原文：
Corporate Governance and Financial Performance of Companies in
Poland
Abstract
The research presented in the paper is aimed at examining the relationship
between the level of corporate governance and the financial performance of listed
companies in Poland. The corporate governance degree is expressed by the outcomes
of a rating of 2003 performed by Polish Corporate Governance Forum. The attempted
models are of ordered multinomial type. Endogenous variable represents the rating
outcome (A-, B+, B, B-, and C+), while the exogenous variables include various
financial indicators evaluated on the basis of the 2002 financial statements. The
estimated ordered logit models show that the level of corporate governance of
companies in Poland is associated with their ability to cope with the financial distress,
as expressed by the degree of liquidity, profitability and the financial leverage
variables. (JEL C10, O57, G30)
Corporate Governance Ratings for Polish Companies
From the perspective of a company, the corporate governance means:
independent and efficient supervising body, transparent and accurate books, strong
shareholders ’ rights and equal treatment of all owners groups. Mechanism of
corporate governance minimizes the agency costs, ., reduces the company’s market
value loss resulting from a potential conflict between the managers and the owners
(Shleifer and Vishny, 1996).
In Poland, the corporate governance questions have been addressed since the
beginning of the first decade in 21st century, both legally and operationally. Good
source of information on current issues in this area is Polish Corporate Governance
Forum (PCFG) founded in 2000 by the Institute for Market Economics
( Warsaw Stock Exchange (WSE) has adopted the corporate
governance principles on the Polish market since 2002, with all listed companies
declaring that they would observe most of the best practice rules
( Since then, the new document, entitled Best practices in
public companies 2005 has been accepted for implementation. The Best practices
express the corporate governance standing of WSE based on practical experience,
opinions and suggestions of market participants over the period of 2003Y2004 and the
recent European Commission recommendations in this field.
Polish Corporate Governance Forum performed two ratings of the companies
quoted on Warsaw Stock Exchange, the first in 2001 and the second in 2004
(Tamowicz et al., 2001; Dzierzanowski et al., 2004). These two ratings are not really
comparable, since the authors significantly changed the scope and methodology for
the second rating.
The last rating of 2004 was carried out for 53 companies with largest
capitalization on WSE. The data on legal standing of the companies ’ corporate
governance issues were collected as of November 2003. According to the authors ’
description, the 2004 rating was based Bon the analysis of statutes, internal
regulations (by-laws) concerning functioning of supervisory and management boards
and shareholders ’ meetings and content of the companies ’ websites. The very
important sources of information were the companies ’ declarations of compliance
with the Warsaw Stock Exchange Code and especially the commentaries to particular
rules.
Authors of PCFG rating indicate at least 60 characteristics that were taken into
account for each company in order to obtain appropriate picture of company ’ s
corporate governance level. The indicator variables for the rating were taken from the
following areas:
1. composition and competence of supervisory board incl. independent board
members
2. supervision over party-related transactions
3. general shareholders meeting accessibility
4. functioning of the management board
5. Auditor’s independence
6. lack of anti-takeover defences
7. regulations on trading in own shares
8. Companies’declared goals and intentions
9. transparency arrangements, including information available from the
companies’websites.
The disclosed rating uses five categories from A- to C+. These were assigned to
53 companies. The authors indicate that the full range of categories include ratings
from A to E. The rating is presented in Table 1. There are five companies rated C+, 17
companies with a B- rating, 14 companies with B, 13 with B+, and 4 with A-.
Econometrics and the Corporate Governance
TABLE 1：PCFG 2004 Rating of Corporate Governance for the Companies
Listed on WSE
The econometric research on corporate governance concerns mainly the
relationships between various categories representing firm ’ s performance and the
variables describing the governance level, such as ownership structure or the
composition of supervising body. Surveys of such research are presented, ., in
Bhagat and Jefferis (2002), Boersch-Supan and Koeke (2002), and Gugler (2001).
