
- o.5
- 1
— 1.5
— 2
— 2.5
顾勇 吴冲锋 冷大丽
AD-30 AD-25 AD-20 AD-1 5 AD-1 0 AD-5 AD
U期 (U)
回归方程的 Adi—R2=0.231 ,回归方程的显著性程度 F=4 5.049 ,可见
CAR 。
一 。. 。
与 CAR。。
. 。 。 + 。 的线性 回归关 系非常显 著 ,CAR 。一 。 与
CAR⋯⋯ 显著正相关 。
表 6显示 了平均累计异常收益率的分布 。第 (1)列指示了交易 Et区间 ,
每个交易区间的前一项表示区间的起始 Et,后一项表示区间的终止 Et;第 (2)
列是平均累计异常收益率 ,由公式 (6)计算得到 ;第 (3)列是参数检验值 ,
由公式 (11)计算得到 。
在表 6中 , (AD一5,AD)的平均累计异常收益率为 一0.70% ,但在 0.05
的水平上不显著为负 。(DD ,DD+5)的平均累计异常收益率为一0.94% ,在
0.0 5的水平上显著 为负 。而在其余时 间区间 ,参数检验均不显著 。同时 ,
(DD ,DD+5)的平均累计异常收益率的绝对值和检验的显著性水平均大于
(AD一5 ,AD)的平均累计异常收益率的绝对值和检验的显著性水平 ,即在转
配股上市 Et后 6天 (包括上市 Et)的时间内 ,股价对信息进一步加深的消极反
应程度要大于公告Et前 6天 (包括公告 Et)股价对信息提前消极反应的程度 。
2.转配股公告 日与上市日之间的统计结果与分析
在前面对样本总体进行分析时 ,由于 AD 与 DD 之间的间隔时间并不统
一 , 间隔Et从 0到 4个交易 Et不等 ,因此我们无法对 AD与 DD的间隔Et进
行分析与检验 。在这 里 ,我们 将分别 计算 间隔 为 1个 交易 Et (子样本数 为
29) 、2个交易 Et(子样本数为 78)和 3个交易 Et(子样本数为 14)这三个
子样本 o的间隔 Et累计异常收益率 ,并进行统计检验 。
c 间隔 日为 3天以上的公司数仅为 1,不具有统计意义 ,这里不再检验 。
维普资讯
E流通股释出对股价的影响 :来 自转配股上市的证据
图 2 上市 日后的 日平均异常收益率和累计异常收益率
表 6 平均累计异常收益率及其 著性检验
DD+20 DD+25
U蟛J(U)
55
t(AD一5,AD)与 (DD 、DD+5)的区间长度为6个交易 日 、其余 为 5个交易 日 。
n零假设是累计异常收益率为 0 。
住 0.05的水半 著 。
AD 与 DD 间隔 日分别为 1个交易 日 、2个交易 日和 3个 交易 日的子样
本的累计异常收益率及其参数和非参数显著性检验的结果列示于表 7 。
在 AD与 DD问隔 日为 1天的 29个子样本 中 ,其 间隔 日平均异常收益率
0.48% ,但在 0.05水平上不显著 ;在该间隔 日仅有 45% 的股票的异常收
益率为负 ,在 非参数检验统计下不显著 。再来看 AD 与 DD 间隔 日为 2天的
异常收益 率情 况 (样 本数 =7 8) ,两 天 间隔期 的 累计平均 异常 收益率 为
0.77% ,且在 0.05的水平下显著为正 ;这两天累计异常收益率 为正的股票数
% 0 一 —
一 一 一
维普资讯
56 顾 勇 吴冲锋 冷大丽
量 占子样本数 的 56% ,且 非参数检验显示 ,正异常收益率所 占的比例在 0.05
的水平上显著 。对于 AD与 DD 间隔 3天的情况 (样本数 =14) ,三天的累
计平均异常收益率为一0.42% ,但在 0.05的水平上不显著 ;有 50%股票的累
计异常收益率为负 ,在 0.05的水平上也不显著 。考虑样本数量的因素 ,间隔
日为 1天和问隔 日为 3天的子样本均为小样本统计 ,尽管其结果的说服力并
不很强 ,但这 两个子样本的累计平均异常收益率并不显著为负 ;而间隔 日为2
天的样本数有 78个 ,且在 0.05的水平下累计平均异常收益率显著为正 。
如表 6所示 ,在转配股上市公告前 6天 (包括公告 日AD)的平均累计异
常收益率为 一0.70% (虽在 0.05的水平上不显著 ,但 在 0.1的水平上显著为
负),那么为什么在公告 日和上市 日之间的间隔 日会出现累计异常收益率为负
不显著甚至异常收益为正的反常情况呢 ?
对于这种在 AD与 DD之间的间隔 日内存在着异常收益率的显著反弹现
象 ,我们 认为有以下两个可能的原因 : (1)过度反应 。在公告 日前 (包括公
告 日),股票的价格对转配股上市这一信息存在过度反应 ,即股票价格被超调
(Over Shooting)了 ,因此在公告 日后股票价格超跌反弹 ; (2)市场主力机
构护盘 。转配股在形成过程中 ,主要掌握在机构投资者手中(如配股的承销券
商),因此转配股上市 ,必将解套这些机构手 中的股票 ,为 了获得高收益 ,这
些机构必然在上市 日以前维持股价 ,以便在转配股上市后以较高的价格抛售股
票 。
但是 ,在表 6中 ,转配股实际上市 日 DD 后 6天 (包括 DD)的累计异
常收益率在 0.05的水平下显著为负 ,这说明转配股上市对股价的影响在上市
日DD后仍然在继续深化 ,显然在公告日前股票价格对转配殴上市这一信息
并不存在过度反应 ,因此可以排除过度反应因素 ,即过度反应并不能解释AD
表 7 AD与 DD 之间的间隔 日的累计异常收益率及其显著性柃验
a 零假设是平均异常收益率为 0 。
零假设是一个交易 目中股票异常收益率为负的比例等于 0.50 。
在 0.05的水平显著 。
维普资讯
非流通股释出对股价的影响 :来 自转配股上市的证据 57
与 DD间隔日的异常收益率为正的现象 。由此可见 ,间隔日异常收益率为正
可能主要是 由于市场主力机构护盘所造成的 。
将上述有关转 配股上市公告 日与上市 日之间的分析结论应用于国有股退
出 ,我 可以认为 ,如果释出的国有股集中在一个或几个投资机构手中 ,再 由
这些机构按市价出售的话 ,那他们就有可能出现为了自身利益最大化而操纵市
场的情况 ,而且这种操纵在流通公告 日与上市 日之间的间隔期间内表现得更为
明 显 。
(二)影响股价变动程度的相关因素分析结果
表 8和表 9分别用最sJ,~-乘法对被解释变量累计异常收益率 (公告前 10
口 (AD一1 0 ,AD)与上市后 10日 (DD ,DD+1 0))与解释变量—— 转配
股 占流通股的比例 RATIO 、1999年净资产收益率 ROE和转配股所对应股
1
票的市净率倒数 —— 进行 了回归分析 。
』‘/D V
上述 两表 中的第 (1)列是 回归方程的截距 ,第 (2)列是转配股 占流通股
总数的比例的回归系数 ,第 (3)列是待释出转配股所对应的股票 1999年的
调整前的净资产收益率的回归系数 ,第 (4)列是市净率倒数的回归系数 。
表 10显示了多元线性回归方程的多重共线性容限度 (To】erance)11。表
1
10显示第三部分中的回归方程 (13)的解释变量 RA TIO 、ROE和 的
』‘/D V
容限度均在 0.9以上 ,因此可以认为回归方程不具有多重共线性 。
1.股票供给因素的检验
在表 8中 ,公告 日前的累计异常收益率和转配股 占流通股的比例 RA TIO
的一元线性 回归模型 (5)显示 ,RA TIO 的回归系数的 t检验值为 一0.7450 ,
在 0.05的水平上不显著 ;而在多元线性模型 (1) 、 (2)和 (3)中 ,RA TIO
的t检验值也十分的不显著 。同样地 ,在表 9中 ,上市 日后的累计异常收益率
和 RATIO 的一元线性 回归模型 (5)显示 ,RATIO 的系数的 t检验值为
一 0.9636 ,在 0.05的水平上不显著 ;在多元线性回归模型 (1)、 (2)和 (3)
中 , RA TIO 的 t检验值 同样也不 显著 。
由此可见 ,至少在统计上 ,待释出的转配股 占该股票已经流通在外 的总股
份的 比例 RA TIO与股票价格的消极反应程度的相关关系并不显著 ,命题 1不
容限度 (Tolerance)是以每个 自变量xi作为被解释变量对其他解释变量回归时
得到的余差比例 ,即 Tolerance
.
