这是用户在 2025-1-10 24:14 为 https://app.immersivetranslate.com/pdf-pro/39022a89-5821-400c-ae7f-1d5ce02cb310 保存的双语快照页面,由 沉浸式翻译 提供双语支持。了解如何保存?

DBAP 5220
Research on Managerial
Accounting, Corporate
Governance, and Executive
DBAP 5220 管理会计、公司治理与高管研究

This material (Reference Number: HCP 2400 53UU) has been copied in accordance with the terms of a licence granted by The Hong Kong Reprographic Rights Licencing Society Limited
本材料(参考编号:HCP 2400 53UU)是根据香港复印权许可协会有限公司授予的许可条款复制的
You are not permitted to make any further copy of this work, or to make it available to others. No re-sale is permitted.
您不被允许对本作品进行任何进一步的复制或将其提供给他人。禁止转售。

Prof Tai-Yuan CHEN  陈泰元教授

DBM
The Hong Kong University of Science and Technology
香港科技大学

January 2025  2025 年 1 月

This course pack has been produced under licences granted by:
本课程包是根据以下授予的许可证制作的:

The Hong Kong Reprographic Rights Licensing Society Limited
香港复印权许可协会有限公司

(Reference Number: HCP 2400 53UU)
(参考编号:HCP 2400 53UU)

Twenty-five years of corporate governance research. . . and counting, 2001
二十五年的公司治理研究……还在继续,2001

Understanding the determinants of managerial ownership and the link between ownership and performance, 1999
理解管理层持股的决定因素以及持股与绩效之间的联系,1999

Wealth and Executive Compensation?, 2006
财富与高管薪酬?, 2006

Do CEOs Matter? Evidence from Hospitalization Events, 2020
首席执行官重要吗?来自 2020 年住院事件的证据
You are not permitted to make any further copy of this work, or to make it available to others. No re-sale is permitted
您不被允许对本作品进行任何进一步的复制或将其提供给他人。禁止转售。

Understanding the determinants of managerial ownership and the link between ownership and performance sh sh ^("sh"){ }^{\text {sh}}
理解管理层持股的决定因素以及持股与绩效之间的联系 sh sh ^("sh"){ }^{\text {sh}}

Charles P. Himmelberg a ^("a "){ }^{\text {a }}, R. Glenn Hubbard a,b,* a,b,*  ^("a,b,* "){ }^{\text {a,b,* }}, Darius Palia a,c a,c  ^("a,c "){ }^{\text {a,c }}
查尔斯·P·希梅尔伯格 a ^("a "){ }^{\text {a }} , R·格伦·哈伯德 a,b,* a,b,*  ^("a,b,* "){ }^{\text {a,b,* }} , 达里乌斯·帕利亚 a,c a,c  ^("a,c "){ }^{\text {a,c }}
a a ^(a){ }^{\mathrm{a}} Graduate School of Business, Columbia University, Uris Hall, 3022 Broadway, New York, NY 10027 USA
哥伦比亚大学商学院,乌里斯大厅,3022 百老汇,纽约,NY 10027 美国
b b ^(b){ }^{\mathrm{b}} The National Bureau of Economic Research, USA
b b ^(b){ }^{\mathrm{b}} 美国国家经济研究局
c 1050 1050 ^("c ")1050{ }^{\text {c }} 1050 Massachusetts Avenue, Cambridge, MA 02138, USA
c 1050 1050 ^("c ")1050{ }^{\text {c }} 1050 马萨诸塞州大道,剑桥,MA 02138,美国

Received 9 March 1998; received in revised form 19 October 1998; accepted 2 March 1999
收到日期:1998 年 3 月 9 日;修订版收到日期:1998 年 10 月 19 日;接受日期:1999 年 3 月 2 日

Abstract  摘要

Both managerial ownership and performance are endogenously determined by exogenous (and only partly observed) changes in the firm’s contracting environment. We extend the cross-sectional results of Demsetz and Lehn (1985) (Journal of Political Economy, 93, 1155-1177) and use panel data to show that managerial ownership is explained by key variables in the contracting environment in ways consistent with the predictions of principal-agent models. A large fraction of the cross-sectional variation in managerial ownership is explained by unobserved firm heterogeneity. Moreover, after
管理层所有权和绩效是由外部(且仅部分可观察的)变化在公司的契约环境中内生决定的。我们扩展了 Demsetz 和 Lehn(1985)的横截面结果(《政治经济学杂志》,93,1155-1177),并使用面板数据表明,管理层所有权是由契约环境中的关键变量解释的,这与委托-代理模型的预测一致。管理层所有权的横截面变异中有很大一部分是由未观察到的公司异质性解释的。此外,在此之后

controlling both for observed firm characteristics and firm fixed effects, we cannot conclude (econometrically) that changes in managerial ownership affect firm performance. © 1999 Elsevier Science S.A. All rights reserved.
在控制了观察到的公司特征和公司固定效应后,我们无法得出(计量经济学上)管理层持股的变化会影响公司绩效的结论。© 1999 Elsevier Science S.A. 保留所有权利。
JEL classification: G14; G32; D23; L14; L22
JEL 分类:G14;G32;D23;L14;L22

Keywords: Managerial ownership; Corporate governance
关键词:管理层持股;公司治理

1. Introduction  1. 引言

Since Berle and Means (1932), the conflict between managers and shareholders has been studied extensively by researchers seeking to understand the nature of the firm. When shareholders are too diffuse to monitor managers, corporate assets can be used for the benefit of managers rather than for maximizing shareholder wealth. It is well known that a solution to this problem is to give managers an equity stake in the firm. Doing so helps to resolve the moral hazard problem by aligning managerial interests with shareholders’ interests. Therefore, Jensen and Meckling (1976) suggest that managers with small levels of ownership fail to maximize shareholder wealth because they have an incentive to consume perquisites. In a similar fashion, some commentators have decried low levels of managerial ownership in U.S. corporations, and the theme has even appeared in discussions by compensation specialists and boards of directors.
自从 Berle 和 Means(1932)以来,研究人员广泛研究了管理者与股东之间的冲突,以理解公司的本质。当股东过于分散以至于无法监控管理者时,企业资产可能会被用于管理者的利益,而不是为了最大化股东财富。众所周知,解决这个问题的一种方法是让管理者在公司中拥有股权。这样做有助于通过将管理者的利益与股东的利益对齐来解决道德风险问题。因此,Jensen 和 Meckling(1976)建议,拥有少量股份的管理者未能最大化股东财富,因为他们有动机消费特权。以类似的方式,一些评论者对美国公司中管理层持股比例低表示谴责,这一主题甚至出现在薪酬专家和董事会的讨论中。
In this paper, we propose an equilibrium interpretation of the observed differences in ownership structures across firms. Rather than interpret low ownership levels as per se evidence of suboptimal compensation design, we argue that the compensation contracts observed in the data are endogenously determined by the contracting environment, which differs across firms in both observable and unobservable ways. In particular, low levels of managerial ownership might well be the optimal incentive arrangement for the firm if the scope for perquisite consumption (or more generally, the severity of the moral hazard problem for managers) happens to be low for that firm. We do not deny the importance of agency problems between stockholders and managers, but rather emphasize the importance of unobserved heterogeneity in the contracting environment across firms.
在本文中,我们提出了对公司间所有权结构差异的平衡解释。我们认为,低所有权水平不应被视为补偿设计不理想的直接证据,而是认为数据中观察到的补偿合同是由合同环境内生决定的,而这一环境在可观察和不可观察的方面在不同公司之间存在差异。特别是,如果某公司的特权消费范围(或更一般地说,管理者的道德风险问题的严重性)恰好较低,那么低水平的管理者所有权可能是该公司的最佳激励安排。我们并不否认股东与管理者之间代理问题的重要性,而是强调不同公司之间合同环境中未观察到的异质性的重要性。
We begin by examining the observable determinants of managerial ownership. This investigation builds upon Demsetz and Lehn (1985), who use crosssectional data to show that the level of managerial ownership is determined by the riskiness of the firm, measured by the volatility of the stock price. They argue that the scope for moral hazard is greater for managers of riskier firms, which therefore means that those managers must have greater ownership stakes to align incentives. They also point out that riskiness makes it costlier for managers
我们首先考察管理层持股的可观察决定因素。这项研究基于 Demsetz 和 Lehn(1985)的工作,他们使用横截面数据表明,管理层持股的水平由公司的风险性决定,风险性通过股票价格的波动性来衡量。他们认为,风险较高的公司的管理者面临的道德风险更大,因此这些管理者必须拥有更大的持股比例以对齐激励。他们还指出,风险性使得管理者的成本更高。

to hold nondiversified portfolios (assuming that equity holdings in the firm are not easily hedged), so the relation between managerial ownership and nondiversifiable stock price risk is not necessarily monotonic.
持有非多样化投资组合(假设公司股权不易对冲),因此管理层持股与不可多样化股票价格风险之间的关系不一定是单调的。
To document the extent to which managerial ownership is endogenously determined by the contracting environment, we extend the empirical specification used by Demsetz and Lehn by including a number of additional explanatory variables other than stock price variability (see also Kole, 1996). Most importantly, we include variables (such as firm size, capital intensity, R&D intensity, advertising intensity, cash flow, and investment rate) designed to control for the scope for moral hazard. To the extent that our additional explanatory variables proxy for moral hazard, our specification clarifies the role of stock price variance as an explanatory variable for managerial ownership. We also use panel data that allow us to estimate the importance of unobserved (time-invariant) firm effects. These results show that a large fraction of the cross-sectional variation in managerial ownership is ‘explained’ by unobserved firm heterogeneity. In our subsequent analysis of the determinants of firm value, we argue that this unobserved heterogeneity generates a spurious correlation between ownership and performance.
为了记录管理层持股在多大程度上是由契约环境内生决定的,我们通过包括除股票价格波动之外的一些额外解释变量,扩展了 Demsetz 和 Lehn 使用的实证规范(参见 Kole,1996)。最重要的是,我们包括了旨在控制道德风险范围的变量(如公司规模、资本密集度、研发密集度、广告密集度、现金流和投资率)。在我们的额外解释变量代理道德风险的程度上,我们的规范阐明了股票价格波动作为管理层持股解释变量的作用。我们还使用了面板数据,使我们能够估计未观察到的(时间不变的)公司效应的重要性。这些结果表明,管理层持股的横截面变异中有很大一部分是由未观察到的公司异质性“解释”的。在我们后续对公司价值决定因素的分析中,我们认为这种未观察到的异质性产生了持股与绩效之间的虚假相关性。
The second goal of this paper is to reexamine theoretical explanations of the empirical link between managerial ownership and firm performance. Mørck et al. (1988) estimate a piecewise-linear relation between board ownership and Tobin’s Q Q QQ and find that Tobin’s Q Q QQ increases and then decreases with managerial ownership. McConnell and Servaes (1990) examine a larger data set than the Fortune 500 firms examined by Mørck et al. and find an inverted U-shaped relation between Q Q QQ and managerial ownership, with an inflection point between 40 % 40 % 40%40 \% and 50 % 50 % 50%50 \% ownership. Hermalin and Weisbach (1991) analyze 142 NYSE firms and find that Q Q QQ rises with ownership up to a stake of 1 % 1 % 1%1 \%; the relation is negative in the ownership range of 1 5 % 1 5 % 1-5%1-5 \%, becomes positive again in the ownership range of 5 20 % 5 20 % 5-20%5-20 \%, and turns negative for ownership levels exceeding 20 % 20 % 20%20 \%. The pattern identified by Mørck et al. has been corroborated for a cross-section of U.S. firms from 1935 by Holderness et al. (1999). Kole (1995) examines the differences in data sources used in several recent studies and concludes that differences in firm size can account for the reported differences between those studies. These studies generally interpret the positive relation at low levels of managerial ownership as evidence of incentive alignment, and the negative relation at high levels of managerial ownership as evidence that managers become ‘entrenched’ and can indulge in non-value-maximizing activities without being disciplined by shareholders. However, these studies do not address the endogeneity problem that confronts the use of managerial ownership as an explanatory variable, a problem noted early by Jensen and Warner (1988, p. 13).
本文的第二个目标是重新审视管理层持股与公司绩效之间经验联系的理论解释。Mørck 等人(1988)估计了董事会持股与托宾的 Q Q QQ 之间的分段线性关系,并发现托宾的 Q Q QQ 随着管理层持股的增加而先上升后下降。McConnell 和 Servaes(1990)研究了一个比 Mørck 等人研究的财富 500 强公司更大的数据集,发现 Q Q QQ 与管理层持股之间存在倒 U 形关系,拐点位于 40 % 40 % 40%40 \% 50 % 50 % 50%50 \% 的持股之间。Hermalin 和 Weisbach(1991)分析了 142 家纽约证券交易所公司,发现 Q Q QQ 随着持股上升至 1 % 1 % 1%1 \% 的股份而上升;在 1 5 % 1 5 % 1-5%1-5 \% 的持股范围内关系为负,在 5 20 % 5 20 % 5-20%5-20 \% 的持股范围内再次变为正,并且在超过 20 % 20 % 20%20 \% 的持股水平时变为负。Mørck 等人识别的模式得到了 Holderness 等人(1999)对 1935 年以来美国公司横截面的证实。Kole(1995)研究了几项近期研究中使用的数据来源的差异,并得出结论认为公司规模的差异可以解释这些研究之间报告的差异。 这些研究通常将低水平的管理层持股与激励一致性之间的正相关关系解释为证据,而将高水平的管理层持股与管理者变得“固守”并可以沉迷于不追求价值最大化的活动而不受股东约束的负相关关系解释为证据。然而,这些研究并没有解决将管理层持股作为解释变量所面临的内生性问题,这一问题早在 Jensen 和 Warner(1988 年,第 13 页)就已被提及。
We investigate the degree to which this heterogeneity makes managerial ownership an endogenous variable in models of firm performance. Following in the tradition of Demsetz and Lehn, we describe the contracting problem faced
我们研究这种异质性在多大程度上使管理层持股成为公司绩效模型中的内生变量。沿袭 Demsetz 和 Lehn 的传统,我们描述了面临的契约问题。