Main factors influencing the corporate governance level, which are usually considered
in empirical research, are:
1. composition of the supervisory board (degree of dependence on
management, board＇s structure)
2. ownership structure (diluted ownership, concentration of ownership,
A- Amica, Agora, BZWBK, Orbis
B+ BPH-PBK, BRE, Computerland, Eldorado, FORTE, KGHM, LPP, Netia,
Pekao, Polfa Kutno, Prokom, TPSA, Stomil Sanok
B Debica, Elektrim, Groclin, Impexmetal, INGBSK, Jelfa, Kety, Kruszwica, Mennica,
PGF, PKNOrlen, Polifarb CW, Rafako, Softbank
B- Bank Handlowy, BIG BG (Millenium), Budimex, Comarch, Farmacol, EFL, Grajewo,
Kogeneracja, Kredyt Bank, Krosno, Lentex, Mostostal SDL, Okocim,Orfe, Rolimpex,
Sokolow, Sterprojekt
C+ Cersanit, Echo, Hoop, Swiecie, Z ˙ ywiec
corporate owners, institutional owners, managerial ownership)
3. acquisitions (including managerial acquisitions) and CEO changing
4. equity structure (debt structure)
5. managerial compensation
Boersch-Supan and Koeke (2002) formulate a number of weak points of
econometric research concerning corporate governance, such as:
1. Structural Reverse Causality: For example, it is not clear what is the
direction of causality between ownership structure and firm ’ s
performance; more concentrated ownership can improve firm
performance, but the reverse relation is also possibleYfirms well assessed
by the market could also attract investors.
2. Missing Variables: In the area of corporate governance it is customary
that major explanatory variables may not be included into the model;
moreover, the of the equations excludes the presence
of higher order terms.
3. Sample Selectivity: Most empirical studies on corporate governance
analyze only the largest companies, usually the listed ones; such samples
are selected by the performance variable and Y in effect Y the studies
have sample selection bias.
4. Measurement Error in Variables: For example, the company ’ s
performance can be measured by different variables, such as market value,
ROA, ROE, EBIT, Tobin’s Q; these variables are sometimes uncorrelated,
., they measure the same performance in different way.
To properly deal with some of these dangers of econometric modeling in
corporate governance it is necessary to use panel data on companies. Boersch-Supan
and Koeke (2002) indicate several such databases for Germany.
Corporate Governance and Economic Performance
The recent survey of corporate governance in OECD countries (Corporate
governance, 2004) indicates that Bstudies that are considered to be best practice
econometric techniques indicate that the corporate governance is an important
determinant of performance (...).Survey authors quote three studies showing
significant association between the level of governance and the firms’ performance.
In another study, Bøhren and Ødegaard (2001) analyze the relationship between
Tobin ’ s Q for the companies quoted on Oslo stock exchange in 1989Y1997 (217
companies by the end of 1997) and several variables representing corporate
governance. Their results indicate ., that ownership concentration has negative
effect on performance, while the effect of insider ownership is positive. Increasing
board size and the use of nonvoting shares decrease performance. The direct
ownership has stronger effect on performance than institutional or state ownership.
Similar outcomes are presented in a study by Lehman and Weigand (2000) for
the sample of 361 German companies in 1991Y1996. In their research, company’ s
performance measured by ROA is related to a number of corporate governance
variables. The panel regression results confirmed that ownership concentration has
negative effect on ROA, while the positive impact of ownership concentration is
found for firms with financial institutions as large shareholders.
Corporate governance indices constructed for various countries also have been
examined with respect to their association both with economic and market
performance of companies. The examples may be found in Gruszczynski (2003). In
the next section, an attempt is made towards examining the relationship between the
corporate governance rating and company’s financial performance.
Ordered Logit
The corporate governance rating represents a typical dependent variable suitable
for ordered response models. The categories are ranked in ascending order and the
distances between neighboring categories are not set as equal. For the exposition we
assume that the observed corporate governance rating variable y can be equal to one
of five rating categories, as appearing in the 2003 Polish rating.
It is assumed that this ordinal variable y is related to the continuous latent
variable y* that indicates the company’s degree of corporate governance. The values
of y* are unknown.
The values of y* determine the outcome represented by y in the following
manner:
yi =1 or C+ ifτ0≤yi * <τ1
yi =2 or B- ifτ1≤yi * <τ2
yi =3 or B ifτ2≤yi * <τ3
yi =4 or B+ ifτ3≤yi * <τ4
yi =5 or A- ifτ4≤yi * <τ5
The τ’s are thresholds (cutpoints), with the values for exteme categories equal toτ
0 =-∞ andτ5=∞.
The linear model for the unobserved y* is as follows:
yi *=xiτβ+εi
where xi is typical [(k + 1) *1] vector of values on k explanatory variables (plus
constant
term) for the ith observation (ith company).
The use of the observed y values requires the assumption about distribution of
errors ε i. Let F be the cumulative distribution function (cdf). The probability of
outcome y = m (m = 1, ..., 5) can be written as:
P(yi=m│xi, β, τ)=F(τm- xiτβ)-F(τm-1- xiτβ)
whereτis the vector of cutpoints. In the equation for P(yi=m│xi, β, τ), the second
term on the right-hand side is equal 0. Also, in P(yi=m│xi, β, τ), the first term equals 1.