=1一 只:
. ,
。 容限度越小 ,说明xi与其他变量
的信息重复性越大 ,其对 Y的边际解释能力就越小 。一般当容限度小于 0.1,
便认 为这一变量与其他变量之间的多重共线性超过了容许的界限 (参见郭志
刚 , 1999) 。
维普资讯
58 顾 勇 吴冲锋 冷大丽
表 8 公告前 l0日 (AD2lO,AD)累计异常收益率的回归统计表
- 0.0451
(2.5246)。
一 0.0093
(-0.9943)
一 0.0418
(一2.4213)
一 0.0489
(-3.0122)
- 0.0091
(-0.9868)
一 0.0137
(一1.9306)
- 0.0459
(-2.91 16)
一 0.0001
(-0.5185)
一 0.0002
(-0.7264)
一 0.0001
(一0.5950)
一 0.0002
(-0.7450)
一 O.0094
(-0.7366)
一 0.0022
(-0.1739)
0.1723 0.0207 2.0223 0.1 135
(2.3382)
0.1 592
(2.2297)
- 0.0100 -0.1758
(-0.7937) (一2.4025)
- 0.0029
(一0.2324)
- 0.0099 0.2907 0.7482
0.0239 2.77 1 0 0.0660
0.0257 2.91 39 0.0575
- 0.003 1 0.5550 0.4575
- 0.0066 0.0540 0.8 1 66
0.1622 0.0282 5.21 13 0.0239
(2.2828)
a 对回归系数进行显著性榆验的 t统计值 。
b对回归方程进行拟合柃验 的 F统计值 。
c F统计检验的显著性概率 。
成立 ,因此在转配股释 出的情况下 ,股票供给因素并不是影响股票价格被影响
程度 的重要 因素 。
2.公司盈利能力因素的检验
在表 8中 ,公告 日前的累计异常收益率和 ROE的一元线性 回归模型 (6)
显示 ,ROE的回归系数的 t检验值 为 一0.2324 ,在 0.05的水平上不显著 ;
而在多元线性模型 (1) 、 (2)和 (4)中 ,ROE的 t检验值也十分不显著 。
同样地 ,在表 9中 ,上市 日后的累计异常收益率和 ROE的一元线性 回归模型
(6)显示 ,ROE的系数的 t检验值 为 0.6576 ,在 0.05的水平上不显著 ;在
多元线性 回归模型 (1) 、 (2)和 (4)中 , ROE的 t检验值同样也不显著 。
因此净资产收益率 ROE与股票价格的消极反应程度的相关关系并不显
著 ,命题2不成立 。在转配股的释出过程中 ,公司股价的消极反应程度并不随
公司业绩的提高而减小 ,可见公司业绩并不是影响非流通股释出所引起的股价
消极反应程度的重要因素 。笔者认为 ,中国证券市场作为新兴的资本市场 ,在
; ; )
维普资讯
_1}流通股释出对股价的影响 :来 自转配股上市的证据 59
会计信息披露方面还存在诸多问题 ,上市公司净资产收益率每年的波动较大 ,
ROEff~可信度存在很大的问题 ,投资者 把RoE作为投资决策关键依据的认
同度较低 ,这导致了命题 2的不成立 。因此关于国有股释出的昕谓 “靓女先
嫁”的说法 ,尚有待商榷 。
3.价值因素的检验
在排除 RATIO 和 ROE作为回归方程的解释变量后 ,在表 8的模型 (7)
1
中 ,市净率倒数的回归系数 为 0.1622,且回归系数在 0·05的水平上显
著为正 ,其 回归方程为
1
c 。一l 。 一0·0459+0.1622P——/B V (1 5)
F检验也显示上述 回归方程在 0.05的水平上显著成立 。如果市净率 由 5
降低为 4 ,则累计异常收益率 由 ~0.0135变为 一0.0054 。可见随着市 净率的
下降 ,异常收益率的绝对值降低 ,股价对非流通释出的消极反应程度变弱了 。
1
同样地 ,表 9的模型 (8)所对应的市 净率 的倒数 的回归系数为
0.1467 ,且 回归系数在 0.05的水平上显著为正 ,其 回归方程 为
1
脚l。一0.0 2+0· 67高 (16)
述 回归方程也在 0.05的水平上显著成立 。同样地 ,在方程 (16)中 ,随着
市净率的下降 ,上市 日后股价对非流通股释 出的消极反应程度也将变弱 。
由此可见 ,无论在公告 日前还是在上市 日后 ,市净率与股价的消极反应程
度都呈现正 向相关关系 ,即市净率越低 ,股价的消极反应程度越小 ,因此命题
3成立 。在转配股释出的过程中 ,公 司的价值因素是影响股价消极反应程度的
重要因素 1 z。
:在研究中 ,笔者还发现了一个非常有趣的现象 ,转配股上市的股价消极反应的
程度 (累计异常收益率)与其股票价格存在着显著的回归关系 (负相关) :
上述结果表明 ,至少从统计意义上说 ,累计异常收益率与股票价格这一指标呈
现明显的显著负相关性 。股票价格每减少 1元 ,公告前 10目的累计异常收益
维普资讯
60 顾勇 吴冲锋 冷大丽
表 9 上市后 10日 (DD,DD+10)累计异常收益率的回归统计表
一 0.0393
(一2.7617)
- 0.0099
(一1.3343)
一 0.0398
(一2.8973)
- 0.0441
(一3.3993)
一 0.0104
(一1.4005)
一 0.0149
(-2.6224)
一 0.0442
(一3.5190)
一 0.0001
(一0.8073)
一 0.0002
(一1.0172)
一 0.0001
(-0.7997)
一 0.0002
(一0.9636)
0.001 5
(0.1442)
0.0073
(0.7360)
0.0007
(0.0649)
0.0065
(0.6576)
0.1414
(2.4067)
0.1435
(2.5249)
0.1458
(2.495O)
0.0289
0.0037
0.0355
0.0312
- 0.0005
2.4363
0.7336
3.6691
3.3367
0.9285
0.0672
0.4820
0.0279
0.0383
0.3369
— 0.0039 0.4324 0.51 18
0.1467 0.0379 6.71 55 0.0105
(2.59 14)
a 对 同归系数进行显著性柃验 的 统计值 。
对同归方程进行拟合检验的 F统 计值 。
。F统 计柃验的显著性慨率 。
率的绝对值将下降 0.43% ,而上市后 10日的累计异常收益率的绝对值将下降
0.21% 。这是 因为 :
首先 ,股价反映了价值因素 。一般来说股价较低的股票盘子(流通股股数)
较大 ,股价的泡沫成分也较小 ,其股价与每股净资产的接近程度也就较好 ,所
以从某种程度上说 ,股价的高低也反映了公司的投资价值 。事实上 ,在本文的
样本中 ,我们也发现股价与市净率具有显著的相关性 。对股票价格和市净率的
倒数作一元线性回归分析 ,以市净率倒数为解释变量 ,股票价格为被解释变
量 ,所得 的 回归方程 如下 :
P=19.677—29.15 面1 ’Adj—R2=。
.231 ,回 归方 程 的 显 著 性 程 度
(-6.712)
F=45.049,市净率倒数的回归系数和回归方程均非常显著 。假设市净率PIBV
由 4变为 3 ,则股票价格将由 12.39元变为 9.96元 ,即股票价格越低 ,则在
统计上其所对应市净率也就越低 。
其次 ,股票价格的高低反映了股票的流动性 。一般来说 ,股票价格越低 ,
则股票的流动性也就越好 ,投资者越愿意接受 ,对于某一事件股价的消极反应
程 度也就越 小 。
、, 、 , 、, 、 , 、 , 、 J )
维普资讯
jF-流通股释m刈股价的影响 :来 自转 配股上市的证据
表 10 回归方程的多重共线性容忍度 (Tolerance)表
6 l
综 [ 所述 ,在转配股释出过程中 ,转配股 占流通股的比例与股价的消极反
应程度并无显著的相关性 ,即股票供给因素并不显著地影响股价对释 出的消极
反应 。同时 ,公司股票对释出的消极反应程度并不随公司业绩的提高而降低 ,
即至少在统计上绩优公司并不一定导致较弱的股价消极反应程度 在实证中无
据显示投资苦对绩优公 司转配股的释出持更加欢迎的态度 。与此相反 ,投资
者艾、f释出的低市净率转配股则持一种更加欢迎的态度 ,这些股票价格的消极反
应程度较小 ,也就是偏向价值型的股票对非流通股释 出的消极反应程度较小 。
六 、结论及局 限
转配股上市流通是 非流通股释 出的一种形式 ,通过对转配股上市流通对股
票价格影响及其相关因素的分析 ,有助于我们认识国有股 、法人股等非流通股
释出的市场影响 。本文对 2000年 3月到 2001年 4月转配股上市对股价的影
响进行了实证研究 ,研究结论如7.-:
1、转配股上市这一事件在短期内对股票价格存在着显著的消极影响 。
2 、 与国外有关大宗股权交易的研究结论相 比 中国在转 配股上市后股
价继续消极反应的延续l卜f问要长于国外大宗股权交易后股价反应的
延续时间 ,这是由于 中国股票市场相对缺乏效率所引起的 。
3、 在转 配股上市公告 日与转 配股实际上市 日之间 ,股票价格存在明显
的反弹现象 。
4 、 转 配股上市前后股票价格的消极反应程度与市净率指标 呈现明显的
显著正相关性 ,即市净率越低 ,转 配股释出所引起的股价消极反应
程度越小 。
第三 ,股价的高低间接反映了有关该股票的信息传播程度 。B rennan和
Hughes(1 991)的实证研究表明 ,证券分析师 (Analyst,在中同被称为股评
家)更倾向于关注股价较低的股票 ;而股评家关注程度越高 ,则该股的信息传
播程度也就越好 ,因此低价股的信息传播程度较好 。同时 ,市场效率理沦和微
观结构理论还 认为如果信息传播越充分 ,则股价在事件区间内对事件的反应程
度就越小 (参见 Easley和 Hara(1 987)) 。关于这一点 ,国内的马春林和吴冲
锋 (2002)也得出中国股票价格对事件的反应程度与信息供应量存在反向关系
的结沦 。
维普资讯
62 顾 勇 吴冲锋 冷大丽
5、 转配股上市所引起的股价消极反应程度与转配股 占流通股的比例无
显著关系 ,即股票供给因素不是引起股价消极反应的重要因素 。
6 、 在转配股的释出过程 中 ,公 司股价的消极反应程度并不随公司业绩
的提高而减小 。
上述 结沦为国有股等非流通股上市流通对市场的冲击及国有股流通的政策
选择提供了一些证据 :
(1)国有股上市将对股票价格产生显著的消极影响 ;
(2)为降低市场对国有股释 出的消极反应程度 ,国有股释出定价不应以上
市公司的收益状况为定价依据 ,而要充分考虑公司的价值因素 ,即定
价应以每股净资产为依据 ,定价越接近每股净资产 ,则股价的消极反
应程度越小 。
(3)如果定价合理 (接近每股净资产),释出的国有股 占已流通股的比例
对股价的消极反应程度无显著影响 。
本文的局限主要在于 : (1)所有转配股上市的时间范 围为 2000年 3月
到 2001年 4月 ,而在该阶段 ,中国股市呈现 “慢 牛”行情 ,因此本文的结论
只是反应在股市上涨阶段的情况 。由于受大势的影响 ,在横盘或下跌 阶段 ,非
流通股释 出对股票价格的影响可能会不 同 。 (2)本文研究的是转配股上市 ,
但从总体上讲转配股上市的规模要远远小于国有股上市规模 ,这将削弱研究结
果的代表性 ,同时这也有可能是转配股上市对股价的消极影响程度与转配股占
流通 股的比例并 无显著关系的原因 。
参考文献
中文参考文献按汉语拼音排序 。
郭 志刚.1 999.《社会统 计分析方法 —— SPSS软件应用》.中国人 民大学 出版社.