by the firm and develop a simple empirical model to illustrate the econometric issues that are encountered when estimating the relation among managerial ownership, its determinants, and its effect on firm performance. Distinct from Demsetz and Lehn, and in contrast to previous papers that attempt to measure the impact of managerial ownership on firm performance, we use panel data to test for the endogeneity of managerial ownership in models linking ownership to performance (measured by Tobin’s Q Q QQ ). In particular, we use panel data to investigate the hypothesis that managerial ownership is related to observable and unobservable (to the econometrician) firm characteristics influencing contracts. If the unobserved sources of firm heterogeneity are relatively constant over time, we can treat these unobserved variables as fixed effects, and use panel data techniques to obtain consistent estimates of the parameter coefficients. This approach provides consistent estimates of the residuals in the Q Q QQ regression, which we use to construct a test for correlation between managerial ownership and unobserved firm heterogeneity.
由公司提出并开发一个简单的实证模型,以说明在估计管理层持股、其决定因素及其对公司绩效影响之间的关系时遇到的计量经济学问题。与 Demsetz 和 Lehn 不同,并且与之前试图衡量管理层持股对公司绩效影响的论文相对,我们使用面板数据来检验管理层持股在将持股与绩效(以 Tobin’s Q Q QQ 衡量)联系起来的模型中的内生性。特别是,我们使用面板数据来研究管理层持股与影响合同的可观察和不可观察(对计量经济学家而言)公司特征相关的假设。如果公司异质性的未观察来源在时间上相对恒定,我们可以将这些未观察变量视为固定效应,并使用面板数据技术获得参数系数的一致估计。这种方法提供了 Q Q QQ 回归中残差的一致估计,我们用它来构建管理层持股与未观察公司异质性之间相关性的检验。
Our principal findings are threefold. First, proxies for the contracting environment faced by the firm (i.e., observable firm characteristics) strongly predict the structure of managerial ownership. We substantially extend the set of explanatory variables examined by Demsetz and Lehn, and we show that many of our results are robust to the inclusion of observed determinants of managerial ownership, industry fixed effects, or firm fixed effects. Second, we show that the coefficient on managerial ownership is not robust to the inclusion of fixed effects in the regression for Tobin’s Q Q QQ. Our formal statistical test rejects the null hypothesis of a zero correlation between managerial ownership and the unobserved determinants of Tobin’s Q Q QQ, thus supporting our conjecture that managerial ownership is endogenous in Q Q QQ regressions. That is, managerial ownership and firm performance are determined by common characteristics, some of which are unobservable to the econometrician. Third, we explore the use of instrumental variables as an alternative to fixed effects to control for the endogeneity of managerial ownership in the Q Q QQ regression. We find some evidence to support a causal link from ownership to performance, but this evidence is tentative because of the weakness of our instruments. We argue that future progress will require a more structural approach to the model.
我们的主要发现有三点。首先,企业面临的合同环境的代理变量(即可观察的企业特征)强烈预测管理层持股的结构。我们大幅扩展了 Demsetz 和 Lehn 所研究的解释变量集,并且我们表明,许多结果在包括管理层持股的观察决定因素、行业固定效应或企业固定效应时仍然稳健。其次,我们表明,在 Tobin’s Q Q QQ 的回归中,管理层持股的系数对固定效应的包含并不稳健。我们的正式统计检验拒绝了管理层持股与 Tobin’s Q Q QQ 的未观察决定因素之间零相关的原假设,从而支持了我们关于管理层持股在 Q Q QQ 回归中是内生的猜想。也就是说,管理层持股和企业绩效是由共同特征决定的,其中一些特征对计量经济学家是不可观察的。第三,我们探讨了使用工具变量作为固定效应的替代方案,以控制 Q Q QQ 回归中管理层持股的内生性。 我们发现了一些证据支持所有权与绩效之间的因果关系,但由于我们工具的弱点,这些证据是初步的。我们认为,未来的进展将需要对模型采取更结构化的方法。
Kole (1996) also argues that managerial ownership is endogenous; she further argues that causality operates in the opposite direction, from performance to ownership. Using a panel-data vector autoregression, we corroborate Kole’s reverse causality evidence (results available upon request). Our research, however, supports the idea that both ownership and performance are determined by similar (observed and unobserved) variables in the firm’s contracting environment. Thus, our interpretation is different from Kole’s interpretation. That is, we find evidence endogeneity caused by unobserved heterogeneity as opposed to reverse causality.
Kole (1996) 还认为管理层持股是内生的;她进一步认为因果关系是相反的,从绩效到所有权。通过使用面板数据向量自回归,我们证实了 Kole 的逆向因果关系证据(结果可根据请求提供)。然而,我们的研究支持这样一种观点,即所有权和绩效都是由公司合同环境中相似的(可观察和不可观察的)变量决定的。因此,我们的解释与 Kole 的解释不同。也就是说,我们发现证据表明内生性是由不可观察的异质性引起的,而不是逆向因果关系。
The paper is organized as follows. In Section 2, we outline a simple model of managerial ownership and explain why it is difficult to estimate the relation between managerial ownership levels and firm performance, particularly in the context of cross-sectional data. Section 3 describes the sample selection criteria and the data we use in our empirical analysis of managerial ownership and firm performance. In Sections 4 and 5, respectively, we present empirical evidence on the determinants of managerial ownership and on the relation between managerial ownership and firm performance. Section 6 concludes.
本文的组织结构如下。在第二节中,我们概述了一个简单的管理层持股模型,并解释了为什么在横截面数据的背景下,估计管理层持股水平与公司绩效之间的关系是困难的。第三节描述了样本选择标准以及我们在管理层持股与公司绩效的实证分析中使用的数据。在第四节和第五节中,我们分别提供了关于管理层持股决定因素的实证证据以及管理层持股与公司绩效之间关系的实证证据。第六节为结论。

2. An empirical framework for analyzing executive contracts
2. 分析高管合同的实证框架

A common approach for estimating the impact of managerial ownership on firm value is to regress Tobin’s Q Q QQ on such variables as the percentage of equity held by managers. In this section, we argue that this regression is potentially misspecified because of the presence of unobserved heterogeneity. Specifically, if some of the unobserved determinants of Tobin’s Q Q QQ are also determinants of managerial ownership, then managerial ownership might spuriously appear to be a determinant of firm performance. To motivate our focus on the endogeneity of managerial ownership, we provide three examples of likely sources of unobservable heterogeneity, and in each case, we discuss their econometric consequences for cross-sectional regressions. We follow this discussion with a more formal exposition, in which we assume that the unobserved heterogeneity is a ‘firm fixed effect’, and we show how, under this assumption, panel data can be used to mitigate the endogeneity problem. In Section 5, we return to this model to describe a test for the endogeneity of ownership in regressions for Tobin’s Q Q QQ.
估计管理层持股对公司价值影响的一个常见方法是将托宾的 Q Q QQ 对管理者持股比例等变量进行回归。在本节中,我们认为这种回归可能存在模型设定错误,因为存在未观察到的异质性。具体而言,如果一些未观察到的托宾的 Q Q QQ 决定因素也是管理层持股的决定因素,那么管理层持股可能会虚假地显现为公司绩效的决定因素。为了引导我们关注管理层持股的内生性,我们提供了三个可能的未观察到的异质性来源的例子,并在每种情况下讨论它们对横截面回归的计量经济学影响。我们在此讨论之后进行了更正式的阐述,在此我们假设未观察到的异质性是“公司固定效应”,并展示在这一假设下,如何使用面板数据来减轻内生性问题。在第 5 节中,我们回到这个模型,描述一个关于托宾的 Q Q QQ 回归中所有权内生性的检验。
For our first example of unobserved heterogeneity, consider two firms that are identical except that the owner of one of the firms has access to a superior monitoring technology. Under the optimal contracting regime, the owners with access to the superior monitoring technology will choose a lower level of managerial ownership to align incentives, and this firm will have a higher valuation because fewer resources will be diverted to managerial perquisites. If measures of the quality of the monitoring technology are omitted from the specification, a regression of firm value on managerial ownership will spuriously (and falsely) indicate a negative relation, because ownership is a negative proxy for the quality of monitoring technology.
在我们未观察到的异质性的第一个例子中,考虑两家完全相同的公司,除了其中一家公司的所有者拥有更优越的监控技术。在最佳契约制度下,拥有更优越监控技术的所有者将选择较低的管理层持股比例以对齐激励,而这家公司将具有更高的估值,因为更少的资源会被转移到管理特权上。如果在规范中省略了监控技术质量的衡量指标,对公司价值与管理层持股比例的回归分析将错误地(并且虚假地)表明存在负相关关系,因为持股比例是监控技术质量的负代理。
Intangible assets provide a second example of unobserved firm heterogeneity. Suppose two firms are identical except that one of the firms operates with a higher fraction of its assets in the form of intangibles. Under the optimal contracting regime, the owners of this firm will require a higher level of managerial ownership to align incentives because the intangible assets are
无形资产提供了未观察到的公司异质性的第二个例子。假设两家公司是相同的,只是其中一家公司以更高比例的无形资产运营。在最佳契约制度下,这家公司的所有者将需要更高水平的管理层持股以对齐激励,因为无形资产是

harder to monitor and therefore subject to managerial discretion. This firm will also have a higher Q Q QQ value because the market will value intangibles in the numerator (market value), but the book value of assets in the denominator will understate the value of intangibles (because Tobin’s Q Q QQ is measured as the ratio of the market value of the firm’s outstanding debt and equity divided by the book value of assets). In this example, the unobserved level of intangibles induces a positive correlation between managerial ownership and Tobin’s Q Q QQ, but this relation is spurious, not causal.
更难以监控,因此受到管理层的自由裁量权。这家公司也会有更高的 Q Q QQ 值,因为市场会在分子中评估无形资产(市场价值),但资产的账面价值在分母中会低估无形资产的价值(因为托宾的 Q Q QQ 是通过公司的流通债务和股本的市场价值与资产的账面价值的比率来衡量的)。在这个例子中,无形资产的未观察水平导致管理层持股与托宾的 {{2 }} 之间存在正相关关系,但这种关系是虚假的,而不是因果关系。

A third example of unobserved heterogeneity is variation in the degree of market power. Suppose there are two firms competing in a market with differentiated products and that one firm enjoys a competitive advantage because (for some historical reason) it has been able to locate its products in such a way that confers more market power. If this market power insulates managerial decision-making from the discipline of competitive product markets, then the optimal contract for managers will call for higher levels of managerial ownership. Hence, unobserved heterogeneity in the form of unobserved differences in market power will (spuriously) induce a positive relation between ownership and performance. Alternatively, causation could run the other way; stockholders might design the manager’s compensation to implicitly encourage collusive outcomes in the product market (Fershtman and Judd, 1987). Attempting to test this proposition using regressions of Tobin’s Q Q QQ on managerial ownership suffers from the same econometric problems we study here. The ownership decision is endogenous because of unobserved firm heterogeneity.
未观察到的异质性的第三个例子是市场力量程度的变化。假设有两家公司在一个具有差异化产品的市场中竞争,其中一家公司由于某种历史原因享有竞争优势,能够以一种赋予更多市场力量的方式定位其产品。如果这种市场力量使管理决策免受竞争产品市场的约束,那么对管理者的最佳合同将要求更高的管理者持股比例。因此,未观察到的异质性以未观察到的市场力量差异的形式将(虚假地)引发所有权与绩效之间的正相关关系。或者,因果关系也可能反向运行;股东可能会设计管理者的薪酬,以隐含地鼓励产品市场中的共谋结果(Fershtman 和 Judd,1987)。尝试使用对管理者持股比例的 Tobin’s Q Q QQ 回归来检验这一命题,面临着我们在这里研究的相同计量经济学问题。由于未观察到的公司异质性,所有权决策是内生的。
It is possible to generalize these examples in a simple analytical framework. We assume that within the general set of contracts agreed to by the firm, the owners of the firm choose a simple management compensation contract that includes a share of the firm’s equity. This equity share (or ‘managerial stake’) is chosen to maximize the owners’ equity return subject to incentive compatibility and participation constraints. For this purpose, we assume that gains from other means for reducing agency costs have been maximized, so that we examine the residual agency cost to be addressed by managerial ownership (we revisit this assumption in Section 5 below). Let x i t x i t x_(it)x_{i t} and u i t u i t u_(it)u_{i t}, respectively, denote observable and unobservable characteristics for firm i i ii at time t t tt related to the firm’s contracting environment (including, e.g., proxies for the potential for moral hazard). In addition to unobserved firm characteristics, we implicitly assume a profitability shock that is observable to the manager, but not to outside shareholders. This shock cannot be contracted upon, giving rise to moral hazard.
可以在一个简单的分析框架中对这些例子进行概括。我们假设在公司达成的一般合同集中,公司的所有者选择一个简单的管理薪酬合同,其中包括公司股权的一部分。这个股权份额(或称“管理者股份”)的选择是为了在激励兼容性和参与约束的条件下最大化所有者的股权回报。为此,我们假设通过其他手段降低代理成本的收益已经最大化,因此我们考察由管理层持股所需解决的剩余代理成本(我们将在下面的第 5 节中重新审视这一假设)。设 x i t x i t x_(it)x_{i t} u i t u i t u_(it)u_{i t} 分别表示与公司 i i ii 在时间 t t tt 相关的可观察和不可观察特征,这些特征与公司的合同环境有关(包括例如道德风险潜在的代理)。除了未观察到的公司特征外,我们还隐含假设存在一个对管理者可观察但对外部股东不可观察的盈利冲击。这个冲击无法通过合同进行约定,从而产生道德风险。
The firm’s owners must decide how much equity to give to managers in order to align incentives for value maximization. This equity share m i t m i t m_(it)m_{i t} depends on such factors as the potential for moral hazard and managers’ exposure to risk, which we assume are partly measured by x i t x i t x_(it)x_{i t}, but are otherwise unobserved and
公司的所有者必须决定给予管理者多少股权,以便使激励与价值最大化保持一致。这部分股权 m i t m i t m_(it)m_{i t} 取决于道德风险的潜在性和管理者的风险暴露等因素,我们假设这些因素部分通过 x i t x i t x_(it)x_{i t} 来衡量,但其他方面则是不可观察的。