For the ordered probit model, errorεis distributed normally with a mean of 0 and
a variance of 1. For the ordered logit model, errorεhas logistic distribution. For the
sake of model’s identification, either constant term or τ 1 is set as equal 0. In the
ordered logit:
F(τm- xiτβ)=eτm- xiτβ/(1+ eτm- xiτβ)
The probability pi for each category of y for the ith company (i = │1,2,...,n) is
equal to:
P(yi=1│xi, β, τ) if y=1
…
pi= P(yi=m│xi, β, τ) if y=m
…
P(yi=5│xi, β, τ) if y=5
The pi’s lead to form the likelihood function, the max I mum of which is attained
for the ML estimates of vectorsβandτ.
Sample and Models
For the purpose of explaining corporate governance rating y of Polish listed
companies, the ordered logit models were specified. Since the rating of 2004 was
based on the November 2003 legal standing, the explanatory variables for financial
performance were calculated on the basis of financial statements for the year 2002.
The sample includes 37 out of 53 rated companies. As many as 16 companies
have been excluded from the sample: Eight banks, one other financial institution, two
companies with the first quotation after January 1, 2002, and five companies excluded
for other reasons (data not accessible, financial statements with missing information).
The endogenous variable y has been defined as:
(1) variable CG with all five ranking categories appearing in original
sample;
(2) variable CG1 with only three categories (first two and the last two
have been combined).
The reason for considering also simpler ranking CG1 is the small size of the
sample as compared to the number of parameters to be estimated in the logit model.
There are 20 financial ratios that have been considered as explanatory variables.
The ratios, calculated on the basis of the companies’ 2002 financial statements are as
follows:
Profitability ratios: P102 gross profit from sales margin, P202 operating profit
margin, P302 gross profit margin, P402 net profit margin, ROE02 return on equity,
and ROA02 return on assets.
Liquidity ratios: L102 current ratio, L202 quick ratio, and L302 acid test.
Activity ratios: A102 amount due turnover, A202 inventory turnover, A302
operating cycle, A402 liabilities turnover, A502 cash conversion cycle, A602 current
assets turnover, and A702 assets turnover.
Debt ratios: D102 fixed assets cover ratio, D202 debt margin, D302 EBITDA/
financial expenses, and D402 debt/EBITDA.