楼洪豪.1997.转配股上市对我国证券市场的影响.《浙江大学学报 (社会科学版)》
第 2期 ,88—90.
唐国正 ,许宁 ,王志诚 ,史树中.1998.利用可交换债券变现国有股为国家财政融
资.《经济研究》第 10期 ,64—69.
吴喜之.1999.《非参数统计》.中国统计出版社.
许圣道 ,耿志民.1997.关于转配股问题的若干思考.《金融理论与实践》第 ll期 ,
8—9.
易晓明.2000.从转配股上市谈对证券市场的影响.《青海金融》第 5期 ,42—44.
马春林 ,吴冲锋.2002.中国股票市场信息供应量影响因素分析.工作论文.上海
交通大 学.
维普资讯
F流通H殳释 出尉股价的影响 :米 自转 配股 』二市的证据 63
朱武祥 ,郭志江.1999.股票市场对非流通股比例的价格反应 —— 兼析释放非流
通 股市冲击效府及策略.《经济研究》第 5期 ,32—38.
Aitken,M.and Frino,A.(1 996),“Asmmetry in Stock Returns Following Block Trades
on the AuStrallan Stock Exchange:A Note”,ABACUS 32:54—6 1.
Arnold,R.and Cowan(1992),“Nonparametric Event Study Test”,Review ofQuanti—
tative Finance and Accounting 2:343—358.
Brennan,M.and Hughes,P.(1 99 1),“Stock Prices and the Supply of Information”,
Journal ofFinance 46:1665——1691.
Corrado,C.I.(1 989),“A Nonparametric Test for Abnormal Security—price Perfor—
mance in Event Studies”,Journal ofFinancial Economics 23:385—395.
Easlev,D.and o’Hara,M.(1 987),Price,Trade Size and Information in Securities
Markets./ournal ofFinancial Economics 1 9:69-90.
Holthausen,R.W.,Leftwich,R.W.and Mayers,D.(1 987),“The Effect Of Large
Block Tran saction s On Securities Prices:A cros s—sectional Analysis”,Journal of
Financial Economics 1 9(2):237—268.
Holthausen,Leftwich,and Mayers(1 990),“Large—Block Transaction s,the Speed of
Response,and Temporary and Permanent Stock—Price Effects,”Journal ofFinancial
Economics 26:71—95.
KeiITI and Madhavan(1 996),“The Up stairs Market for Large—Block Tran saction s:
Analvsis and Measurement of Price Effects,’’Review ofFinancial Studies 9:1-36.
Kraus,A.and Stoll,H.R.(1 972),“Price Impacts of Block Trading on the New York
Stock Exchange,”Journal ofFinance 27:569—588.
Mikkel son,W.H.and Partch,M.M.(1985),“Stock P rice Effects and Co sts of
Secondarv Distribution s,”Journal ofFinancial Economics 1 4:1 65—1 94.
Peterson,P.(1 989),“Event Studies:A Review of I ssues and Methodology”,Quarterly
Journal ofBusiness and Economics 28:36——66.
维普资讯
64
中 国 会 计
C h i n a A c c o u n t i n
与
g a n d
2002年 9月 第 4卷 第3期
财 务 研 究
F i n a n c e R e v i e w
Volume 4,Number 3,September 2002
THE EFFECT 0F N0N-NEGO ABLE SHARES RELEASE
O N S TO C K P R l C E S: E V l D E N C E F R O M T H E
DISTRIBUTION OF RIGHTS ISSUES’
Yong Gu,Chongfeng W u,and Dali Leng
Translated by Xuelin Zhou
ABSTRACT
The distribution of non—negotiable state—owned shares and legal person shares is a natural
choice for China in its development of the capital market.Our study has selected the rights
issues distributed from M arch 2000 to April 200 l as research samples,and discussed the
impact of their distributions on stock prices as well as the factors involved in price fluctuations.
The empirical results indicate that the distribution of rights issues has a significant impact
on stock prices in the short run.and that the negative response of stock prices after distribu—
tion lasts much longer in China than in overseas markets.as the Chinese stock market is
relatively inefficient.Between the announcement date(AD)and the distribution date(DD)
of rights issues.stock prices had an obvious rebound,and it was observed that the degree of
negative response before and after the distribution has a significant negative correlation
This is a translation of the paper on PP.39-63.
Our research is sponsored by the National Science Fund for Distinguished Young Scholars
under the project of‘Financial Engineering and Financial Complexity’(Project No.
70025303)and the Trans—Century Fund of the Ministry of Education under the project of
‘Price Behaviors of the Financial Sector’.
Our special thanks should go to Dr.Gongmeng Chen,Director of the China Accounting
and Finance Research Centre under The Hong Kong Polvtechnic University,who has of—
fered valuable advice on the research subject and methodology.We are also grateful to Miss
Leung,who has provided relevant overseas literature.Dr.Hailong Liu,Dr.Xuezhi Oin,Dr.
Chunlin Ma,and Mr.He Li from the M anagement School of Shanghai Jiaotong University
have helped us in one way or another.Last but not the least.we are much indebted to the
anonymous referees at the China Accounting and, ance Review for their important advices.
We are,however,responsible for all the errors in this paper.
Yong Gu,MA,Fullgoal Fund Management Co.,Ltd.,1 565 Nanjing Road,Shanghai(200040),
PRC;Chongfeng Wu,Professor;Dali Leng.Director Assistant of the Institution of Contem—
porary Finance;Management School of Shanghai Jiaotong University。535 Fahuazhen Road.
Shanghai(200052).PRC.
维普资讯
EVIDENCE FROM THE DISTRIBUTION OF RIGHTS ISSUES 65
with stock price/net asset per share(P/BV),but has no significant correlation with the ratio
of rights issues to the total number of negotiable shares(RATIO).The negative response of
stock prices does not reduce with the improvement of company performance.
Ke
.
vwords:stock price response,rights issues,distribution,event study,abnormal returns
I.RESEARCH BACKGROUND
Listed companies in China have issued two types of shares:negotiable and non—
negotiable.The former refers to public shares held by individuals.and the latter
includes state.owned shares and legal person shares(rights issues were also non—
negotiable before April 2000).This phenomenon is peculiar to China,and accord—
ing to international practices,all shares of a listed company should circulate in a
unified market.In addition.there are only common shares and preferred shares in
equity structure,but no such things as negotiable public shares or non—negotiable
state—owned and legal person shares.
W ith the increasing standardization of China’s stock market and its integration
with international practices.all rights issues have been made negotiable.and the
distribution of state—owned shares and legal person shares seems to be the only
choice jn the development of China’s capitaJ market.Indeed.since J 999 the coun.
try has been discussing the schemes for non.negotiable shares to be distributed on
the market.
In December 1 999,China Jialing(600877)and Oian Tires(000589)became the
first companies to pilot the‘allocation of state.owned shares to holders of public
shares’.3 Between M arch 2000 and April 2001.the non.negotiable rights issues were
distributed on the market in batches.
‘Rights issues’refers to shares whose rights are partially or wholly transferred to
public shareholders from state—owned and legal person shareholders as the latter
fail to subscribe them for lack of funds or other reasons(Lou,l 997).These rights
issues were non—negotiable before April 2000. By the end of M arch 2000(i.e.be一
China Jialing transferred l 00 million state—owned shares to its public shareholders at a ratio
of 10 t0 8.36l shares.with the reduction ratio of 21.10 per cent in the tota1 capita1.The
percentage of state—owned shares to the tota1 capita1 fel1 from 74.76 percent to 53.66 percent.