included in u i t u i t u_(it)u_{i t}. We assume that the functional relation is linear, and that u i t = u i u i t = u i u_(it)=u_(i)u_{i t}=u_{i} is time-invariant for the firm, so that
包含在 u i t u i t u_(it)u_{i t} 中。我们假设功能关系是线性的,并且 u i t = u i u i t = u i u_(it)=u_(i)u_{i t}=u_{i} 对于公司是时间不变的,因此
m i t = β 1 x i t + γ 1 u i + e i t , m i t = β 1 x i t + γ 1 u i + e i t , m_(it)=beta_(1)x_(it)+gamma_(1)u_(i)+e_(it),m_{i t}=\beta_{1} x_{i t}+\gamma_{1} u_{i}+e_{i t},
where e i t e i t e_(it)e_{i t} represents independent measurement error.
其中 e i t e i t e_(it)e_{i t} 代表独立测量误差。

Faced with this contract, managers choose an optimal ‘effort level’, y i t y i t y_(it)y_{i t}, which could include a range of participation in non-value-maximizing activities. This effort choice depends on the managerial ownership stake, m i t m i t m_(it)m_{i t}, and, like the optimal contract itself, depends on both observed and unobserved characteristics of the firm, x i t x i t x_(it)x_{i t} and u i u i u_(i)u_{i}. Assuming a linear functional form, we can represent the manager’s effort choice by the following relation:
面对这一合同,管理者选择一个最佳的“努力水平”, y i t y i t y_(it)y_{i t} ,这可能包括参与一系列非价值最大化活动。这个努力选择取决于管理者的所有权份额, m i t m i t m_(it)m_{i t} ,并且像最佳合同本身一样,取决于公司的观察到的和未观察到的特征, x i t x i t x_(it)x_{i t} u i u i u_(i)u_{i} 。假设线性函数形式,我们可以通过以下关系表示管理者的努力选择:
y i t = θ m i t + β 2 x i t + γ 2 u i + v i t . y i t = θ m i t + β 2 x i t + γ 2 u i + v i t . y_(it)=thetam_(it)+beta_(2)x_(it)+gamma_(2)u_(i)+v_(it).y_{i t}=\theta m_{i t}+\beta_{2} x_{i t}+\gamma_{2} u_{i}+v_{i t} .
Using firm value as a summary measure of expected firm performance, we assume that firm value depends on managerial effort plus the vector of observed and unobserved firm characteristics. Denoting the value of firm i i ii at time t t tt by Q i t Q i t Q_(it)Q_{i t}, we assume that
将公司价值作为预期公司绩效的总结指标,我们假设公司价值依赖于管理努力加上观察到的和未观察到的公司特征向量。我们用 Q i t Q i t Q_(it)Q_{i t} 表示时间 t t tt 时公司 i i ii 的价值,我们假设
Q i t = δ y i t + β 3 x i t + γ 3 u i + w i t . Q i t = δ y i t + β 3 x i t + γ 3 u i + w i t . Q_(it)=deltay_(it)+beta_(3)x_(it)+gamma_(3)u_(i)+w_(it).Q_{i t}=\delta y_{i t}+\beta_{3} x_{i t}+\gamma_{3} u_{i}+w_{i t} .
We can now combine Eqs. (2) and (3) to derive the following relation among firm managerial ownership, firm characteristics, and firm performance:
我们现在可以将方程(2)和(3)结合起来,推导出公司管理层持股、公司特征和公司绩效之间的以下关系:
Q i t = δ θ m i t + ( δ β 2 + β 3 ) x i t + ( δ γ 2 + γ 3 ) u i + δ v i t + w i t . Q i t = δ θ m i t + δ β 2 + β 3 x i t + δ γ 2 + γ 3 u i + δ v i t + w i t . Q_(it)=delta thetam_(it)+(deltabeta_(2)+beta_(3))x_(it)+(deltagamma_(2)+gamma_(3))u_(i)+deltav_(it)+w_(it).Q_{i t}=\delta \theta m_{i t}+\left(\delta \beta_{2}+\beta_{3}\right) x_{i t}+\left(\delta \gamma_{2}+\gamma_{3}\right) u_{i}+\delta v_{i t}+w_{i t} .
Simplifying the notation reveals the regression specification commonly used in the empirical literature:
简化符号揭示了实证文献中常用的回归规范:
Q i t = a 0 + a 1 m i t + a 2 x i t + ε i t . Q i t = a 0 + a 1 m i t + a 2 x i t + ε i t . Q_(it)=a_(0)+a_(1)m_(it)+a_(2)x_(it)+epsi_(it).Q_{i t}=a_{0}+a_{1} m_{i t}+a_{2} x_{i t}+\varepsilon_{i t} .
In a cross-section of firms, as long as the error term, ε i t = ( δ γ 2 + γ 3 ) u i + ε i t = δ γ 2 + γ 3 u i + epsi_(it)=(deltagamma_(2)+gamma_(3))u_(i)+\varepsilon_{i t}=\left(\delta \gamma_{2}+\gamma_{3}\right) u_{i}+ δ v i t + w i t δ v i t + w i t deltav_(it)+w_(it)\delta v_{i t}+w_{i t} - is uncorrelated with both m i t m i t m_(it)m_{i t} and x i t x i t x_(it)x_{i t}, one can consistently estimate the reduced-form coefficient on managerial ownership in the regression for firm value. However, because the choice of managerial ownership depends on unobserved firm characteristics, m i t m i t m_(it)m_{i t} depends on u i u i u_(i)u_{i}, and is therefore correlated with ε i ε i epsi_(i)\varepsilon_{i}. Specifically,
在一组公司的横截面中,只要误差项 ε i t = ( δ γ 2 + γ 3 ) u i + ε i t = δ γ 2 + γ 3 u i + epsi_(it)=(deltagamma_(2)+gamma_(3))u_(i)+\varepsilon_{i t}=\left(\delta \gamma_{2}+\gamma_{3}\right) u_{i}+ δ v i t + w i t δ v i t + w i t deltav_(it)+w_(it)\delta v_{i t}+w_{i t} m i t m i t m_(it)m_{i t} x i t x i t x_(it)x_{i t} 都不相关,就可以在公司价值的回归中一致地估计管理层持股的简化形式系数。然而,由于管理层持股的选择依赖于未观察到的公司特征, m i t m i t m_(it)m_{i t} 依赖于 u i u i u_(i)u_{i} ,因此与 ε i ε i epsi_(i)\varepsilon_{i} 相关。具体来说,
E ( m i t ε i t ) = E ( ( β 1 x i t + γ 1 u i ) ( δ γ 2 + γ 3 ) u i ) = γ 1 ( δ γ 2 + γ 3 ) σ v 2 E m i t ε i t = E β 1 x i t + γ 1 u i δ γ 2 + γ 3 u i = γ 1 δ γ 2 + γ 3 σ v 2 E(m_(it)epsi_(it))=E((beta_(1)x_(it)+gamma_(1)u_(i))(deltagamma_(2)+gamma_(3))u_(i))=gamma_(1)(deltagamma_(2)+gamma_(3))sigma_(v)^(2)E\left(m_{i t} \varepsilon_{i t}\right)=E\left(\left(\beta_{1} x_{i t}+\gamma_{1} u_{i}\right)\left(\delta \gamma_{2}+\gamma_{3}\right) u_{i}\right)=\gamma_{1}\left(\delta \gamma_{2}+\gamma_{3}\right) \sigma_{v}^{2}
In general, the expectation in Eq. (6) will be zero only in the unlikely event that the optimal contract does not depend on observed firm characteristics ( γ 1 = 0 γ 1 = 0 gamma_(1)=0\gamma_{1}=0 ), or in the event that neither effort nor Q i t Q i t Q_(it)Q_{i t} do ( γ 2 = γ 3 = 0 ) γ 2 = γ 3 = 0 (gamma_(2)=gamma_(3)=0)\left(\gamma_{2}=\gamma_{3}=0\right). Hence one cannot estimate Eq. (5) using ordinary least squares. A natural solution to this problem would be to use instrumental variables for ownership, but this approach is difficult in practice because the natural instruments - the observed firm characteristics x i t x i t x_(it)x_{i t} - are already included on the right-hand side of the equation for firm valuation in Eq. (5). Hence it is difficult to identify instrumental variables that
一般来说,方程(6)中的期望值仅在不太可能的情况下为零,即最优合同不依赖于观察到的公司特征( γ 1 = 0 γ 1 = 0 gamma_(1)=0\gamma_{1}=0 ),或者在努力和 Q i t Q i t Q_(it)Q_{i t} 都不做 ( γ 2 = γ 3 = 0 ) γ 2 = γ 3 = 0 (gamma_(2)=gamma_(3)=0)\left(\gamma_{2}=\gamma_{3}=0\right) 的情况下。因此,无法使用普通最小二乘法估计方程(5)。解决这个问题的一个自然方法是使用所有权的工具变量,但在实践中这种方法是困难的,因为自然工具 - 观察到的公司特征 x i t x i t x_(it)x_{i t} - 已经包含在方程(5)中公司估值的右侧。因此,很难识别出可以用作工具变量的内容。

would permit identification of a 1 a 1 a_(1)a_{1}. With panel data, however, one can use a fixed-effects estimator, assuming that the unobserved heterogeneity is constant over time.
将允许识别 a 1 a 1 a_(1)a_{1} 。然而,使用面板数据时,可以使用固定效应估计量,假设未观察到的异质性在时间上是恒定的。
In contrast to the model for Tobin’s Q Q QQ, the model for the optimal choice of managerial ownership levels in Eq. (1) is more easily identified because it requires only the much weaker assumption that the unobserved firm characteristics are uncorrelated with observed characteristics. Hence the focus of our results in Section 4 is on Eq. (1).
与 Tobin 的 Q Q QQ 模型相比,方程(1)中关于最佳管理所有权水平选择的模型更容易识别,因为它仅需要一个更弱的假设,即未观察到的公司特征与观察到的特征不相关。因此,我们在第 4 节中的结果重点关注方程(1)。
The above discussion suggests four lines of empirical inquiry. First, we explore whether the observed firm characteristics (proxies for the potential for moral hazard and risk) influence managerial ownership in ways that are consistent with theoretical predictions. Second, we investigate the importance of unobserved characteristics as determinants of managerial ownership. Third, we investigate the extent to which the empirical relation between managerial ownership and firm performance (measured by Tobin’s Q Q QQ ) can be explained by the omission of observed and unobserved firm characteristics (i.e., by uncontrol-led-for or unobserved heterogeneity). Fourth, we explore the possibility of using instrumental variables to recover the parameter values in Eq. (5). We describe these results in Sections 4 and 5 after describing our sample and data in Section 3.
上述讨论提出了四条实证研究的思路。首先,我们探讨观察到的公司特征(道德风险和风险潜力的代理变量)是否以与理论预测一致的方式影响管理层持股。其次,我们研究未观察到的特征作为管理层持股决定因素的重要性。第三,我们调查管理层持股与公司绩效(通过托宾的 Q Q QQ 测量)之间的实证关系在多大程度上可以通过遗漏观察到的和未观察到的公司特征(即,未控制或未观察的异质性)来解释。第四,我们探讨使用工具变量来恢复方程 (5) 中参数值的可能性。在描述我们的样本和数据后,我们将在第 3 节中在第 4 节和第 5 节中描述这些结果。