Source: Markek Gruszczynski*,“ Governance and Financial
Performance of Companies in Poland”. International Advances in Economic Research,
,pp. 251–259.
二、翻译文章
译文：
波兰的公司治理与财务绩效
摘要
本文目的是研究波兰上市公司的治理水平和财务业绩之间的关系。公司治理
的水平由 2003 年波兰公司治理论坛进行的评级来表示。本模型为复合式模型。
内生变量表示评级结果（A-, B+, B, B-, 和 C+），而外生变量为以 2002 年财务报
表为基础的各种财务指标。多元 Logit 模型的估计结果显示，波兰上市公司的企
业治理水平是与他们应对财务困境的能力、盈利能力及财务杠杆变量相关联。
（JELC10，O57，G30）
波兰上市公司的公司治理评级
从一个公司的角度来看，公司治理是指：强大的股东权利和平等待遇的所有
业主群体组成的机构对公司独立和高效的监督。公司治理机制的代理成本最小化，
即减少了公司的管理人员和业主之间由于市场价值损失而造成的潜在冲突
（Shleifer 和 Vishny，1996）。
在波兰，自 21 世纪第一个十年开始时，无论在法律上还是业务上，公司治
理问题已经得到解决。在当前问题的这一领域很好的信息来源是年由市场经济研
究所于 2000 年成立的波兰公司治理论坛(PCFG)（
证券交易所（WES）自 2002 年起通过了关于波兰市场的企业管治原则，所有上
市公司也宣布，他们将遵守大部分的最佳实践规则（
从那时起，就产生新的文件，题为 2005 年上市公司的最佳做法已经被接受并执
行实施。最佳做法显示，WSE 公司治理委员会在企业实践经验的基础上，意见
和与会者建议超过了 2003-2004 期间的和最近的欧洲委员会在这方面的建议。
波兰公司治理论坛对华沙证券交易所进行了两次公司评级，第一次在 2001
年，第二次在 2004 年（Tamowicz 等,2001；Dzierzanowski 等, 2004 年）。这两
个级别中并没有真正可比性，因为提交的第二次评级，大大改变了范围和方法。
2004 年的最后评级是对华尔街上最大市值的 53 家公司所进行的。这些对公
司的管治事项的法律地位的数据收集时间为 2003 年 11 月。根据作者的描述，
2004 年的排名是基于法规的分析、内部规定（附例）中关于监督和管理董事会
的运作和股东大会以及公司网站的内容。非常重要的信息来源是公司的声明与华
沙证券交易所守则，尤其是特定规则的评注。
波兰公司治理论坛的创始人的评级表明，为了获得公司的企业管治水平相应
的图像，每个公司至少有 60 个特征被送往考虑。评级的指标变量是从以下几个
方面：
1. 债务重组协议和监事会能力，独立董事成员
2. 党有关的交易监督
3. 一般股东大会的可达性
4. 运作管理委员会
5. 审计的独立性
6. 反收购防御的缺乏性
7. 购买自己股票的规定
8. 公司声明的目标和意图
9. 透明安排，包括信息公司的网站查阅。
公开披露的评级有五类，从 A-到 C+。这些被使用到 53 家公司。作者指出，
信用评级的门类齐全，包括从 A 至 E 的等级评分。有五家公司的评级为 C+，17
家是 B-的评级，另外还有 14 家是 B，13 家是 B+， 4 家是 A-。
计量经济学与公司治理
表 1：2004 年在华尔街上市的公司治理评级
对公司治理的经济学研究涉及各方的代表公司各类的业绩水平和变量描述
的治理水平之间的主要关系，如股权结构，或监督机构的组成。这些研究调查多
次被提出，例如，在 Bhagat 和 Jefferis(2002)，Boersch-Supan 和 Koeke（2002）
A- Amica, Agora, BZWBK, Orbis
B+ BPH-PBK, BRE, Computerland, Eldorado, FORTE, KGHM, LPP,
Netia, Pekao, Polfa Kutno, Prokom, TPSA, Stomil Sanok
B Debica, Elektrim, Groclin, Impexmetal, INGBSK, Jelfa, Kety, Kruszwica,
Mennica, PGF, PKNOrlen, Polifarb CW, Rafako, Softbank
B- Bank Handlowy, BIG BG (Millenium), Budimex, Comarch, Farmacol,
EFL, Grajewo, Kogeneracja, Kredyt Bank, Krosno, Lentex, Mostostal
SDL, Okocim,Orfe, Rolimpex, Sokolow, Sterprojekt
C+ Cersanit, Echo, Hoop, Swiecie, Z ˙ ywiec
和 Gugler（2001）。影响公司治理水平主要因素，通常认为在实证研究中被证实，
主要有：
1. 监事会的构成（对管理层，董事会的结构依赖度）
2. 所有权结构（稀释后的所有权，所有权集中，企业业主，机构投资人，
管理权）
3. 收购（包括经理收购）和首席执行官变化
4. 产权结构（债务结构）
5. 经理薪酬
Boersch-Supan 和 Koeke（2002）制定的关于公司薄弱治理的计量研究，有
这几个问题：
1. 反向结构因果关系：例如，目前还不清楚股权结构与公司业绩的因果关
系的方向是什么，较集中的所有权可以提高公司绩效，但反向关系也可能广受市
场评估，也可以吸引投资者。
2. 缺失变量：在公司治理方面的习惯，主要是解释变量可能不被纳入模型
中；此外，该方程 linear3.的规格排除了高阶项的存在。