Oian Tires transferred 1 7.1 0 million state—owned shares to its genera1 investors at a ratio of
1 0 to 1.59097,with the release ratio of 6.73 per cent in the tota1 capita1.The percentage of
state—owned shares to the tota1 capita1 of the two companies fel1 respectively from 74.76 per
cent and 57.3 per cent to 53.66 per cent and 5 l per cent.
Regarding to the rights issues,the Nationa1 Administration of State—owned Assets issued
No.I 2 Document(I 994)一 ‘Emergency Circular on Protecting the State Equities in Rights
Issues’,requiring that‘state—owned shareholders should not vote for rights issues on the
general meeting of shareholders unless it is deemed as necessary.or when they do have no
capability r0r subscription’.State—owned shareholders should not give up rights issues
arbitrarily,and when such transfer is necessary.negotiated transfer should be the priority
.
The China Securities Regulatory Commission issued No
. 1 6 l Document on 0ctober 27.
1 994~ ‘Specific Regulations r0r Listed Companies on Rights Issues Application and Info卜
维普资讯
66 Gu,W u and Leng
fore the distribution of rights issues).over l 70 listed companies had historical prob—
lems of rights issues,taking up l 5.6 per cent of the total,and the total number of
rights issues reached 3.5 billion。taking up 4 per cent of A—shares.We fo und l 5
companies with rights issues in 1994.132 from 1995 to 1997.and 23 in 1998.The
prices of these rights issues were normally low.with the highest seen in Dongda
Erpai f6007 1 8),14 yuan per share,and the lowest seen in ST Yue Haifa,1.5 yuan
per share.Three companies had rights issues with prices over l 0 yuan,twenty—
seven companies with prices between 6-1 0 yuan,seventy—six companies with prices
between 4-6 yuan,and the rest with prices below 4 yuan.The average transfer
charge was below 0.2 yuan fYi.2000).
In June 200 1.the State Council promulgated the‘Interim Regulations on Reduc.
tion of State—owned Shares to Raise Social Security Funds’5 prescribing that the
‘reduction of state—owned shares.including state shares and state—owned legal per—
son shares.should be interpreted as a behaviour to transfer state.owned shares of
listed companies.including would—be listed companies.to public investors of indi.
viduals and securities investment funds’.and that‘stock issue should be the key
channel for the reduction of state.owned shares’.Since then.the reduction of non.
negotiable state—owned shares was put on agenda.
The distribution of non—negotiable shares will invariably increase the stock sup.
ply on the market,and affect market confidence and thereby stock prices.Out of
this concern,the regulatory authorities are very cautious in developing reduction
schemes.and investors take the event as an important basis for their investment
decisions.Based on the data of the listed rights issues.this paper performs a pre
1iminary empirical analysis on the impact of their distribution on stock prices.and
aims to provide advice for research and policymaking concerning the distribution
of non—negotiable state—owned shares.
Rights issues were selected as research samples not only because they were one
of the major category of non—negotiable shares and their distribution is quite similar
to that of state—owned shares and legal person shares.but also because all the rights
issues were distributed on the market by M ay 200 1.which ensured a sensible size
for empirical studies.
This paper is structured as fo·llows:Part I introduces the research background;
Part II reviews the related literature;Part III discusses the research methodology;
mation Disclosure’.pointing out that‘the increased shares after the transfer should not be
traded on the market for the time being,and this should apply to listed companies with
rights issues in l 994’.Since then.the authorities reiterated the policy of suspending the
distribution of rights issues several times.In this context.1isted companies qualified for
rights issues and with a large financial demand began to sell the rights to the general public
as the state—owned and legal person shareholders were not able to subscribe.But by now,
China has stopped the practice to prevent the aggravation of the historical problem.(Please
refer to Xu and Geng.1 997.1
However,within three months after the promulgation of the‘Regulations’.both markets in
Shanghai and Shenzhen dropped over 30 per cent,which led to the decision by the China
Securities Regulatory Commission to suspend its implementation.
维普资讯
EVIDENCE FROM THE DISTRIBUTION OF RIGHTS ISSUES 67
Part IV provides the research data and descriptive statistics;Part V describes the
research results;and Part VI presents the research conclusions and limitations.
.¨LITERATURE REVIEW
There are no such things as‘reduction of state—owned shares’in the Western literature.
as shares are not divided between individual public shares.state—owned shares.and
legal person shares in Western securities market.In addition.privatization of state—
owned enterprises in the West refers to direct sales or initial public offering fIP0).
which has no impact on stock prices.
From the perspective of the distribution of non—negotiable rights issues.the re—
duction of state—owned shares is close to‘large block trading’in overseas markets.
In defining‘large block trading’,Kraus and Stoll f1972)equated it to any transac—
tion with the number of shares over the market capacity.Aitken and Frino f 1 996)
further defined it as a transaction with the number of shares over l per cent in the
total of negotiable shares.Kraus and Stoll(1 972).and Holthausen,Leftwich.and
Mayers(1987;l990),tested the negative impact of large block trading on stock
prices on the New York Stock Exchange fNYSE).and found that such an impact
was statistically significant.Keim and Madhavan f l 996)performed similar tests on
the NYSE,AMEX,and NASDAQ NMS,and the results were the same.Similar
conclusions can be found in Aitken and Frino f 1 996).who performed an empirical
study on the impact of large block trading on stock prices on the Australia Stock
Exchange fASE).
Also close to large block trading is‘secondary distribution’.which refers to the
sale of outstanding shares not on the stock exchanges due to the enormous amount
involved but after the close of the day at a price equal to or below the closing price
.
Mikkelson and Partch f l 985)tested the impact of secondary distribution on stock
prices,and found that,similar to large block trading.it had a significant negative
impact on the returns of individual stocks.
W hy should large block trading and secondary distribution have such a remark—
able impact on stock prices?Kraus and Stoll(1 972)provided three key factors:f 1)
information factor-- large block trading can convey to investors a message and
change their expectations for stock returns;f2)investors’preferences for specific
stocks;and(3)liquidity cost in the short run.Holthausen,Leftwich.and Mayers
(1 987)also argued that liquidity cost,inelasticity of stock demand curve.and info卜
matlon were the main factors that led to a significant impact of large block trading
on stock prices.
Event study is often used for significance test of the impact of large block trading
on stock prices.Peterson(1 989)defined the method as a number of procedures:f 1)
determination of event day;(2)selection of event period;(3)selection of estimation
period;and(4)calculation of abnormal return fAR).Significance test can be para—
metric(Mikkelson and Partch,l 985;Peterson,1 989)and non—parametric fCorrado.
1 989;Arnold.1 992).
维普资讯
68 Gu,W u and Leng
In analyzing the factors involved in stock price fluctuations。overseas scholars
often use multilinearity regression.Mikkelson and Partch(1 985)argued that the
information factor and supply.demand factor resulting from secondary distribution
led to the abnormal fluctuations of stock prices.The information factor was mea—
sured by the type of seller fi.e.whether the sellers were the management of the
company)。the holding percentage of the secondary distribution shareholders before
and after the distribution.and the market value of the sold shares.The supply—
demand factor was measured by the percentage of secondary di stribution shares in
the total outstanding shares.They found that AR had a significant negative correla—
tion with the type of secondary distribution.0 but not with the number and value of
the shares.M ikkelson and Partch f l 985)concluded that information was the key
factor involved in the price fluctuations.Kraus and Stoll(1972)found that the ef-
fect of stock price fluctuations was significantly correlated with the volume of large
block trading.
In China,the reduction of state.owned shares was iust initiated,most of the exist.
ing studies are qualitative.and few empirical studies can be found on the impact of
the distribution of non.negotiable shares on stock prices.11ang et a1.(1 998)took the
average prices of 626 shares on Shanghai and Shenzhen markets in April l 998 as
the dependent variables,and the earnings per share(EPS)and percentage of nego.
tiable shares in the prior year f l 997)as the independent variables.and performed a
regression analysis to examine the response of stock prices to the percentage of
state.owned shares.Zhu and Guo f l 999)included the average return of the industry
in the independent variables and perform ed a similar regression analysis.Their studies
indicated that stock prices were negatively correlated to the percentage of nego.
tiable shares.However,these qualitative studies are static analysis,which examines
only the correlation between stock prices and percentage of negotiable shares。and
ignores the impact of the release of non.negotiable shares on stock prices.
1¨.RESEARCH DESIGN
The research method used in this paper is event study,and empirical study has been
performed on the effect of the‘release’of non—negotiable shares.In the meantime,
regression analysis has been made to test which factors are involved in the relations
between the release of non—negotiable shares and stock prices.
1.Measure of Impact on Stock Prices
Kraus and Stoll(1 972)defined the impact on stock prices(or the stock response to
the event)as follows:
= +DI 、 .r,『、 厂m (1)
Is the management of the company the seller of the secondary distribution shares?If yes
,
the dumb variable is set as 1.and as 0 ifnot.
维普资讯
EVIDENCE FROM THE DISTRIBUTION OF RIGHTS ISSUES 69
where P
t,
D t,and l t represent stock prices,stock dividends,and market index on t—
day respectively.With the maturity of event study in the 1 980s,the measurement of
the impact on stock prices has been greatly improved.