3. The data  3. 数据

Our sample consists of firms from the Compustat universe. We restrict ourselves to firms that have no missing data (on sales, the book value of capital, and the stock price) over the three-year period 1982-1984. (We cannot avoid this conditioning because we cannot use firms with missing data or fewer than three years of data for the variables of interest.) We then select 600 firms by random sampling, and we collect data for all subsequent periods. Our panel is therefore balanced at 600 firms from 1982 through 1984, but the number of firms declines to 551 by 1985 , and falls to a low of 330 by 1992, the last year in the sample. Because of this attrition from Compustat (principally due to mergers and acquisitions), our panel is systematically less random over time. However, we avoid exacerbating the scope for sampling bias by not requiring a balanced panel.
我们的样本由来自 Compustat 宇宙的公司组成。我们限制在 1982 年至 1984 年的三年期间内没有缺失数据(关于销售、资本的账面价值和股票价格)的公司。(我们无法避免这种条件限制,因为我们不能使用缺失数据或少于三年数据的公司进行感兴趣变量的分析。)然后,我们通过随机抽样选择 600 家公司,并收集所有后续时期的数据。因此,我们的面板在 1982 年至 1984 年期间是平衡的 600 家公司,但到 1985 年公司数量减少到 551,到 1992 年样本的最后一年降至 330。由于 Compustat 的这种流失(主要是由于合并和收购),我们的面板随着时间的推移系统性地减少了随机性。然而,我们通过不要求平衡面板来避免加剧抽样偏差的范围。
For this unbalanced panel of firms, we attempt to collect the following additional data for each firm-year observation: the number of top managers and directors (as reported in the proxy statement), the percentage of the firm’s shares owned by those managers and directors, and the date of the proxy statement from which these two numbers are collected. For those observations for which we can locate proxy statements, we collect the managerial ownership variables and merge this information with the Compustat data. Because smaller firms (in
对于这个不平衡的公司面板,我们试图为每个公司-年份观察收集以下额外数据:高管和董事的数量(如代理声明中所报告),这些高管和董事所持有的公司股份的百分比,以及收集这两个数字的代理声明的日期。对于我们能够找到代理声明的观察,我们收集管理层持股变量,并将这些信息与 Compustat 数据合并。因为较小的公司(在
Table 1  表 1
Sample of Compustat firms by year
按年份的 Compustat 公司样本

We start out with 600 firms randomly sampled from the universe of Compustat firms with data available over the period 1982-1984 on sales, book value of capital, and stock price. The number of firms declines after 1984, principally due to mergers and acquisitions. The number of available ownership observations represents firms for which we are able to obtain proxy statements with the number of top managers and directors and their collective percentage share ownership.
我们从随机抽取的 600 家 Compustat 公司开始,这些公司在 1982-1984 年期间有销售额、资本账面价值和股票价格的数据。1984 年后,公司的数量减少,主要是由于合并和收购。可用的所有权观察数量代表了我们能够获得代理声明的公司,这些声明中包含高管和董事的数量及其集体持股比例。
Year  

可用的 Compustat 观察数量
Number of available
Compustat observations
Number of available Compustat observations| Number of available | | :--- | | Compustat observations |

可用所有权观察的数量
Number of available
ownership
observations
Number of available ownership observations| Number of available | | :--- | | ownership | | observations |
1982 600 398
1983 600 425
1984 600 427
1985 549 408
1986 518 385
1987 482 359
1988 442 330
1989 422 329
1990 396 300
1991 382 296
1992 330 293
Year "Number of available Compustat observations" "Number of available ownership observations" 1982 600 398 1983 600 425 1984 600 427 1985 549 408 1986 518 385 1987 482 359 1988 442 330 1989 422 329 1990 396 300 1991 382 296 1992 330 293| Year | Number of available <br> Compustat observations | Number of available <br> ownership <br> observations | | :--- | :--- | :--- | | 1982 | 600 | 398 | | 1983 | 600 | 425 | | 1984 | 600 | 427 | | 1985 | 549 | 408 | | 1986 | 518 | 385 | | 1987 | 482 | 359 | | 1988 | 442 | 330 | | 1989 | 422 | 329 | | 1990 | 396 | 300 | | 1991 | 382 | 296 | | 1992 | 330 | 293 |
terms of the number of shareholders) are not required to file proxies with the Securities and Exchange Commission, we are unable to obtain proxy information for all firms. We end up with managerial ownership information for about 70 % 70 % 70%70 \% of the Compustat firms. Table 1 summarizes the number of firms in our sample as a result of the sample selection process.
(股东人数的条款)不需要向证券交易委员会提交代理文件,因此我们无法获得所有公司的代理信息。我们最终获得了大约 70 % 70 % 70%70 \% 家 Compustat 公司的管理层持股信息。表 1 总结了由于样本选择过程而导致的我们样本中公司的数量。
Despite the problems of attrition and proxy availability (which are not unique to our study), our sample provides several distinct advantages over datasets used in previous studies. First, in contrast to studies that focus on the Fortune 1000, our sample includes a much larger number of small firms and is more representative of the typical firm in Compustat. Second, we have a panel of firms rather than a single cross-section. This allows us to control for firm-level fixed effects. Third, we deliberately construct our panel in such a way that we can control for sample selection bias because of lack of data (for ownership) and attrition. In fact, it is possible to describe the significance of the bias imposed on the level of managerial ownership by a requirement that the panel be balanced; looking over the 1982-1992 period, the average ownership share varies from 16.2 % 16.2 % 16.2%16.2 \% to 19.4 % 19.4 % 19.4%19.4 \%, and for the balanced panel, for the firms removed by the balancing criterion, the ownership share varies between 22.4 % 22.4 % 22.4%22.4 \% and 25.3 % 25.3 % 25.3%25.3 \%. The availability of data on managerial ownership is well predicted by variables such as firm size and fixed
尽管存在流失和代理可用性的问题(这些问题并非我们研究所独有),但我们的样本相较于以往研究中使用的数据集提供了几个明显的优势。首先,与专注于财富 1000 强的研究相比,我们的样本包含了更多的小型企业,更能代表 Compustat 中的典型企业。其次,我们拥有一组企业的面板数据,而不是单一的横截面数据。这使我们能够控制企业层面的固定效应。第三,我们故意构建我们的面板,以便能够控制由于缺乏数据(关于所有权)和流失造成的样本选择偏差。事实上,可以描述由于要求面板平衡而对管理层所有权水平施加的偏差的重要性;在 1982-1992 年期间,平均所有权份额从 16.2 % 16.2 % 16.2%16.2 \% 变化到 19.4 % 19.4 % 19.4%19.4 \% ,而对于平衡面板,被平衡标准移除的企业,所有权份额在 22.4 % 22.4 % 22.4%22.4 \% 25.3 % 25.3 % 25.3%25.3 \% 之间变化。管理层所有权的数据可用性可以通过企业规模和固定等变量很好地预测。
Table 2  表 2
Managerial ownership stakes by firm size, 1982
1982 年按公司规模划分的管理层持股比例

For the 398 Compustat firms for which we have data on sales, book value of capital, stock price, number of top managers and directors, and collective equity ownership of top managers and directors, we report the average number of managers and their average collective ownership stake by firm size.
对于 398 家我们拥有销售额、资本账面价值、股票价格、顶级管理者和董事人数以及顶级管理者和董事的集体股权所有权数据的 Compustat 公司,我们报告了按公司规模划分的管理者平均人数及其平均集体所有权股份。
Firm size class  公司规模类别
  公司数量
Number
of firms
Number of firms| Number | | :--- | | of firms |

每家公司平均经理人数
Average number
of managers
per firm
Average number of managers per firm| Average number | | :--- | | of managers | | per firm |

平均总管理层持股比例
Average total
managerial
ownership
stake
Average total managerial ownership stake| Average total | | :--- | | managerial | | ownership | | stake |
Sales < $ 22 < $ 22 < $22<\$ 22 million  销售 < $ 22 < $ 22 < $22<\$ 22 百万 111 7.2 32.0 % 32.0 % 32.0%32.0 \%
$ 22 $ 22 $22\$ 22 million <=\leqslant Sales $ 188 $ 188 <= $188\leqslant \$ 188 million
$ 22 $ 22 $22\$ 22 百万 <=\leqslant 销售 $ 188 $ 188 <= $188\leqslant \$ 188 百万
147 12.4 25.4 % 25.4 % 25.4%25.4 \%
Sales > $ 188 > $ 188 > $188>\$ 188 million  销售 > $ 188 > $ 188 > $188>\$ 188 百万 140 22.3 13.4 % 13.4 % 13.4%13.4 \%
Firm size class "Number of firms" "Average number of managers per firm" "Average total managerial ownership stake" Sales < $22 million 111 7.2 32.0% $22 million <= Sales <= $188 million 147 12.4 25.4% Sales > $188 million 140 22.3 13.4%| Firm size class | Number <br> of firms | Average number <br> of managers <br> per firm | Average total <br> managerial <br> ownership <br> stake | | :--- | :--- | :--- | :--- | | Sales $<\$ 22$ million | 111 | 7.2 | $32.0 \%$ | | $\$ 22$ million $\leqslant$ Sales $\leqslant \$ 188$ million | 147 | 12.4 | $25.4 \%$ | | Sales $>\$ 188$ million | 140 | 22.3 | $13.4 \%$ |
capital intensity, but as we explain in our discussion of empirical results, controlling for this ‘missing data bias’ does not qualitatively affect our results. While our sample design allows us to estimate and control for the effects of attrition bias, exit from Compustat due to mergers and acquisitions or bankruptcies is, in practice, difficult to predict using observable firm characteristics. A simple probit model for exit reveals that size is the principal explanatory variable; many more firms exit because of mergers than because of failure. When we include the inverse Mills ratio in our Q Q QQ regressions, we find no statistically significant effect of selection bias. We therefore decide not to correct formally for attrition bias.
资本密集度,但正如我们在对实证结果的讨论中所解释的,控制这种“缺失数据偏差”并不会对我们的结果产生质的影响。虽然我们的样本设计使我们能够估计和控制流失偏差的影响,但由于并购或破产而退出 Compustat 在实践中很难通过可观察的公司特征进行预测。一个简单的退出概率模型显示,规模是主要的解释变量;由于并购而退出的公司远多于因失败而退出的公司。当我们在我们的 Q Q QQ 回归中包含逆米尔斯比率时,我们发现选择偏差没有统计学上显著的影响。因此,我们决定不正式纠正流失偏差。
To illustrate differences between small and large firms, Table 2 shows, by size class, the average number of managers per firm and the percentage of shares outstanding owned collectively by those managers in 1982. The frequency distributions of managerial ownership and the number of managers are reported in Fig. 1. Note that the percentage of shares owned by insiders is much higher for small firms, measuring 32 % 32 % 32%32 \% on average for firms in (roughly) the bottom third of the size distribution of firms. By contrast, existing studies typically oversample large firms, and report average ownership shares of approximately 10 % 10 % 10%10 \%. This figure is consistent with the ownership stakes in firms in the top third of our size distribution (for comparison, the sales cutoff for Fortune 1000 firms is approximately $ 1 $ 1 $1\$ 1 billion).
为了说明小型企业和大型企业之间的差异,表 2 按规模类别显示了 1982 年每家企业的平均经理人数和这些经理共同拥有的流通股份百分比。管理层所有权和经理人数的频率分布在图 1 中报告。请注意,小型企业内部人士所拥有的股份百分比要高得多,对于(大致上)规模分布底部三分之一的企业,平均为 32 % 32 % 32%32 \% 。相比之下,现有研究通常对大型企业进行过度抽样,并报告平均所有权股份约为 10 % 10 % 10%10 \% 。这一数字与我们规模分布顶部三分之一企业的所有权份额一致(作为比较,财富 1000 强企业的销售门槛约为 $ 1 $ 1 $1\$ 1 十亿)。

4. Determinants of managerial ownership: empirical evidence
4. 管理层持股的决定因素:实证证据

4.1. Firm characteristics
4.1. 公司特征

The simple model outlined in Section 2 indicates the need to identify observable variables that relate to potential moral hazard and influence optimal
第二节中概述的简单模型表明,需要识别与潜在道德风险相关并影响最优的可观察变量