3. 样品选择性：大多数公司治理的实证研究通常只分析上市的大公司，这
样的样本会选择有效的业绩变量和选择研究样本偏差。
4. 测量中的变量错误：例如，该公司的表现可通过不同的变量，就像市场
价值，资产报酬率，净资产收益率，税前利润，托宾 Q 值；这些变量有时是不
相关的，也就是说，他们用不同的方式衡量相同的性能。
要正确处理在公司治理中的这些计量经济模型的危险，必须使用公司的面板
部分数据。Boersch-Supan 和 Koeke（2002）说明了德国几个这样的数据库。
公司治理与经济绩效
最近在经合组织国家对公司治理的调查(企业管理，2004)表明，学习那些计
量技术被认为是最好的实践，而计量经济学方法是公司治理性能的一个重要决定
因素。三项研究显示，公司治理水平和企业的经营状况之间存在着相关性。
在另一项研究，Bøhren 和 Ødegaard（2001）分析了在奥斯陆证券交易所上
市的公司在 1989 到 1997 年的托宾 Q 值和代表公司治理的几个变量之间的关系
（到 1997 年底的 217 家公司）。他们的研究结果表明，例如，股权集中度对性能
产生负面影响，而内部人持股的影响是积极的。提高董事会规模能减少股份使用
性能。直接所有权在性能上的影响要远远大于对机构或国家所有制的影响。
类似的结果被发表于一份 1991 年到 1996 年由雷曼兄弟和韦根（2000）对于 361
德国企业的抽样调查中。在他们的研究，公司的业绩由资产报酬率来衡量相关的
公司治理变数。该小组回归结果证实，股权集中度对资产报酬率有负面影响，而
所有权集中的积极影响是作为大股东的公司和金融机构发现的。
构建了各国公司治理的指标，也已审查了其与经济和市场的公司业绩。这些
例子被发现于 Gruszczynski (2003)。在下一节中，将尝试实现在企业之间的治理
评级和财务绩效的关系。
排序多元化
公司治理评级是一个典型的因变量，适合有序反应模型。设置升序排名的类
别和邻国之间距离的类别并不相同。对于假设的论述，我们所观察到的公司治理
评级变量 y 可以等于五个等级类别之一，如在 2003 年波兰出现的评级。据推测，
这个有序变量 y 是关系到连续潜变量 y *，它表示该公司的企业管治的程度。这
个 y *是已知的。
y*的值确定了以下方式 y 为代表的结果：
yi =1 or C+ ifτ0≤yi * <τ1
yi =2 or B- ifτ1≤yi * <τ2
yi =3 or B ifτ2≤yi * <τ3
yi =4 or B+ ifτ3≤yi * <τ4
yi =5 or A- ifτ4≤yi * <τ5
这个τ的值是阈值（切点），它的值相当于τ0 =-∞ 和τ5=∞。
这个没有确定的线形模型如下：
yi *=xiτβ+εi
其中 xi 是典型的[(k + 1) *1]向量的 k 解释变量（加长期常数）为第 i 个的值
（第 i 个公司）。
观测到的 y 值的使用需要对误差分布的εi 进行假设。设 F 是累积分布函数
（CDF），概率 y=m （m= 1，...，5）可以写成：
P(yi=m│xi, β, τ)=F(τm- xiτβ)-F(τm-1- xiτβ)
其中τ为切点向量。在方程的P（yi=m│xi, β, τ），使右边的第二项等于0。此
外，在P（yi=m│xi, β, τ），第一项等于1。
对于有序反应模型，误差εi 的分布式通常用均值 0 和方差为 1 来表示。对于
排序多元反应模型，误差ε依据物流来配送。对于模型的识别，无论是常数项还
是τ1 设定为 0 的缘故。在有序反应中：
F(τm- xiτβ)=eτm- xiτβ/(1+ eτm- xiτβ)
第 i 公司的 y 值的每个类别的概率（i=│1,2 ,..., n）是相等的：
P(yi=1│xi, β, τ) if y=1
…
pi= P(yi=m│xi, β, τ) if y=m
…
P(yi=5│xi, β, τ) if y=5
在 Pi 的影响下，形成了相似函数，其最大值达到了分母向量β，τ的 ML 估计
值。
样品和模型
为了更好地解释波兰上市公司的公司治理评级 y 的目的，以及有序反应模型
的指定。2004 年评级是建立在 2003 年 11 月的法律地位的基础之上，以 2002 年
财务绩效的解释变量计算的财务报表为基础。
样本包括 37 家公司中的 53 个评级。多达 16 个公司已经被排除在样本之外：
第八银行，其他金融机构之一，2002 年 1 月 1 日后两个首次报价公司，与五家
排除了其他原因的公司（数据无法访问，缺少信息的财务报表）。
内生变数 y 已被定义为：
股东变量，以及在样本中出现的所有五个等级类别；
第一大股东变量，只有三个类别（前两个和最后两个已被合并）。
为第一大股东简单排名的原因是与小规模样品相比，参数的数量在有序反应
模型中是估计值。
有 20 个解释变量已被考虑作为财务比率。对公司 2002 年财务比率计算如下：
盈利能力比率：P102 销售利润率，P202 营业利润率，P302 毛利率，P402
净利率，ROE02 股东权益回报，ROA02 总资产回报利润。
流动比率：L102 流动比率，L202 速动比率。
速动比率：A102 款项周转率，A202 存货周转率，A302 营业周期，A402 负
债额，A502 现金转换周期，A602 流动资产周转率，A702 资产和营业额。
负债比率：D102 固定资产负债率，D202 债务保证金，D302 的息税前利润/
财务费用，D402 债务/息税前利润。
出 处 :Markek Gruszczynski. 波 兰 的 公 司 治 理 与 财 务 绩 效 . 国 际 经 济 发 展 研
究,2006(12):251-259.