First of al1.CAPM modeliS used to estimate the normal return.that is:
, = & + ,+£ , (2)
where R
,,represents the daily return of stock J on t-day,R, r represents the daily
return of the market index on t-day.and£ iS the random error that iS subject to
standard normal distribution.Based on the data of individual stocks and market
index during the estimation period,parameters of&,and f are obtained with the
least square method.With these parameters and the daily return of the market index
during the event period.the normal return of each stock iS calculated.The A尺 Of
each stock during the event period iS the actual daily return less the normal return.
0r
At,= ,一(岛+ )
The average abnormal return(A R)of N share on t-day is
M R =
1善I A V —
In addition to the impact on stock prices in one trading day,we have also examined
the cumulative effect during the event period,that is,the cumulative abnormal re—
turn(CAR).
CAR . =∑At,
,=Dt
Dl is the opening day to calculate CAR,and D,is the closing day.The average
cumulative abnormal return(ACAR)of N share during the event period is
A
.
=专 N萎D,A
2.Significance Test
Statistical tests must be performed on the above AR and CAR to examine the
significance of the impact on stock prices.Here we will use both parametric and
non—parametric methods for the significance test.
(1)Parametric Test
To perform a significance test on AR
, all the statistical figures must be unified, and
our study is based on Mikkelson and Panch(1 985).
First of all,the daily AR is unified as follows:
SARa,=ARi 'J
i
维普资讯
70
where
r- f
+
Gu,W u and Leng
(8)
In the above equation.V represents the regression variance of company J in regres-
sion model(2),ED represents the length of the estimation period(trading days),R
represents the daily return of the market index on t-day during the event period,and
represents the average daily return of the market index during the estimation
period.The average standardized abnormal return SAR)is
ASAR,= 1 N
It iS generally assumed that A尺 fregression variance)iS SUbJect to normal distribu—
tion, SAR iS subject to r distribution,and the variance iS£D/(ED一2).According to
the centrallimit theorem,if AR f =l,...,Ⅳ)iS independent,then ASAR iS sub—
iect to the normal distribution when variance iS『 D/(ED一2)(Ⅳ)1.As the value of
D iS big,and D/(ED一2)iS close to l,the variance ofA阴 R iS close to 1/N,or
A 尺 ~N(0.1/N).The standard statistics to test the significance ofAR can be as一
一
sumed as follOWS:
z,A ="~-N(ASAR,1 (10)
where z
,
A
~ N(O,1).Similarly,the standard statistics to test the significance of CAR
can be constructed as follows:
CA R :
/)2
From zz N(0,1),we can first obtain∑zA N(O,02一Dj+1),and then
—DI
CA
,
R
~ N(0,1).
W hen the standard statistics for daily AR and CAR is constructed,and the
significance level(0.05 in this paper)is determined,we can test the significance of
A尺.
(2)Non-parametric Test
To further test the significance ofAR during the event period,we use the W ilcoxon
signed—rank test(See also Wu,1 999)as follows:
(i) Forj=1,.. .,N,calculate ,I
Please refer to Mikkelson and Pa~ch(1 985)
, 一 .
盟
二
肋∑
叻
维普资讯
EVIDENCE FROM THE DISTRIBUT10N OF RIGHTS ISSUES
(ii) For Set ,={lag.,Illag, l≠0},find their respective ranks.If there are mul—
tiple sample Doirits,use the average for each rank.
fiii1 Set w equal to the sum ofranks of lag 1 when A .,>0,and W『equal to the
sum of ranks of laR.1 when A .<0.
For the bilateral test,W is usually close to under zero assumpnon.
Thus.when either of them is too small,we should doubt the zero assumption,
and set W=rain( . ).In a similar way,we can set W= or W= f0r
the unilateral test.
When the sample size is too small f~<30),we can look up the distribution
table of the W ilcoxon signed—rank test.W hen N is big.we can use the
normal distribution approximation and construct the statistics as follows:
W一~f~+1)/4
Z,m=—4—N—(N—+— 1)(—2N— + 1)一/24
where Z ~N(0,1).When the significance level(0.05 in this paper)is
determin d.we can test the significance of AR.
3.Factors Involved in Price FIuctuations of IndividuaI Stocks
The release of non—negotiable shares will produce an impact on stock prices,and
the degree of the impact can be measured by CAR that difiers among individual
stocks.W hat are the factors contributing to the difference?Based on the AR—related
factors mentioned in the‘literature review’and the difference between the release
of non—negotiable shares in China and the overseas large block trading or secondary
distribution,we propose three major factors that can better reflect the realities of
C:hina and provide an important basis for policymaking.
(1)Stock Supply Factor
If a company’s stock cannot be fullv substituted by another in investment decisions,
the investor’s demand is not perfectly elastic,which means the demand curve moves
right downward.Increase of stock supply will lead to the supply curve moving right
and the equilibrium price falling down. As the distribution of rights issues increases
stock supply,the supply curve moves right downward.In the meantime,the stock
demand is not perfectly elastic,and therefore,stock prices must respond negatively
to the distribution of rights issues.The stock supply curve moves right further and
the price fluctuation increases with the rising number of negotiable shares distrib—
uted on the market.Therefore,if we want to test the significance of impact of stock
supply factor on stock fluctuations,we need to test whether Hypothesis 1 holds
water.
Hypothesis 1:The higher the percentage of a company’s rights is-
sues to the total outstanding shares,the larger the negative response
of its stock price to their release.
Please rcfer to Mikkelson and Partch(1 985)
维普资讯
72 Gu,W u and Leng
If Hypothesis 1 holds water,the supply—demand relation is an important factor
contributing to the negative response of stock prices to the release of non’
negotiable shares.
(2)Profitability Factor
M any schemes are in favor of giving a priority to high—performing companies,on
the assumption that‘the most pretty girl should get married first’.It is generally
accepted that investors are more ready to invest in shares of the high performers,
and therefore the release of non.negotiable shares will have a smaller negative impact.
The opposite is also true:when a company has a bad performance,the release of its
non.negotiable shares will produce a large negative impact on stock price.Therefore,
if we want to test the consequences of releasing non—negotiable shares,we need to
test whether Hypothesis 2 holds water.
Hypothesis 2:The better a company’s performance。the smaller the
negative response of its stock price to the release of non·negotiable
shares.
In our later empirical test,wc will use the return on equity(ROE)in 1 999 to
measure a listed company’s perform ance.As most rights issues happened from April
2000 to M arch 200 1,when the 2000 annual reports were not yet disclosed,it is only
natural to measure a company’s performance with the information disclosed in the
1 999 annual reports.
(3)Value Factor
In genera1.the intrinsic investment value of a stock is the most important factor for
investors.W hen non.negotiable shares are released.stock prices will have a smaller
negative response if their prices are slightly higher than the intrinsic investment
value.In contrast.if their prices are much higher than the intrinsic investment value.
investors will‘vote with their feet’and the negative response will be much stronger.
In asset valuation in China.net asset is the key indicator of a company’s intrinsic
value.In the meantime.investors also value net asset per share as it has greater
reliability than ROE.The closer the stock prices are to their net asset per share,the
happier investors are towards the release of non.negotiable shares and thereby the
smaller the negative response of stock prices.
To further assess the impact of value factor on stock prices.we include the indi.
cator of P|B to examine the relations between CAR and P|Bv The smaller the
P|BV.the happier nvestors are towards the release of non.negotiable shares and
therefore the smaller the negative response of stock prices.To this end.we will test
Hypothesis 3.
In the US,P/BV is an important indicator used to evaluate whether a stock is a value stock.
M orningStar,for example,uses the average P/BV of portfolios as an important criterion to
judge whether the mutual fund is a value fund or a growth fund.
维普资讯
EVIDENCE FR0M THE DISTRIBUTION OF RIGHTS ISSUES 73
Hypothesis 3:The smaller the P/BV,the weaker the negative response
of stock prices.
To calculate P/B we take the average closing price of twenty days before AD as
numerator.and the net asset per share disclosed in the 1 999 annual reports as
denominator.As most of the companies with rights issues did not disclose the 2000
annual reports during the period.the net asset per share disclosed in the 1 999 annual
reports becomes the key measurement of a company’s value.
In later analysis we will use multilinearity regression and take CAR as depen—
dent variables to measure the negative response of stock prices to the release of
l
rights issues,and RATIO,the 1999 ROE,and P{Bv as independent variables.On
this basis,we construct a regression equation as follows
CA
,
=oh+),IiRATIO+71 2ROE+),I 3 P{Bv +£I (1 3)
Through significance tests we can find which variables have a significant impact
on C Of individual stocks.
IV.DATA SOURCES AND SAMPLE SELECT10N
1.Data Sources
The data used in this paper come mainly from the Reuters Information System.
Shanghai Securities Daily,China Securities Daifv.and StockStar.com.Tb be specific,
the daily trading data are from the Reuters Information System in the Reuters Fi—
nance Lab at Shanghai Jiaotong University;the number of distributed rights issues,
AD.and DD are from the announcements of related companies on Shangha Secu—
rities Daily and China Securities Dai/y:and ROE and the number of negotiable
shares are from the l 999 annual reports fthe actual number has been adjusted if
there were rights iSSHeS between early 2000 and AD).The definition and value of
the related variables are shown in Table 1.
The口value in Table 1 is based on the data of 120 trading days from the l50th to
the 30th day before AD fdata are weighted when there were rights issues during the
calculation period)and calculated according to the market mode1.In line with the
listing markets,the Shanghai A.share composite index and Shenzhen A.share sub.
index are taken as the benchmark.In selecting the average stock prices.we use the
average closing prices of twenty days before AD to exclude any potential impact on
the final results brought by the uncertainty of date.