Fig. 1. Frequency distribution of managerial ownership and number of managers, 1982.
图 1. 管理层持股和经理人数的频率分布,1982 年。

managerial stakes. If the scope for managerial discretion differs across firms according to observable differences in the composition of assets, then a prediction of the theory is that firms with assets that are difficult to monitor will have higher levels of managerial ownership. The specification used by Demsetz and Lehn (1985) to explain ownership concentration includes stock price volatility and industry dummies, but does not include proxies for the scope for managerial discretion (though managerial discretion is one interpretation offered for stock price volatility). We extend their specification by adding a large number of explanatory variables designed to proxy for the scope for managerial discretion, namely, size, capital intensity, cash flow, R&D intensity, advertising intensity, and gross investment rates. As we show below, this expanded variable set dramatically improves the R 2 R 2 R^(2)R^{2} statistic, and the coefficient estimates are all statistically different from zero with the predicted signs.
管理层的利益。如果管理层的自由裁量权在不同公司之间因可观察到的资产组成差异而有所不同,那么理论的一个预测是,拥有难以监控资产的公司将拥有更高水平的管理层持股。Demsetz 和 Lehn(1985)用于解释所有权集中度的规范包括股票价格波动和行业虚拟变量,但不包括管理层自由裁量权的代理变量(尽管管理层自由裁量权是对股票价格波动的一种解释)。我们通过添加大量旨在代理管理层自由裁量权的解释变量来扩展他们的规范,即规模、资本密集度、现金流、研发强度、广告强度和总投资率。正如我们下面所示,这一扩展的变量集显著改善了 R 2 R 2 R^(2)R^{2} 统计量,且系数估计值均在统计上显著不同于零,并具有预测的符号。
Size. Firm size has an ambiguous effect a priori on the scope for moral hazard. On the one hand, monitoring and agency costs can be greater in large firms, increasing desired managerial ownership. In addition, large firms are likely to employ more skilled managers, who are consequently wealthier, suggesting a higher level of managerial ownership. On the other hand, large firms might enjoy economies of scale in monitoring by top management and by rating agencies, leading to a lower optimal level of managerial ownership. We use the log log log\log of firm sales, L N ( S ) L N ( S ) LN(S)L N(S), and its square, ( L N ( S ) ) 2 ( L N ( S ) ) 2 (LN(S))^(2)(L N(S))^{2}, to measure size.
规模。公司规模对道德风险的范围有模糊的影响。一方面,大型公司的监控和代理成本可能更高,从而增加所需的管理层持股比例。此外,大型公司可能会雇佣更多技术熟练的管理者,他们因此更富有,这表明管理层持股比例更高。另一方面,大型公司可能在高层管理和评级机构的监控中享有规模经济,从而导致管理层持股的最佳水平较低。我们使用公司销售的 log log log\log L N ( S ) L N ( S ) LN(S)L N(S) 及其平方 ( L N ( S ) ) 2 ( L N ( S ) ) 2 (LN(S))^(2)(L N(S))^{2} 来衡量规模。

Scope for discretionary spending. To the extent that investments in fixed capital are observable and more easily monitored, firms with a greater concentration of fixed or ‘hard’ capital in their inputs will generally have a lower optimal level of managerial ownership (Gertler and Hubbard, 1988). We use the firm’s capital-to-sales ratio, K / S K / S K//SK / S, and its square, ( K / S ) 2 ( K / S ) 2 (K//S)^(2)(K / S)^{2}, as measures of the relative importance of hard capital in the firm’s technology.
可自由支配支出的范围。在固定资本的投资可观察且更易于监控的情况下,投入中固定或“硬”资本集中度较高的公司通常会有较低的最佳管理层持股比例(Gertler 和 Hubbard,1988)。我们使用公司的资本与销售比率 K / S K / S K//SK / S 及其平方 ( K / S ) 2 ( K / S ) 2 (K//S)^(2)(K / S)^{2} 作为公司技术中硬资本相对重要性的衡量标准。
Beyond hard capital, other firm spending is more discretionary and less easily monitored. The greater the role of these ‘soft capital’ inputs in the firm’s technology, all else being equal, the higher is the desired level of managerial ownership. By including the capital-to-sales ratio, we have controlled (inversely) for soft capital, but some soft capital is ‘softer’ than others and hence more vulnerable to managerial discretion. To refine our proxies for the scope for discretionary spending, we use the ratio of R & D R & D R&D\mathrm{R} \& \mathrm{D} spending to capital, ( R & D ) / K ( R & D ) / K (R&D)//K(R \& D) / K, the ratio of advertising spending to capital, A / K A / K A//KA / K, and dummy variables for whether the firm reports R & D R & D R&D\mathrm{R} \& \mathrm{D} spending ( R D U M R D U M RDUMR D U M ) and advertising spending ( A D U M ) ( A D U M ) (ADUM)(A D U M) in that year. We include dummy variables when R & D R & D R&D\mathrm{R} \& \mathrm{D} and advertising are missing to control for the possibility that nonreporting firms are discretely different from reporting firms. By far the most common reason for not complying with the disclosure requirement is that the level of R & D R & D R&D\mathrm{R} \& \mathrm{D} or advertising expenditure is negligible. Simply eliminating observations with missing values for these variables is undesirable because it significantly reduces the sample size and biases the sample in favor of R&D-intensive and advertising firms.
除了硬资本,其他公司的支出更具选择性且不易监控。这些“软资本”投入在公司技术中所占的比重越大,其他条件相同,所需的管理层持股水平就越高。通过包括资本与销售比率,我们已经控制了(反向)软资本,但某些软资本比其他软资本“更软”,因此更容易受到管理层的自由裁量影响。为了细化我们对选择性支出范围的代理,我们使用了 R & D R & D R&D\mathrm{R} \& \mathrm{D} 支出与资本的比率, ( R & D ) / K ( R & D ) / K (R&D)//K(R \& D) / K ,广告支出与资本的比率, A / K A / K A//KA / K ,以及虚拟变量来表示公司是否在该年报告了 R & D R & D R&D\mathrm{R} \& \mathrm{D} 支出( R D U M R D U M RDUMR D U M )和广告支出 ( A D U M ) ( A D U M ) (ADUM)(A D U M) 。当 R & D R & D R&D\mathrm{R} \& \mathrm{D} 和广告缺失时,我们包括虚拟变量,以控制未报告公司与报告公司可能存在的离散差异。未遵守披露要求的最常见原因是 R & D R & D R&D\mathrm{R} \& \mathrm{D} 或广告支出的水平微不足道。 简单地消除这些变量缺失值的观察是不理想的,因为这会显著减少样本大小,并使样本偏向于研发密集型和广告公司。
As a proxy for the link between high growth and opportunities for discretionary projects, we use the firm’s investment rate measured by the ratio of capital expenditures to the capital stock, I / K I / K I//KI / K. Finally, we use the ratio of operating income to sales Y / S Y / S Y//SY / S to measure market power or a firm’s ‘free cash flow’ (the difference between cash flow and spending on value-enhancing investment projects). As suggested by Jensen (1986), the higher is a firm’s free cash flow, all else being equal, the higher is the desired level of managerial ownership. While free cash flow is itself unobservable, it is presumably correlated with operating income.
作为高增长与自由支出项目机会之间联系的代理,我们使用公司的投资率,该投资率通过资本支出与资本存量的比率来衡量, I / K I / K I//KI / K 。最后,我们使用营业收入与销售额的比率 Y / S Y / S Y//SY / S 来衡量市场力量或公司的“自由现金流”(现金流与价值提升投资项目支出之间的差额)。正如詹森(1986)所建议的,其他条件不变的情况下,公司的自由现金流越高,所期望的管理层持股水平就越高。虽然自由现金流本身是不可观察的,但它与营业收入可能是相关的。

Managerial risk aversion. Because higher managerial ownership levels, all else being equal, imply less portfolio diversification for managers, the optimal contract involves a tradeoff between diversification and incentives for performance. The higher is the firm’s idiosyncratic risk, the lower is optimal managerial ownership. Demsetz and Lehn (1985) offer a second interpretation of this relation, suggesting that higher volatility indicates more scope for managerial discretion and thereby increases equilibrium managerial ownership levels. Unlike their specification, ours includes measures of intangible capital to control for managerial discretion. We therefore focus on the first interpretation of risk. As an empirical proxy for volatility, we use the standard deviation of the idiosyncratic component of daily stock prices (constructed from residuals from a standard CAPM regression), denoted by SIGMA, although our results are not qualitatively changed by the substitution of total stock return variance for our definition of SIGMA. Analogous to our treatment of missing values of ( R & D ) / K ( R & D ) / K (R&D)//K(R \& D) / K and A / K A / K A//KA / K, we set missing values of SIGMA equal to zero, and then also include in the regression a dummy variable SIGDUM equal to unity when SIGMA is not missing, and zero otherwise.
管理层风险厌恶。因为在其他条件相同的情况下,较高的管理层持股比例意味着管理者的投资组合多样化程度较低,最佳合同涉及多样化与绩效激励之间的权衡。公司的特有风险越高,最佳的管理层持股比例就越低。Demsetz 和 Lehn (1985) 提出了这种关系的第二种解释,认为更高的波动性表明管理者的自由裁量权更大,从而增加了均衡的管理层持股水平。与他们的模型不同,我们的模型包括无形资本的衡量,以控制管理者的自由裁量权。因此,我们专注于风险的第一种解释。作为波动性的实证代理,我们使用每日股票价格特有成分的标准差(由标准资本资产定价模型回归的残差构建),记作 SIGMA,尽管我们的结果在用总股票收益方差替代我们对 SIGMA 的定义时并没有发生质的变化。 类似于我们对 ( R & D ) / K ( R & D ) / K (R&D)//K(R \& D) / K A / K A / K A//KA / K 缺失值的处理,我们将 SIGMA 的缺失值设为零,然后在回归中还包括一个虚拟变量 SIGDUM,当 SIGMA 不缺失时等于 1,否则等于 0。
To deal with zero-volume trading days in the daily data, we construct n n nn-day returns by summing the Center for Research in Security Prices (CRSP) daily returns over the days in the period to create an approximate n n nn-day return. We then divide this return (as well as n n nn-day returns created by weekends and holidays) by the number of days in the period to obtain an average daily return. The variance of this average return will equal the variance of the daily return, and the same will be true for the idiosyncratic variance in a CAPM regression. Converting n n nn-day returns to average daily returns thus removes the heteroskedasticity introduced by combining n n nn-day-return observations with daily returns. Out of our initial universe of 600 firms for the 1984 period, the CRSP NYSE/AMEX and NASDAQ daily files contains 525 firms reporting returns in 1984. Of these, there are 502 firms with enough data to construct at least 20 observations on daily returns using only days for which trading volume is positive.
为了处理日数据中的零交易量交易日,我们通过对该期间内的 Center for Research in Security Prices (CRSP)日收益进行求和,构建 n n nn 天收益,以创建一个近似的 n n nn 天收益。然后,我们将该收益(以及由周末和假期产生的 n n nn 天收益)除以该期间的天数,以获得平均日收益。该平均收益的方差将等于日收益的方差,在 CAPM 回归中,特有方差也是如此。因此,将 n n nn 天收益转换为平均日收益可以消除通过将 n n nn 天收益观测值与日收益结合而引入的异方差性。在我们 1984 年初始的 600 家公司中,CRSP NYSE/AMEX 和 NASDAQ 日文件包含 525 家在 1984 年报告收益的公司。其中,有 502 家公司有足够的数据,仅使用交易量为正的日子构建至少 20 个日收益观测值。

In addition to the problem of days with zero trading volume, there are days for which closing prices are not available, or days for which CRSP uses the average of the closing bid-ask spread instead of the closing price. This introduces
除了零交易量的天数问题,还有一些天的收盘价不可用,或者 CRSP 使用收盘买卖差价的平均值而不是收盘价。这引入了

nonclassical measurement error into the return calculation, which could bias our ordinary least squares (OLS) estimates of beta, return variance, and residual variance. To check the robustness of our results, we experiment with smaller samples that include only firms for which we can construct at least 20 observations based on (i) positive trading volume and (ii) transactions prices rather than the average of bid-ask prices. This substantially reduces the sample in 1984 to 328 firms (from 502). In practice, however, the results do not differ qualitatively from the larger sample. Hence, we report results in the paper using the larger sample, which, for many firms, relies on the closing average of the bid-ask spread rather than an actual closing price.
将非经典测量误差纳入收益计算,这可能会偏倚我们对 beta、收益方差和残差方差的普通最小二乘法 (OLS) 估计。为了检查我们结果的稳健性,我们尝试使用较小的样本,仅包括我们可以基于 (i) 正的交易量和 (ii) 交易价格而不是买卖价差的平均值构建至少 20 个观察值的公司。这大大减少了 1984 年的样本,从 502 家减少到 328 家。然而,在实践中,结果与较大样本的定性并没有差异。因此,我们在论文中报告使用较大样本的结果,对于许多公司,这依赖于买卖价差的收盘平均值,而不是实际的收盘价。
While we have addressed the most obvious examples of nonsynchronous trading (namely, days on which no trading occurs or on which CRSP cannot obtain a valid transaction price), there remain days on which CRSP calculates returns using the last price transacted rather than the closing price. As Scholes and Williams (1976), among others, point out, the inclusion of nonsynchronous trading days produces biased OLS estimates of beta. To check the robustness of our results against the possibility of biased beta estimates due to nonsynchronous trading days, we follow the approach recommended by Dimson (1979) by including leads and lags of the market return in the beta regression. These additional regressors are occasionally significant for some firms in some years, but using the alternative estimates of the idiosyncratic variance does not materially affect our results.
虽然我们已经解决了最明显的非同步交易示例(即没有交易发生的日子或 CRSP 无法获得有效交易价格的日子),但仍然存在一些日子,CRSP 使用最后成交价格而不是收盘价来计算收益。正如 Scholes 和 Williams(1976)等人指出的,包含非同步交易日会导致 OLS 的 beta 估计偏差。为了检验我们的结果在非同步交易日导致的 beta 估计偏差下的稳健性,我们遵循 Dimson(1979)推荐的方法,在 beta 回归中包含市场收益的滞后和领先项。这些额外的回归变量在某些年份对某些公司偶尔是显著的,但使用替代的特异方差估计并不会实质性影响我们的结果。