2.Sample Selection
Between M arch 2000 and April 200 1,there were 1 67 A—share companies⋯with
distribution announcements.Ninety—nine of them were listed on the Shanghai mar-
” The data of another three companies were not available on China Securities Daily and
Shanghai Seeurities Dail)’.
维普资讯
74
Table 1 Description of Variables and Their Values
Gu,W u and Leng
ket and sixty.eight were listed on the Shenzhen market,with a total of 3.467 billion
shares,taking uD 2O.7 per cent in the total of negotiable shares.
It can be seen from Table 2 that eighteen companies had a ratio of rights issues to
negotiable shares below l per cent,totaling 7.069 million shares,less than 0.2O per
cent in the total number of rights issues.In the meantime,the daily exchange rate
fdaily trading volume/total number of negotiable shares)iS often higher than l per
cent for most stocks.As Kraus and Stoll f l 9721 argued that the stock number of
large block trading was much larger than that of the market capacity,we have ex-
eluded the eighteen companies with a ratio below l per cent and focused our analy-
sis on the remaining l49 shares.In addition,PT Wangdian(600833)has also been
excluded from our samples because it was under particular transfer fPT)“during
the distribution of rights issues.Therefore.the total number of samples iS l 48 shares
in our study.among which fifty.two are from the Shenzhen market and ninety-six
are from the Shanghai market.
Listed companies are required to disclose the number and date before the distil-
bution of rights issues in the China Securities Daily or Shanghai Securities Daily.
The interval between AD and DD often varies from 0 to 4 trading days.In the l48
samples,twenty.six had an interval of 0 day(1 7.6 per cent of the tota1),twenty-nine
had an interval of one day f l 9.6 per cent in the tota1),seventy—eight had an interval
PT iS the acronym for‘Particular Transfer’.W hen listed companies are suspended from
trading due to losses for three consecutive years.the stock exchange and its members
provide‘particular transfer service’on Friday rexcept legal holidays)for holders or inves—
tors of the initial negotiable shares.PT shares can only be applied fo r transfer at the open—
ing time each Friday(except legal holidays)and fixed through collective bidding after the
closing of the day.In other words,PT shares only have one transaction and one price each
week.
维普资讯
EVIDENCE FR0M THE DISTRIBUT10N OF RIGHTS ISSUES
Table 2 Ratios of Rights Issues to Negotiable Shares in 1 67 Companies
75
nutuber of
companies with
r1 ,~hts i sues
nutuber of r1ghts
i sues¨ 0
29 22 0
706 9 7377 l l5600 0 55l 35 8 65764 5 47776 4 27736 1 93344 3 33250 4 346691 4
thousand shares)
percentageintotal 0 20 2 1 3 4 50 1 5 90 1 8 97 1 3 78 8 00 26 92 9 59 100
rights shares(% )
RATIO f%)
R0E f% 1
P/BV
28.54
— 2.21
5.02
17.71
8.02
5.57
1.09
564.71
166.59
223.89
38.04
1.75
of two days(52.7 per cent of the tota1),and the rest had an interval of three or more
days.
Table 3 lists the descriptive statistics of factors affecting stock prices.The data
are treated with EXCEL and SPSS.
V_RESUL_TS AND ANALYSlS
1.AbnormaI Return and SjgnⅢcance Test
(1)StatisticaI Results and Analysis before AD and after DD
AS the interval of each stock between AD and DD varies from 0 to four days.our
calculation of AR and significance test can only be based on the data of twenty—five
days before AD and twenty—five days after DD.
The A尺for twenty—five days before AD and the results of both parametric and
non—parametric significance tests are listed in Table 4.and the AR for twenty—five
days after DD and the results of both parametric and non—parametric significance
tests are listed in Table 5.
In Table 4 and Table 5,column(1)represents the sequence of trading days.fAD—
m)represents m—day before AD,and(DD+m)represents m—day after DD.Column
(2)represents the AR of shares obtained from equation 4.Column f3)iS the
significance test of the data in column(2),and the test value Z iS obtained from
equation 1 0.Column(4)represents the percentage of shares with negative A尺.Col—
维普资讯
76
Table 4 AR and Test Results of 25 Days before AD
Gu,W u and Leng
Zero assumption means the AAR is 0.
Zero assumption means the percentage of stocks with negative AR is 0.50 on one trading
day.
Significant at 0.05 leve1.
Significant at 0.0 1 leve1.
umn f5)iS the non—parametric significance test of the data in column(4)and the test
value iS obtained from equation f l 21.
Table 4 shows that the average AR of sample companies on AD iS-0.38 per cent.
and significantly negative at 0.05 leve1.On the same day.68 per cent of shares have
an AR below 0 and are significant at 0.05 leve1.This indicates that stock prices have
responded negatively to the information of distribution.The average AR iS negative
for ten trading days before AD fAD—l 0,AD-1),despite the fact that the parametric
test iS significantly negative at 0.05 level only on fAD一10).The non—parametric test
shows that the AR iS negative and significant at 0.O5 level on(AD一1),(AD一2),
维普资讯
EVIDENCE FROM THE DISTRIBUTION OF RIGHTS ISSUES
Table 5 ARTestResults of25 Days afterDD
77
“ Zero assumption means the AAR is 0.
Zero assumption means the percentage of stocks with negative AR is 0.50 on one trading
day.
Significant at 0.05 leve1.
Significant at 0.0 1 leve1.
fAD-4),fAD一9),and fAD—l0).Between the eleventh day beforeAD and the twnry一
行fth day before AD fAD一25,AD—l1),the average AD has been negative for nine
days.but not significantly negative at 0.05 leve1.These phenomena indicate that the
response of stock prices to the distribution of rights issues concentrates on these ten
days(AD一1 0,AD一1),and is not all significant before these days(See Figure 1).
This finding is consistent with the statistical results in the US mflrket by Kraus and
Stoll f 1 972)and Mikkelson and Partch f l 985).Such phenomena can usually be
explained by the information factor:investors are so concerned about the distribu—
tion of rights issues that they collect.analyze.and distribute any usefu1information
which has thereby been reflected in stock prices well before AD.
维普资讯
78
Figure 1 Daily AR and CAR before AD
(%)
0.5
O
_o.5
— 1.5
- 2
— 2.5
Gu,W u and Leng
AD一30 AD一25 AD一20 AD一1 5 AD一1 0 AD一5 AD
Let US turn to the figures after the distribution of rights iSSHeS.Table 5 shows that
the average AR iS-0_26 percent on DD.and significant at 0.05 leve1.Fifty.five per
cent of the stocks have a negative AR.but not significant at 0.05 leve1.This indi.
cates that the distribution of non.negotiable shares does produce a negative impact
on stock prices on DD.Within seven days after DD(DD+l,DD+7),the AR iS nega.
tive on(DD+2)and(DD+7)and significant at 0.05 leve1.The non.parametric test
shows that the AR iS negative on(DD+2),(DD+4),and(DD+7)and significant at 0.
05 leve1.From the eight day to the twenty.fifth day after DD fDD+9,DD+25),
eleven days have a negative AR.but not significant in the parametric test.This tells
US that the negative response of stock prices after the distribution of rights iSSHeS
concentrates on seven days in『DD+l,DD+71,but iS not significant thereafter(see
Figure 21.
The distribution of rights iSSHeS in China are quite difierent from the‘large block
trading’and‘secondary distribution’in overseas markets in that the former has a
much longer AR than the latter.Kraus and Stoll(1 972).Mikkelson and Partch(1 985),
and Aitken and Frino f l 996)f0und that the AR in the US had been significant for
only two or three days after the event.which iS obviously much less time than in
China.A possible reason iS that the Chinese capital market lacks e币 ciency com.
pared with the US and other developed markets,and the response of stock prices iS
not as quick as in these markets.Although the stock prices have absorbed the info卜
mation before AD the impact remains or is even enhanced after DD.
This can also be observed from the significant positive correlation between the
CAR after DD and the C 尺before AD.Take C 尺
dD-ln D
as independent variables,
and cA尺肋
.肋 +l0
as dependent variables,and obtain the following equation after
monolinearity regression:
维普资讯
EVIDENCE FROM THE DISTRIBUTIoN OF RIGHTS ISSUES
Figure 2 Daily AR and CAR after DD
(%)
0.5
0
- 0.5
— 1
— 1.5
— 2
— 2 5
DD DD+5 DD+1 O DD+1 5 DD+2O DD+25
79
CARDD
.DD+Io=一0.0106+0.329CA尺 D—I() D (14)
(5.401)
where Adj—R!=0.23 l,and F=45.049,which indicates that cA尺协10 AD has a
significant linear regression relationship with cA尺
,J,J
and cA尺AD
— Io D
is posi—
tively correlated to CARDD
.
DD+10.
Table 6 iS the distribution ofAC .Column f 1)lists the interval of trading days.
with the first item as the opening day and the second item as the closing day of each
interva1.Column(2)iS the ACAR obtained from equation f6).Column f3)iS the
parametric test value obtained from equation f 11 1.
In Table 6,the ACAR of(AD一5,AD)iS-0.70 per cent,but not significant at 0.05
level,while the ACAR of(DD,DD+5)iS-0.94 per cent and significant at 0.05 level
(but significant at 0.1】eve1).In other intervals.the parametric test iS not significant.
Obviously,the absolute value and significance level ofACAR iS larger in fDD.DD+5)
than in(AD一5,AD),which means that the stock prices have a stronger negative
response during 6 days after DD fincluding DD1 than during 6 days before AD
fincluding AD1.
(2)StatisticaI Results and Analysis between AD and DD
In discussing the samples in general,the interval between AD and DD iS not ana—
lyzed and tested as it varies from 0 to four trading days
. Now we will calculate and
test the C 尺 of three groups of sub—samples :companies with an interval of one
trading day(twenty—nine in tota1),companies with an interval of two trading days
(seventy—eight in tota1),and companies with an interval of three trading days ffou~een
in tota1).
There is only one company with an interval of over three trading days
, which obviously
has no statistical significance.
维普资讯
80
Table 6 ACAR and Significance Test
Gu,W u and Leng
The interval between(AD一5,AD)and(DD,DD+5)is six trading days,and the rest is five
trading days.
Zero assumption means ACAR is 0.
Significant at 0.05 leve1.
Table 7 CAR and Significance Test of the Interval between AD and DD
Zero assumption means ACAR is 0.
Zero assumption means the percentage of stocks with negative AR is 0.50 on one trading
day.
Significant at 0.05 leve1.
Table 7 lists the C Of the sub—samples with an interval of one.two.and three
trading days between AD and DD,and the results of the parametric and non—
parametric significance tests.
In the twenty—nine sub—samples with an interval of one trading day between AD
and DD,the daily AAR iS 0.48 per cent.but not significant at 0.05 leve1.Only 45 per
cent of stocks have a negative CAR that iS not significant under the non—parametric
test.The AC of the sub—samples fseventy—eight in tota1)with an interval of two
trading days between AD and DD iS 0.77 per cent,and significantly positive at 0.05
leve1.Fifty—six per cent of stocks have a positive CAR that iS significant at 0.05 level
维普资讯
EVIDENCE FROM THE DISTRIBUTION OF RIGHTS ISSUES 8
under the non—parametric test.The ACAR of the sub—samples(fourteen in tota1)is一
0.42 per cent.but not significant at 0.05 leve1.Fifty per cent of stock have a nega—
tive CAR that is not significant at 0.05 leve1.Given the small size of the samples
with an interval of one and three trading days,the results are not very convincing.
However.the ACAR of these two sub—samples is not significantly negative,and
seventy—eight companies have an interval of two trading days with their ACAR
significantly positive at 0.05 leve1.
As shown in Table 6.the ACAR of six days before AD (including AD)is
significantly negative at 0.05 leve1.W hy should the CAR of the interval between
AD and DD not be significantly negative or even positive?
The rebound of the CAR between AD and DD can possibly be explained by two
factors:(1 1 overreaction-- stock prices overreact to the distribution information of
rights issues so that they rebound after AD (i.e.overshooting);and(2)manipulation
of large institutions-- most of the rights issues are controlled by institutional inves—
tors(e.g.underwriters of rights issues),who will try their best to maintain the stock
prices before DD so that they can sell at much higher prices on DD.
However.Table 6 shows that the C.A of rights issues during the six days after
DD (including DD)is significantly negative at 0.05 level,indicating that the impact
Of rights issues distribution is increasin g after DD.In other words,stock prices have
not overreacted to the distribution infc}rmation before AD.Therefore.the positive
AR during the jnterval between AD and DD cannot be explained by the ove~eac—
tion factor.but rather can be attributed to the manipulation of the large institutions.
W hen the above analysis about rights issues between AD and DD is applied to the
withdrawal of state—owned shares.we believe that.if the released state—owned shares
concentrate on one or a few institutional investors,market manipulation is possible
as profit maximization is their sole purpose.and this will be more obvious during
the interval between AD and DD.
2.Factors Involved in Stock Price FIuctuatiOns
Table 8 and Table 9 provide a regression analysis with the least square method of
the dependent variables,i.e.the CAR(AD—l 0,AD)(DD,DD+l 0)and the indepen一
1
dent variables.i.e.RAT10.R0E in l 999 and .
P{Bv
Column(1)in both tables represents the intercept of the regression equation;
column(2)is the regression coem cient of RATIO:column(3)is the regression
coefficient of the unadjusted ROE in l 999;and column(4)is the regression coefficient
of ———————’’—————————●
P/BV
Table 10 lists the tolerance of multilinearity regression equation.It shows that
Tolerance is the difference percentage between X
i
and other independent variables,i.e.,
Tolerance
,
= 1一 .^The smaller the tolerance is,the lager the information repetition of
and other variables,and the smaller its marginal explanatory power towards Y.When the
tolerance is smaller than 0.1,the multilinearity of this variable and other variables is thought
to exceed the tolerance(see Guo,1 999).
维普资讯
82
Table 8 Regression Statistics of CAR of 10 Days before AD (AD一10,AD)
Gu,W u and Leng
(1)
(2)
(3)
(4)
(5)
(6)
(7)
一 0.O45l
f2.5246)
- 0.0093
(一0.9943)
一 0.04l8
(一2.4213)
- 0.0489
f一3.0122)
- 0.009l
(一0.9868)
- 0.0l37
(一1.9306)
- 0.O459
(一2.91 16)
- 0.000l
(一0.5185)
- 0.0002
(一0.7264)
一 0.000l
(一0.5950)
- 0.0002
(一0.7450)
- 0.0094
(一0.7366)
- 0.0022
(一0.1739)
0.1723
(2.3382)
0.1592
(2.2297)
- 0.0l00 —0.1758
(一0.7937) (一2.4025)
- 0.0029
(一0.2324)
0.1622
(2.2828)
0.0207 2.0223
- 0.0099 0.2907
0.0239 2.77l0
0.0257 2.9l39
- 0.003 l 0.5550
- 0.0066 0.0540
0.0282 5.2ll3
0.1l35
0.7482
0.0660
0.0575
0.4575
0.8l66
0.0239
T value of significance test on regression coeffi cient
F value of test on regression equation.
Significance probability of F statistical test.
the tolerance of the independent variables of RATIO,ROE,and — in equation
Pf BV
(1 3)is above 0.9,indicating that the regression equation does not have multilinearity.
(1)Test of Stock Supply Factor
In Table 8,the monolinearity regression model(5)of CAR and RATIO before AD
shows that the t-test value iS一0.7450.but not Significant at 0.05 leve1.In the
multilinearity regression models(1),(2)and(3),the t-test value of RATIO iS not
significant.either.Similarly.in Table 9 the t-test value of the CAR and RATIO iS一
0.9636 under the monolinearity regression model(5),but not significant at 0.05
leve1.It jS the same with the t-test value of RATIO under the multilinearity regres.
sion models(1),(2)and(3).
Therefore.RATIO iS not significantly correlated to the negative response of stock
prices,thus negating Hypothesis 1.In other words,the stock supply is not an impor-
tant factor jn price fluctuations when rights issues are distributed.
(2)Test of Company’s Profitability
In Table 8,the monolinearity regression model(6)of CAR and ROE before AD
shows that the t-test value of ROE is一0.2324.but not significant at 0.05 leve1.In
维普资讯
EVIDENCE FR0M THE DISTRIBUT10N OF RIGHTS ISSUES
Table 9 Regression Statistics of CAR of 1 0 Days after DD (DD,DD+l 0)
83
- 0.0393
r一2.761 7)
- 0.0099
f—1.3343)
- 0.0398
r一2.8973)
- 0.044l
(一3.3993)
- 0.0l04
(一1.4005)
- 0.0l49
r一2.6224)
- 0.0442
f一3.5 l90)
- 0.000l
(一0.8073)
- 0.0002
r—1.0l721
- 0.000l
(一0.7997)
- 0.0002
(一0.9636)
0.00l5
(0.1442)
0.0073
(0.7360)
0.14l4 0.0289 2.4363 0.0672
(2.4067)
0.1435
(2.5249)
0.0007 0.1458
(0.0649) (2.4950)
0.0065
(0.6576)
- 0.0037 0.7336 0.4820
0.0355 3.669 l 0.0279
0.03 l 2 3.3367 0.0383
- 0.0005 0.9285 0.3369
- 0.0039 0.4324 0.5 l1 8
0.1467 0.0379 6.7l55 0.0l05
r2.59l41
T value of significance test on regression coefficient
F value of test on regression equation.
Significance probability of F statistical test.
Table 10 M ultilinearity Tolerance of Regression Equation
the multilinearity regression models(1),(2),and(4),the t-test value of ROE iS not
significant either.Similarly.in Table 9 the t-test value of the CAR and ROE iS
0.6576 under the monolinea ritv regression model(6) but not significant at 0.05
leve1.It iS the same with the t-test value of ROE under the multilinearity regression
models(1) (2),and(4).
Therefore,ROE iS not significantly correlated to the negative response of stock
prices,thus negating Hypothesis 2.In other words,performance is not an important
factor in price fluctuations when rights issues are distributed.because stock prices
do not reduce with performance improvement.
; ; ) ¨
维普资讯
Gu,W u and Leng
China is an emerging capital market where information disclosure still has many
problems.The ROE of listed companies has a large fluctuation each year,and in-
vestors will not take ROE as an important criterion for investment decisions as it
often lacks reliability.Therefore,Hypothesis 2 does not hold water,and the argu-
ment that‘the most pretty daughter gets married first’is still questionable.