Summary. Combining these observable variables associated with moral hazard yields the following reduced-form expression for managerial ownership:
摘要。将与道德风险相关的这些可观察变量结合起来,得出以下关于管理层持股的简化形式表达:
m i t = f ( L N ( S ) i t , ( K / S ) i t , ( R & D / K ) i t , R D U M i t , ( A / K ) i t , A D U M i t , ( I / K ) i t , ( Y / S ) i t , S I G M A i t , S I G D U M i t ) + u i + η i t , m i t = f L N ( S ) i t , ( K / S ) i t , ( R & D / K ) i t , R D U M i t , ( A / K ) i t , A D U M i t , ( I / K ) i t , ( Y / S ) i t , S I G M A i t , S I G D U M i t + u i + η i t , {:[m_(it)=f(LN(S)_(it),(K//S)_(it),(R&D//K)_(it),RDUM_(it),(A//K)_(it),ADUM_(it),(I//K)_(it),:}],[{:(Y//S)_(it),SIGMA_(it),SIGDUM_(it))+u_(i)+eta_(it)","]:}\begin{aligned} m_{i t}= & f\left(L N(S)_{i t},(K / S)_{i t},(R \& D / K)_{i t}, R D U M_{i t},(A / K)_{i t}, A D U M_{i t},(I / K)_{i t},\right. \\ & \left.(Y / S)_{i t}, S I G M A_{i t}, S I G D U M_{i t}\right)+u_{i}+\eta_{i t}, \end{aligned}
where i i ii and t t tt represent the firm and time, respectively, u i u i u_(i)u_{i} is a firm-specific effect, and η i t η i t eta_(it)\eta_{i t} is a white-noise error term. Our list of variables is summarized in Table 3.
其中 i i ii t t tt 分别代表公司和时间, u i u i u_(i)u_{i} 是公司特定效应, η i t η i t eta_(it)\eta_{i t} 是白噪声误差项。我们的变量列表在表 3 中总结。

4.2. Evidence  4.2. 证据

Table 4A reports our estimates of the determinants of managerial stakes. The dependent variable in each case is L N ( m / ( 1 m ) ) L N ( m / ( 1 m ) ) LN(m//(1-m))L N(m /(1-m)). Each of the specifications includes year dummies (not reported). In specifications including fixed firm effects, we control for the unobserved firm heterogeneity represented by u i u i u_(i)u_{i} in Eq. (6).
表 4A 报告了我们对管理层股份决定因素的估计。在每种情况下,因变量为 L N ( m / ( 1 m ) ) L N ( m / ( 1 m ) ) LN(m//(1-m))L N(m /(1-m)) 。每个规格都包括年份虚拟变量(未报告)。在包括固定公司效应的规格中,我们控制了在公式(6)中由 u i u i u_(i)u_{i} 表示的未观察到的公司异质性。
The first column reports results from a baseline specification using pooled data for all firm-years. Increases in firm size, all else being equal, are associated with a reduction in managerial stakes. Increases in fixed capital intensity (which
第一列报告了使用所有公司年份的合并数据的基线规范结果。在其他条件相同的情况下,企业规模的增加与管理层持股的减少相关联。固定资本密集度的增加(这
Table 3  表 3
Variable descriptions  变量描述
Q Q QQ

托宾的 Q Q QQ ,即公司的价值与资产的替代价值之比。对于公司价值,我们使用普通股的市场价值加上优先股的估计市场价值(大致估计为优先股股息的十倍)再加上总负债的账面价值;对于资产的替代价值,我们使用总资产的账面价值。这个定义与市值与账面价值比率密切相关,这可以通过从分子和分母中减去总负债来轻易看出。
Tobin's Q Q QQ, that is, the ratio of the value of the firm divided by the replacement value of
assets. For firm value, we use the market value of common equity plus the estimated
market value of preferred stock (roughly estimated as ten times the preferred dividend)
plus the book value of total liabilities, and for replacement value of assets we use the
book value of total assets. This definition is closely related to the market-to-book ratio,
which is easily seen by subtracting total liabilities from both the numerator and
denominator
Tobin's Q, that is, the ratio of the value of the firm divided by the replacement value of assets. For firm value, we use the market value of common equity plus the estimated market value of preferred stock (roughly estimated as ten times the preferred dividend) plus the book value of total liabilities, and for replacement value of assets we use the book value of total assets. This definition is closely related to the market-to-book ratio, which is easily seen by subtracting total liabilities from both the numerator and denominator| Tobin's $Q$, that is, the ratio of the value of the firm divided by the replacement value of | | :--- | | assets. For firm value, we use the market value of common equity plus the estimated | | market value of preferred stock (roughly estimated as ten times the preferred dividend) | | plus the book value of total liabilities, and for replacement value of assets we use the | | book value of total assets. This definition is closely related to the market-to-book ratio, | | which is easily seen by subtracting total liabilities from both the numerator and | | denominator |
The total common equity holdings of top-level managers as a fraction of common
顶级管理者的普通股权益总持有量占普通股的比例
equity outstanding  流通股本
Q "Tobin's Q, that is, the ratio of the value of the firm divided by the replacement value of assets. For firm value, we use the market value of common equity plus the estimated market value of preferred stock (roughly estimated as ten times the preferred dividend) plus the book value of total liabilities, and for replacement value of assets we use the book value of total assets. This definition is closely related to the market-to-book ratio, which is easily seen by subtracting total liabilities from both the numerator and denominator" The total common equity holdings of top-level managers as a fraction of common equity outstanding | $Q$ | Tobin's $Q$, that is, the ratio of the value of the firm divided by the replacement value of <br> assets. For firm value, we use the market value of common equity plus the estimated <br> market value of preferred stock (roughly estimated as ten times the preferred dividend) <br> plus the book value of total liabilities, and for replacement value of assets we use the <br> book value of total assets. This definition is closely related to the market-to-book ratio, <br> which is easily seen by subtracting total liabilities from both the numerator and <br> denominator | | :--- | :--- | | The total common equity holdings of top-level managers as a fraction of common | | | equity outstanding | |
we associate with lower monitoring costs) also lead to a decline in managerial stakes. Among our proxies for discretionary spending (R&D, advertising, investment rates, and operating income relative to capital), R & D R & D R&D\mathrm{R} \& \mathrm{D} intensity appears to have a negative effect on ownership stakes, while advertising intensity, operating
我们与较低的监控成本相关联)也导致管理层持股的下降。在我们对自由支出的代理指标(研发、广告、投资率以及相对于资本的营业收入)中, R & D R & D R&D\mathrm{R} \& \mathrm{D} 强度似乎对所有权股份产生负面影响,而广告强度、营业

income, and the investment rate appear to have positive effects on ownership stakes. Increases in idiosyncratic risk, as measured by SIGMA, raise the cost of managerial ownership in terms of reduced portfolio diversification and also reduce managerial ownership.
收入和投资率似乎对所有权份额有积极影响。特有风险的增加(以 SIGMA 衡量)提高了管理层所有权的成本,表现为投资组合多样化的减少,同时也减少了管理层所有权。
Table 4  表 4
(A) Determinants of total equity ownership by top managers
(A) 高管总股权拥有的决定因素
The specifications reported in this table all model the fraction of common equity held by top managers, m m mm, by regressing the transformed dependent variable L N ( m / ( 1 m ) ) L N ( m / ( 1 m ) ) LN(m//(1-m))L N(m /(1-m)) on the explanatory variables indicated below. Intercept terms and year dummies are included for all regressions, but not reported. Fixed effects at the industry or firm level are included where indicated, but not reported. Variable definitions for the acronyms are given in Table 3.
本表中报告的规格均建模了高管持有的普通股比例 m m mm ,通过对转化后的因变量 L N ( m / ( 1 m ) ) L N ( m / ( 1 m ) ) LN(m//(1-m))L N(m /(1-m)) 进行回归,使用下面所示的解释变量。所有回归均包含截距项和年份虚拟变量,但未报告。行业或公司层面的固定效应在指示的地方包含,但未报告。缩略语的变量定义见表 3。
Variable  变量 All firms (Pooled)  所有公司(合并) All firms (SIC3 effects)
所有公司(SIC3 效应)
All firms (Firm effects)
所有公司(公司效应)