(3)"l est ol Value Factor
Excluding RATIO and ROE as the independent variables,the regression coefficient
of— in model(7)of Table 8 is 0.1 622,and significantly positive at 0.05 leve1.
P}BV
The regression equation is:
CARA 仙 =一。.0459+。.1622 1 (15)
The F-test also shows that the above regression equation holds water at the 0.05
leve1.If PiBV is reduced from 5 to 4.CAR will change from 一0.0135 to_0.0054.
indicating that the negative response of stock prices is lessening with the decrease
l
of P and ·simi a Y,‘he regres i∞ coe饰cien‘of in mode (8)of
Table 9 is 0.1 467.and significantly positive at 0.05 leve1.The regression equation
is:
CARDD
.
DD+l0=一0.0442+0.1 467
Pf Bv
The above regression equation also holds water at 0.05 leve1.and the negative
response of stock prices after AD is lessening with the decrease of P/B .
弋he above two equations show that P|BV s positively correlated to the degree of
negative response of stock prices.regardless of whether it is before AD or after DD.
The lower the P/B the weaker the negative response.which confirms Hypothesis
3:company value is an important factor affecting the negative response of stock
prices.
We have observed an interesting phenomenon in our study:the negative response of stock
prices to the distribution of rights issues(i.e.CAR)has a significant regression correlation
with stock prices(negatively correlated):
维普资讯
EVIDENCE FROM THE DISTRIBUTION OF RIGHTS ISSUES 85
To conclude.RATIO has no obvious correlation with the degree of negative re。
sponse of stock prices.In other words,the stock supply does not produce any
significant impact on the negative response of stock prices towards the release ot
non—negotiable shares.In the meantime,the negative response of company stocks
does not reduce with their performance improvement,that 1s,top performers do not
necessarily lead to smaller negative response.There 1s no empirical evidence to
show that investors are more ready to accept the distribution ot rights issues by
good performers.On the contrary,they are more 1nterested in the rights issues with
lOW P/BV.and the negative response of these stocks 1s relatively smal1.
VI.C0NCLUSIONS AND LIMITATIONS
The distribution of rights issues is one form of the release of non—negotiable shares,
and analysis of its impact on stock prices and the factors involved can contribute to
our understanding of the impact of the release of state—owned and legal person shares.
Our empirical study on the distribution of rights issues between M arch 2000 to
April 2001 comes up with six major conclusions:
The above results show that,at least in statistics,CAR has a significant negative correlation
with stock prices.W hen stock prices are reduced by every 1 yuan,the absolute value of
CAR of ten days before AD will go down by 0.43 per cent,and the absolute value of CAR
of ten days after DD will go down by 0.2 1 per cent.This is because:
First of al1.stock prices reflect the value factor.In general,stocks with a low price and
a huge number of negotiable shares have fewet bubbles in their prices that are closer to net
asset per share.To some extent,stock prices reflect the investment value of companies.In
our samples,we also find that stock prices have a significant correlation with P/BV.When
1
wetake P|Bv as independent variables and stock prices as dependent variables,we
get the following regression equation:
P=19.677—29.155— ,Adj—R!=0.231,F=45.049,andthe regression coefficient
ElBV
(-6.712)
1
。 nd the re si0n equa i。n 。 h igI1i行 “ · i d 。d什。m 4 。 ,
the stock price will go down from 1 2.39 to 9.96.In other words.the lower the stock prices,
the、ower the P|BV.
Second,stock prices reflect stock liquidity.In genera1.the lower the stock prices,the
better the stock liquidity,and thereby the smaller the negative response of stock prices.
Third,stock prices indirectly reflect the information spread.The empirical study by
Brennan and Hughes f 1 99 1 1 shows that analysts tend to be more interested in stocks with
low prices,and the more tbcus they give,the better the information spreads.Therefore,
stocks with low prices have a better information spread.According to the market efficiency
and micro.structural theories,the more adequate the information spreads.the less the stock
prices react to the event.(See Easley and Hara,1 987.)In China,Ma and Wu(2002)also
conclude that there is an inve~ed relationship between the response of stock prices and the
information supply.
维普资讯
86 Gu,W u and Leng
1. The distribution of rights issues has a significant negative impact on stock
2
4
5
6
prices in the short run.
Compared with ‘large block trading’.stock prices in China have a much
longer period of negative response after the distribution of rights issues,which
iS probably caused by the lack of efficiency in China’S securities market.
Stock prices have an obvious rebound during the interval between AD and
DD.
The degree of negative response of stock prices before and after DD has a
significant positive correlation with P/BV.In other words.the lower the
PIBv.the weaker the negative response.
There iS no significant correlation between the degree of negative response
of stock prices and the ratio of rights issues to the total of negotiable shares,
indicating that the stock supply is not an important factor in causing the
negative response.
The degree of negative response of a company’S stock price does not decline
with its performance improvement during the release of rights issues.
The above conclusions provide important evidence for the impact of the distribu—
tion of non·-negotiable shares and policy·-making concerning the distribution of state·-
owned shares:
f 1 1 The distribution of non—negotiable shares such as state—owned shares will
have a significant negative impact on stock prices.
f2) To reduce the negative response of the market,the pricing of state—owned
shares should not be based on the earnings of listed companies.but rather
on their values fi.e.net asset per share).The closer the pricing to net asset
per share,the smaller the negative response.
(3) If the pricing iS close to net asset per share.the ratio of the released state—
owned shares to the total number of negotiable shares will not have any
S1gnificant impact on the degree of negative response of stock prices.
There are two maior limitations in this paper.First.all the rights shares were
distributed between M arch 2000 and April 200 1.during which China’S securities
market was bullish.Th erefore,our conclusions only reflect the situation when stock
prices gO up,and the impact might be difierent when they go down or remain stagnant.
Second,our foCUS of study iS the distribution of rights issues,whose size iS,however,
much smaller than that of state—owned shares.This invariably reduces the reliability
of our study,and also explains why the degree of negative impact brought by the
distribution of rights issues has no significant correlation with its ratio to the total
number of negotiable shares.
维普资讯
EVIDENCE FROM THE DISTRIBUTION OF RIG HTS ISSUES
REFERENCES
87
Aitken,M.and Frino,A.(1 996),‘Asymmetry in Stock Returns Following Block Trades on the
Australian Stock Exchange:A Note’.ABACUS 32:54—6 1.
Arnold,R.C.(1992),‘Non—parametric Event Study Test’,Review ofQuantitatix’e Finance and
Accounting 2:343—358.
Brennan,M.and Hughes,P.(1991),‘Stock Prices and the Supply ofInformation’,Journal of
Finance 46:l665-l691.
Corrado C.J.(1 989),‘A Non—parametric Test for Abnormal Security—price Performance in Event
Studies’,Journal ofFinancial Economics 23:385—395.
Easley,D.and O’Hara,M .(1987),‘Price,Trade Size and Information in Securities Markets’,
Journal of Financial Economics l 9:69-90.
GUO,Z.G.(1 999),Analytical Method ofSocial Statistics." 尸 Application,Renmin University
Press.(in Chinese)
Holthausen,Leftwich,and Mayers(1 990),‘Large—Block Transactions,the Speed of Response,
and Temporary and Permanent Stock—Price Effects’.Journal ofFinancial Economics 26:7l一
95.
Holthausen,R.W ,Leftwich,R.W.,and Mayers,D.(1 987),‘The Efrect 0f Large Block Transac—
tions on Securities Prices:A Cross—sectional Analysis’.Journal Of Financial Economics l 9
f2):237-268.
Keim and Madhavan f l 996).‘The Upstairs Market for Large—Block Transactions:Analysis and
Measurement of Price Effects’.Review of Financial Studies 9:l-36.
Kraus A.and Stoll H.R.f】972).‘Price Impacts of Block Trading on the New York Stock
Exchange’,Journal of Finance 27:569-588.
Lou.H.H.(1 997),‘Impact of the Distribution of Rights Issues on China’s Securities M arket’,
Zhejiang Universitv Journal fSocial Sciences Edition)3:88-90 fin Chinese).
Ma,C.L.and Wu,C.F (2002),‘Analysis of Information Supply Factor in China’s Securities
Market’,Working Paper.Shanghai Jiaotong University(in Chinese).
Mikke1son,W .H.and Partch,M.M .(1 985),‘Stock Price Effects and Costs of Secondary
Dist rjbutions’.Journal ofFinancial Economics l4:l65-l94.
Peterson,P(1 989),‘Event Studies:A Review of Issues and Methodology’,Quarterly Journal
D,Business and Economics 28:36-66.
Tang,G.Z.,Xu,N.,Wang,Z,C.,and Shi,S.Z.(1 998),‘Financing through Convertible Bonds’,
Economic Research l 0:6-69 fin Chinese).
Wu,X.Z.(1 999),Non—parametric Statistics,China Statistics Press(in Chinese).
Xu,S.D.and Geng,Z.M.(1 997),‘Some PreIiminary Thoughts on Rights Issues’,Financial
Theories and尸ractices l l:8-9 fin Chinese).
Yi,x.M.(2000),‘Impact of the Distribution of Rights Issues on Securities Market’,Qinghai
Finance 5:42—44 fin Chinese).
Zhu,W.X.and Guo,Z.J.(1 999),‘The Price Reaction of the Stock Market to Percentage of Non—
negotiable Shares:The Release Efiect and Strategies’.Economic Research 5:32—38 fin
Chinese).
维普资讯