财富 500 强(公司效应)
Fortune 500
(Firm effects)
Fortune 500 (Firm effects)| Fortune 500 | | :--- | | (Firm effects) |
  非 500(公司效应)
Non-500
(Firm effects)
Non-500 (Firm effects)| Non-500 | | :--- | | (Firm effects) |
L N ( S ) L N ( S ) LN(S)L N(S) 0.195 ( 0.050 ) 0.195 ( 0.050 ) {:[-0.195],[(0.050)]:}\begin{gathered} -0.195 \\ (0.050) \end{gathered} 0.182 ( 0.053 ) 0.182 ( 0.053 ) {:[-0.182],[(0.053)]:}\begin{gathered} -0.182 \\ (0.053) \end{gathered} 0.058 ( 0.095 ) 0.058 ( 0.095 ) {:[0.058],[(0.095)]:}\begin{gathered} 0.058 \\ (0.095) \end{gathered} 1.288 ( 0.697 ) 1.288 ( 0.697 ) {:[-1.288],[(0.697)]:}\begin{gathered} -1.288 \\ (0.697) \end{gathered} 0.252 ( 0.121 ) 0.252 ( 0.121 ) {:[0.252],[(0.121)]:}\begin{gathered} 0.252 \\ (0.121) \end{gathered}
( L N ( S ) ) 2 ( L N ( S ) ) 2 (LN(S))^(2)(L N(S))^{2} 0.027 ( 0.005 ) 0.027 ( 0.005 ) {:[-0.027],[(0.005)]:}\begin{gathered} -0.027 \\ (0.005) \end{gathered} 0.027 ( 0.005 ) 0.027 ( 0.005 ) {:[-0.027],[(0.005)]:}\begin{gathered} -0.027 \\ (0.005) \end{gathered} 0.038 ( 0.010 ) 0.038 ( 0.010 ) {:[-0.038],[(0.010)]:}\begin{gathered} -0.038 \\ (0.010) \end{gathered} 0.040 ( 0.045 ) 0.040 ( 0.045 ) {:[0.040],[(0.045)]:}\begin{gathered} 0.040 \\ (0.045) \end{gathered} 0.067 ( 0.016 ) 0.067 ( 0.016 ) {:[-0.067],[(0.016)]:}\begin{array}{r} -0.067 \\ (0.016) \end{array}
K/S 1.131 ( 0.250 ) 1.131 ( 0.250 ) {:[-1.131],[(0.250)]:}\begin{array}{r} -1.131 \\ (0.250) \end{array} 0.826 ( 0.274 ) 0.826 ( 0.274 ) {:[-0.826],[(0.274)]:}\begin{gathered} -0.826 \\ (0.274) \end{gathered} 0.826 ( 0.259 ) 0.826 ( 0.259 ) {:[-0.826],[(0.259)]:}\begin{gathered} -0.826 \\ (0.259) \end{gathered} 1.05 ( 0.543 ) 1.05 ( 0.543 ) {:[-1.05],[(0.543)]:}\begin{gathered} -1.05 \\ (0.543) \end{gathered} 0.448 ( 0.296 ) 0.448 ( 0.296 ) {:[-0.448],[(0.296)]:}\begin{array}{r} -0.448 \\ (0.296) \end{array}
( K / S ) 2 ( K / S ) 2 (K//S)^(2)(K / S)^{2} 0.023 ( 0.157 ) 0.023 ( 0.157 ) {:[-0.023],[(0.157)]:}\begin{gathered} -0.023 \\ (0.157) \end{gathered} 0.011 ( 0.145 ) 0.011 ( 0.145 ) {:[-0.011],[(0.145)]:}\begin{gathered} -0.011 \\ (0.145) \end{gathered} 0.301 ( 0.122 ) 0.301 ( 0.122 ) {:[0.301],[(0.122)]:}\begin{gathered} 0.301 \\ (0.122) \end{gathered} 0.440 ( 0.228 ) 0.440 ( 0.228 ) {:[0.440],[(0.228)]:}\begin{gathered} 0.440 \\ (0.228) \end{gathered} 0.143 ( 0.141 ) 0.143 ( 0.141 ) {:[0.143],[(0.141)]:}\begin{gathered} 0.143 \\ (0.141) \end{gathered}
SIGMA 5.20 ( 1.96 ) 5.20 ( 1.96 ) {:[-5.20],[(1.96)]:}\begin{array}{r} -5.20 \\ (1.96) \end{array} 3.84 ( 1.86 ) 3.84 ( 1.86 ) {:[-3.84],[(1.86)]:}\begin{array}{r} -3.84 \\ (1.86) \end{array} 5.13 ( 1.43 ) 5.13 ( 1.43 ) {:[-5.13],[(1.43)]:}\begin{array}{r} -5.13 \\ (1.43) \end{array} 0.707 ( 13.3 ) 0.707 ( 13.3 ) {:[-0.707],[(13.3)]:}\begin{gathered} -0.707 \\ (13.3) \end{gathered} 4.84 ( 1.38 ) 4.84 ( 1.38 ) {:[-4.84],[(1.38)]:}\begin{array}{r} -4.84 \\ (1.38) \end{array}
SIGDUM 0.098 ( 0.098 ) 0.098 ( 0.098 ) {:[0.098],[(0.098)]:}\begin{gathered} 0.098 \\ (0.098) \end{gathered} 0.142 ( 0.092 ) 0.142 ( 0.092 ) {:[0.142],[(0.092)]:}\begin{gathered} 0.142 \\ (0.092) \end{gathered} 0.083 ( 0.111 ) 0.083 ( 0.111 ) {:[0.083],[(0.111)]:}\begin{gathered} 0.083 \\ (0.111) \end{gathered} 1.49 ( 0.568 ) 1.49 ( 0.568 ) {:[1.49],[(0.568)]:}\begin{aligned} & 1.49 \\ & (0.568) \end{aligned} 0.092 ( 0.090 ) 0.092 ( 0.090 ) {:[-0.092],[(0.090)]:}\begin{gathered} -0.092 \\ (0.090) \end{gathered}
Y / S Y / S Y//SY / S 0.143 ( 0.240 ) 0.143 ( 0.240 ) {:[0.143],[(0.240)]:}\begin{gathered} 0.143 \\ (0.240) \end{gathered} 0.020 ( 0.232 ) 0.020 ( 0.232 ) {:[-0.020],[(0.232)]:}\begin{array}{r} -0.020 \\ (0.232) \end{array} 0.219 ( 0.178 ) 0.219 ( 0.178 ) {:[0.219],[(0.178)]:}\begin{gathered} 0.219 \\ (0.178) \end{gathered} 0.683 ( 0.678 ) 0.683 ( 0.678 ) {:[0.683],[(0.678)]:}\begin{gathered} 0.683 \\ (0.678) \end{gathered} 0.191 ( 0.175 ) 0.191 ( 0.175 ) {:[0.191],[(0.175)]:}\begin{gathered} 0.191 \\ (0.175) \end{gathered}
( R & D ) / K ( R & D ) / K (R&D)//K(R \& D) / K 1.084 ( 0.197 ) 1.084 ( 0.197 ) {:[-1.084],[(0.197)]:}\begin{array}{r} -1.084 \\ (0.197) \end{array} 0.239 ( 0.206 ) 0.239 ( 0.206 ) {:[-0.239],[(0.206)]:}\begin{gathered} -0.239 \\ (0.206) \end{gathered} 0.502 ( 0.284 ) 0.502 ( 0.284 ) {:[0.502],[(0.284)]:}\begin{gathered} 0.502 \\ (0.284) \end{gathered} 3.08 ( 1.21 ) 3.08 ( 1.21 ) {:[3.08],[(1.21)]:}\begin{gathered} 3.08 \\ (1.21) \end{gathered} 0.546 ( 0.289 ) 0.546 ( 0.289 ) {:[0.546],[(0.289)]:}\begin{gathered} 0.546 \\ (0.289) \end{gathered}
RDUM 0.191 ( 0.061 ) 0.191 ( 0.061 ) {:[-0.191],[(0.061)]:}\begin{array}{r} -0.191 \\ (0.061) \end{array} 0.056 ( 0.090 ) 0.056 ( 0.090 ) {:[-0.056],[(0.090)]:}\begin{array}{r} -0.056 \\ (0.090) \end{array} 0.332 ( 0.105 ) 0.332 ( 0.105 ) {:[0.332],[(0.105)]:}\begin{gathered} 0.332 \\ (0.105) \end{gathered} 0.665 ( 0.322 ) 0.665 ( 0.322 ) {:[0.665],[(0.322)]:}\begin{gathered} 0.665 \\ (0.322) \end{gathered} 0.242 ( 0.105 ) 0.242 ( 0.105 ) {:[0.242],[(0.105)]:}\begin{gathered} 0.242 \\ (0.105) \end{gathered}
A/K 0.227 ( 0.217 ) 0.227 ( 0.217 ) {:[0.227],[(0.217)]:}\begin{gathered} 0.227 \\ (0.217) \end{gathered} 0.953 ( 0.332 ) 0.953 ( 0.332 ) {:[0.953],[(0.332)]:}\begin{gathered} 0.953 \\ (0.332) \end{gathered} 0.184 ( 0.438 ) 0.184 ( 0.438 ) {:[0.184],[(0.438)]:}\begin{gathered} 0.184 \\ (0.438) \end{gathered} 3.60 ( 1.19 ) 3.60 ( 1.19 ) {:[3.60],[(1.19)]:}\begin{gathered} 3.60 \\ (1.19) \end{gathered} 0.067 ( 0.413 ) 0.067 ( 0.413 ) {:[-0.067],[(0.413)]:}\begin{gathered} -0.067 \\ (0.413) \end{gathered}
A D U M A D U M ADUMA D U M 0.143 ( 0.061 ) 0.143 ( 0.061 ) {:[0.143],[(0.061)]:}\begin{gathered} 0.143 \\ (0.061) \end{gathered} 0.082 ( 0.067 ) 0.082 ( 0.067 ) {:[-0.082],[(0.067)]:}\begin{gathered} -0.082 \\ (0.067) \end{gathered} 0.042 ( 0.072 ) 0.042 ( 0.072 ) {:[0.042],[(0.072)]:}\begin{gathered} 0.042 \\ (0.072) \end{gathered} 0.033 ( 0.215 ) 0.033 ( 0.215 ) {:[0.033],[(0.215)]:}\begin{gathered} 0.033 \\ (0.215) \end{gathered} 0.037 ( 0.077 ) 0.037 ( 0.077 ) {:[-0.037],[(0.077)]:}\begin{gathered} -0.037 \\ (0.077) \end{gathered}
I/K 0.440 ( 0.156 ) 0.440 ( 0.156 ) {:[0.440],[(0.156)]:}\begin{gathered} 0.440 \\ (0.156) \end{gathered} 0.114 ( 0.152 ) 0.114 ( 0.152 ) {:[0.114],[(0.152)]:}\begin{gathered} 0.114 \\ (0.152) \end{gathered} 0.157 ( 0.099 ) 0.157 ( 0.099 ) {:[0.157],[(0.099)]:}\begin{gathered} 0.157 \\ (0.099) \end{gathered} 0.280 ( 0.191 ) 0.280 ( 0.191 ) {:[0.280],[(0.191)]:}\begin{gathered} 0.280 \\ (0.191) \end{gathered} 0.144 ( 0.106 ) 0.144 ( 0.106 ) {:[0.144],[(0.106)]:}\begin{gathered} 0.144 \\ (0.106) \end{gathered}
# Obs.
Adj. R 2 R 2 R^(2)R^{2}
# Obs. Adj. R^(2)| # Obs. | | :--- | | Adj. $R^{2}$ |
2630 0.407 2630 0.407 {:[2630],[0.407]:}\begin{aligned} & 2630 \\ & 0.407 \end{aligned} 2630 0.584 2630 0.584 {:[2630],[0.584]:}\begin{aligned} & 2630 \\ & 0.584 \end{aligned} 2630 0.884 2630 0.884 {:[2630],[0.884]:}\begin{aligned} & 2630 \\ & 0.884 \end{aligned} 764 0.884 764 0.884 {:[764],[0.884]:}\begin{aligned} & 764 \\ & 0.884 \end{aligned} 1866 0.831 1866 0.831 {:[1866],[0.831]:}\begin{aligned} & 1866 \\ & 0.831 \end{aligned}
Variable All firms (Pooled) All firms (SIC3 effects) All firms (Firm effects) "Fortune 500 (Firm effects)" "Non-500 (Firm effects)" LN(S) "-0.195 (0.050)" "-0.182 (0.053)" "0.058 (0.095)" "-1.288 (0.697)" "0.252 (0.121)" (LN(S))^(2) "-0.027 (0.005)" "-0.027 (0.005)" "-0.038 (0.010)" "0.040 (0.045)" "-0.067 (0.016)" K/S "-1.131 (0.250)" "-0.826 (0.274)" "-0.826 (0.259)" "-1.05 (0.543)" "-0.448 (0.296)" (K//S)^(2) "-0.023 (0.157)" "-0.011 (0.145)" "0.301 (0.122)" "0.440 (0.228)" "0.143 (0.141)" SIGMA "-5.20 (1.96)" "-3.84 (1.86)" "-5.13 (1.43)" "-0.707 (13.3)" "-4.84 (1.38)" SIGDUM "0.098 (0.098)" "0.142 (0.092)" "0.083 (0.111)" "1.49 (0.568)" "-0.092 (0.090)" Y//S "0.143 (0.240)" "-0.020 (0.232)" "0.219 (0.178)" "0.683 (0.678)" "0.191 (0.175)" (R&D)//K "-1.084 (0.197)" "-0.239 (0.206)" "0.502 (0.284)" "3.08 (1.21)" "0.546 (0.289)" RDUM "-0.191 (0.061)" "-0.056 (0.090)" "0.332 (0.105)" "0.665 (0.322)" "0.242 (0.105)" A/K "0.227 (0.217)" "0.953 (0.332)" "0.184 (0.438)" "3.60 (1.19)" "-0.067 (0.413)" ADUM "0.143 (0.061)" "-0.082 (0.067)" "0.042 (0.072)" "0.033 (0.215)" "-0.037 (0.077)" I/K "0.440 (0.156)" "0.114 (0.152)" "0.157 (0.099)" "0.280 (0.191)" "0.144 (0.106)" "# Obs. Adj. R^(2)" "2630 0.407" "2630 0.584" "2630 0.884" "764 0.884" "1866 0.831"| Variable | All firms (Pooled) | All firms (SIC3 effects) | All firms (Firm effects) | Fortune 500 <br> (Firm effects) | Non-500 <br> (Firm effects) | | :---: | :---: | :---: | :---: | :---: | :---: | | $L N(S)$ | $\begin{gathered} -0.195 \\ (0.050) \end{gathered}$ | $\begin{gathered} -0.182 \\ (0.053) \end{gathered}$ | $\begin{gathered} 0.058 \\ (0.095) \end{gathered}$ | $\begin{gathered} -1.288 \\ (0.697) \end{gathered}$ | $\begin{gathered} 0.252 \\ (0.121) \end{gathered}$ | | $(L N(S))^{2}$ | $\begin{gathered} -0.027 \\ (0.005) \end{gathered}$ | $\begin{gathered} -0.027 \\ (0.005) \end{gathered}$ | $\begin{gathered} -0.038 \\ (0.010) \end{gathered}$ | $\begin{gathered} 0.040 \\ (0.045) \end{gathered}$ | $\begin{array}{r} -0.067 \\ (0.016) \end{array}$ | | K/S | $\begin{array}{r} -1.131 \\ (0.250) \end{array}$ | $\begin{gathered} -0.826 \\ (0.274) \end{gathered}$ | $\begin{gathered} -0.826 \\ (0.259) \end{gathered}$ | $\begin{gathered} -1.05 \\ (0.543) \end{gathered}$ | $\begin{array}{r} -0.448 \\ (0.296) \end{array}$ | | $(K / S)^{2}$ | $\begin{gathered} -0.023 \\ (0.157) \end{gathered}$ | $\begin{gathered} -0.011 \\ (0.145) \end{gathered}$ | $\begin{gathered} 0.301 \\ (0.122) \end{gathered}$ | $\begin{gathered} 0.440 \\ (0.228) \end{gathered}$ | $\begin{gathered} 0.143 \\ (0.141) \end{gathered}$ | | SIGMA | $\begin{array}{r} -5.20 \\ (1.96) \end{array}$ | $\begin{array}{r} -3.84 \\ (1.86) \end{array}$ | $\begin{array}{r} -5.13 \\ (1.43) \end{array}$ | $\begin{gathered} -0.707 \\ (13.3) \end{gathered}$ | $\begin{array}{r} -4.84 \\ (1.38) \end{array}$ | | SIGDUM | $\begin{gathered} 0.098 \\ (0.098) \end{gathered}$ | $\begin{gathered} 0.142 \\ (0.092) \end{gathered}$ | $\begin{gathered} 0.083 \\ (0.111) \end{gathered}$ | $\begin{aligned} & 1.49 \\ & (0.568) \end{aligned}$ | $\begin{gathered} -0.092 \\ (0.090) \end{gathered}$ | | $Y / S$ | $\begin{gathered} 0.143 \\ (0.240) \end{gathered}$ | $\begin{array}{r} -0.020 \\ (0.232) \end{array}$ | $\begin{gathered} 0.219 \\ (0.178) \end{gathered}$ | $\begin{gathered} 0.683 \\ (0.678) \end{gathered}$ | $\begin{gathered} 0.191 \\ (0.175) \end{gathered}$ | | $(R \& D) / K$ | $\begin{array}{r} -1.084 \\ (0.197) \end{array}$ | $\begin{gathered} -0.239 \\ (0.206) \end{gathered}$ | $\begin{gathered} 0.502 \\ (0.284) \end{gathered}$ | $\begin{gathered} 3.08 \\ (1.21) \end{gathered}$ | $\begin{gathered} 0.546 \\ (0.289) \end{gathered}$ | | RDUM | $\begin{array}{r} -0.191 \\ (0.061) \end{array}$ | $\begin{array}{r} -0.056 \\ (0.090) \end{array}$ | $\begin{gathered} 0.332 \\ (0.105) \end{gathered}$ | $\begin{gathered} 0.665 \\ (0.322) \end{gathered}$ | $\begin{gathered} 0.242 \\ (0.105) \end{gathered}$ | | A/K | $\begin{gathered} 0.227 \\ (0.217) \end{gathered}$ | $\begin{gathered} 0.953 \\ (0.332) \end{gathered}$ | $\begin{gathered} 0.184 \\ (0.438) \end{gathered}$ | $\begin{gathered} 3.60 \\ (1.19) \end{gathered}$ | $\begin{gathered} -0.067 \\ (0.413) \end{gathered}$ | | $A D U M$ | $\begin{gathered} 0.143 \\ (0.061) \end{gathered}$ | $\begin{gathered} -0.082 \\ (0.067) \end{gathered}$ | $\begin{gathered} 0.042 \\ (0.072) \end{gathered}$ | $\begin{gathered} 0.033 \\ (0.215) \end{gathered}$ | $\begin{gathered} -0.037 \\ (0.077) \end{gathered}$ | | I/K | $\begin{gathered} 0.440 \\ (0.156) \end{gathered}$ | $\begin{gathered} 0.114 \\ (0.152) \end{gathered}$ | $\begin{gathered} 0.157 \\ (0.099) \end{gathered}$ | $\begin{gathered} 0.280 \\ (0.191) \end{gathered}$ | $\begin{gathered} 0.144 \\ (0.106) \end{gathered}$ | | # Obs. <br> Adj. $R^{2}$ | $\begin{aligned} & 2630 \\ & 0.407 \end{aligned}$ | $\begin{aligned} & 2630 \\ & 0.584 \end{aligned}$ | $\begin{aligned} & 2630 \\ & 0.884 \end{aligned}$ | $\begin{aligned} & 764 \\ & 0.884 \end{aligned}$ | $\begin{aligned} & 1866 \\ & 0.831 \end{aligned}$ |
Table 4. Continued.  表 4. 续。
(B) Determinants of average equity ownership per manager
(B) 每位经理的平均股权拥有权的决定因素
The specifications reported in this table all model the average equity owned by top managers, e, by regressing the dependent variable L N ( e ) L N ( e ) LN(e)L N(e) on the explanatory variables indicated below. Intercept terms and year dummies are included for all regressions, but not reported. Fixed effects at the industry or firm level are included where indicated, but not reported. Variable definitions for the acronyms are given in Table 3.
本表中报告的规格均通过将因变量 L N ( e ) L N ( e ) LN(e)L N(e) 对下面所示的解释变量进行回归,来建模高管所拥有的平均股权 e。所有回归均包含截距项和年份虚拟变量,但未报告。行业或公司层面的固定效应在指示的地方包含,但未报告。缩略语的变量定义见表 3。
Variable  变量 All firms (Pooled)  所有公司(合并) All firm (SIC3 effects)
所有公司(SIC3 效应)
All firm (Firm effects)
所有公司(公司效应)

财富 500 强(公司效应)
Fortune 500
(Firm effects)
Fortune 500 (Firm effects)| Fortune 500 | | :--- | | (Firm effects) |
  非 500(公司效应)
Non-500
(Firm effects)
Non-500 (Firm effects)| Non-500 | | :--- | | (Firm effects) |
L N ( S ) L N ( S ) LN(S)L N(S) 0.334 ( 0.056 ) 0.334 ( 0.056 ) {:[0.334],[(0.056)]:}\begin{gathered} 0.334 \\ (0.056) \end{gathered} 0.387 ( 0.067 ) 0.387 ( 0.067 ) {:[0.387],[(0.067)]:}\begin{gathered} 0.387 \\ (0.067) \end{gathered} 0.066 ( 0.112 ) 0.066 ( 0.112 ) {:[0.066],[(0.112)]:}\begin{gathered} 0.066 \\ (0.112) \end{gathered} 0.328 ( 0.742 ) 0.328 ( 0.742 ) {:[-0.328],[(0.742)]:}\begin{gathered} -0.328 \\ (0.742) \end{gathered} 0.053 ( 0.145 ) 0.053 ( 0.145 ) {:[0.053],[(0.145)]:}\begin{gathered} 0.053 \\ (0.145) \end{gathered}
( L N ( S ) ) 2 ( L N ( S ) ) 2 (LN(S))^(2)(L N(S))^{2} 0.008 ( 0.005 ) 0.008 ( 0.005 ) {:[-0.008],[(0.005)]:}\begin{gathered} -0.008 \\ (0.005) \end{gathered} 0.012 ( 0.006 ) 0.012 ( 0.006 ) {:[-0.012],[(0.006)]:}\begin{gathered} -0.012 \\ (0.006) \end{gathered} 0.030 ( 0.011 ) 0.030 ( 0.011 ) {:[0.030],[(0.011)]:}\begin{gathered} 0.030 \\ (0.011) \end{gathered} 0.032 ( 0.049 ) 0.032 ( 0.049 ) {:[0.032],[(0.049)]:}\begin{gathered} 0.032 \\ (0.049) \end{gathered} 0.041 ( 0.018 ) 0.041 ( 0.018 ) {:[0.041],[(0.018)]:}\begin{gathered} 0.041 \\ (0.018) \end{gathered}
K/S 1.044 ( 0.255 ) 1.044 ( 0.255 ) {:[1.044],[(0.255)]:}\begin{gathered} 1.044 \\ (0.255) \end{gathered} 1.629 ( 0.302 ) 1.629 ( 0.302 ) {:[1.629],[(0.302)]:}\begin{gathered} 1.629 \\ (0.302) \end{gathered} 0.830 ( 0.300 ) 0.830 ( 0.300 ) {:[0.830],[(0.300)]:}\begin{gathered} 0.830 \\ (0.300) \end{gathered} 0.510 ( 0.600 ) 0.510 ( 0.600 ) {:[0.510],[(0.600)]:}\begin{gathered} 0.510 \\ (0.600) \end{gathered} 0.888 ( 0.355 ) 0.888 ( 0.355 ) {:[0.888],[(0.355)]:}\begin{gathered} 0.888 \\ (0.355) \end{gathered}
( K / S ) 2 ( K / S ) 2 (K//S)^(2)(K / S)^{2} 0.783 ( 0.154 ) 0.783 ( 0.154 ) {:[-0.783],[(0.154)]:}\begin{gathered} -0.783 \\ (0.154) \end{gathered} 0.892 ( 0.160 ) 0.892 ( 0.160 ) {:[-0.892],[(0.160)]:}\begin{gathered} -0.892 \\ (0.160) \end{gathered} 0.188 ( 0.137 ) 0.188 ( 0.137 ) {:[-0.188],[(0.137)]:}\begin{array}{r} -0.188 \\ (0.137) \end{array} 0.095 ( 0.256 ) 0.095 ( 0.256 ) {:[-0.095],[(0.256)]:}\begin{gathered} -0.095 \\ (0.256) \end{gathered} 0.253 ( 0.161 ) 0.253 ( 0.161 ) {:[-0.253],[(0.161)]:}\begin{gathered} -0.253 \\ (0.161) \end{gathered}
SIGMA 18.5 ( 2.14 ) 18.5 ( 2.14 ) {:[-18.5],[(2.14)]:}\begin{gathered} -18.5 \\ (2.14) \end{gathered} 18.3 ( 2.12 ) 18.3 ( 2.12 ) {:[-18.3],[(2.12)]:}\begin{gathered} -18.3 \\ (2.12) \end{gathered} 14.5 ( 1.68 ) 14.5 ( 1.68 ) {:[-14.5],[(1.68)]:}\begin{gathered} -14.5 \\ (1.68) \end{gathered} 24.6 ( 15.3 ) 24.6 ( 15.3 ) {:[-24.6],[(15.3)]:}\begin{array}{r} -24.6 \\ (15.3) \end{array} 12.6 ( 1.67 ) 12.6 ( 1.67 ) {:[-12.6],[(1.67)]:}\begin{gathered} -12.6 \\ (1.67) \end{gathered}
SIGDUM 0.089 ( 0.101 ) 0.089 ( 0.101 ) {:[0.089],[(0.101)]:}\begin{gathered} 0.089 \\ (0.101) \end{gathered} 0.915 ( 0.100 ) 0.915 ( 0.100 ) {:[0.915],[(0.100)]:}\begin{gathered} 0.915 \\ (0.100) \end{gathered} 0.598 ( 0.115 ) 0.598 ( 0.115 ) {:[0.598],[(0.115)]:}\begin{gathered} 0.598 \\ (0.115) \end{gathered} 1.63 ( 0.526 ) 1.63 ( 0.526 ) {:[1.63],[(0.526)]:}\begin{aligned} & 1.63 \\ & (0.526) \end{aligned} 0.412 ( 0.107 ) 0.412 ( 0.107 ) {:[0.412],[(0.107)]:}\begin{gathered} 0.412 \\ (0.107) \end{gathered}
Y / S Y / S Y//SY / S 1.58 ( 0.326 ) 1.58 ( 0.326 ) {:[1.58],[(0.326)]:}\begin{aligned} & 1.58 \\ & (0.326) \end{aligned} 1.14 ( 0.335 ) 1.14 ( 0.335 ) {:[1.14],[(0.335)]:}\begin{gathered} 1.14 \\ (0.335) \end{gathered} 1.80 ( 0.242 ) 1.80 ( 0.242 ) {:[1.80],[(0.242)]:}\begin{aligned} & 1.80 \\ & (0.242) \end{aligned} 4.63 ( 0.830 ) 4.63 ( 0.830 ) {:[4.63],[(0.830)]:}\begin{aligned} & 4.63 \\ & (0.830) \end{aligned} 1.56 ( 0.235 ) 1.56 ( 0.235 ) {:[1.56],[(0.235)]:}\begin{gathered} 1.56 \\ (0.235) \end{gathered}
( R & D ) / K ( R & D ) / K (R&D)//K(R \& D) / K 0.174 ( 0.201 ) 0.174 ( 0.201 ) {:[-0.174],[(0.201)]:}\begin{array}{r} -0.174 \\ (0.201) \end{array} 0.154 ( 0.223 ) 0.154 ( 0.223 ) {:[0.154],[(0.223)]:}\begin{gathered} 0.154 \\ (0.223) \end{gathered} 0.380 ( 0.348 ) 0.380 ( 0.348 ) {:[0.380],[(0.348)]:}\begin{gathered} 0.380 \\ (0.348) \end{gathered} 3.79 ( 1.48 ) 3.79 ( 1.48 ) {:[3.79],[(1.48)]:}\begin{gathered} 3.79 \\ (1.48) \end{gathered} 0.282 ( 0.353 ) 0.282 ( 0.353 ) {:[0.282],[(0.353)]:}\begin{gathered} 0.282 \\ (0.353) \end{gathered}
RDUM 0.212 ( 0.065 ) 0.212 ( 0.065 ) {:[0.212],[(0.065)]:}\begin{gathered} 0.212 \\ (0.065) \end{gathered} 0.003 ( 0.100 ) 0.003 ( 0.100 ) {:[0.003],[(0.100)]:}\begin{gathered} 0.003 \\ (0.100) \end{gathered} 0.430 ( 0.116 ) 0.430 ( 0.116 ) {:[0.430],[(0.116)]:}\begin{gathered} 0.430 \\ (0.116) \end{gathered} 0.597 ( 0.315 ) 0.597 ( 0.315 ) {:[0.597],[(0.315)]:}\begin{gathered} 0.597 \\ (0.315) \end{gathered} 0.379 ( 0.125 ) 0.379 ( 0.125 ) {:[0.379],[(0.125)]:}\begin{gathered} 0.379 \\ (0.125) \end{gathered}
A/K 0.139 ( 0.225 ) 0.139 ( 0.225 ) {:[-0.139],[(0.225)]:}\begin{gathered} -0.139 \\ (0.225) \end{gathered} 0.834 ( 0.418 ) 0.834 ( 0.418 ) {:[0.834],[(0.418)]:}\begin{gathered} 0.834 \\ (0.418) \end{gathered} 0.235 ( 0.658 ) 0.235 ( 0.658 ) {:[0.235],[(0.658)]:}\begin{gathered} 0.235 \\ (0.658) \end{gathered} 3.34 ( 1.08 ) 3.34 ( 1.08 ) {:[3.34],[(1.08)]:}\begin{gathered} 3.34 \\ (1.08) \end{gathered} 0.121 ( 0.662 ) 0.121 ( 0.662 ) {:[0.121],[(0.662)]:}\begin{gathered} 0.121 \\ (0.662) \end{gathered}
ADUM 0.314 ( 0.067 ) 0.314 ( 0.067 ) {:[0.314],[(0.067)]:}\begin{gathered} 0.314 \\ (0.067) \end{gathered} 0.080 ( 0.076 ) 0.080 ( 0.076 ) {:[-0.080],[(0.076)]:}\begin{gathered} -0.080 \\ (0.076) \end{gathered} 0.030 ( 0.082 ) 0.030 ( 0.082 ) {:[0.030],[(0.082)]:}\begin{gathered} 0.030 \\ (0.082) \end{gathered} 0.276 ( 0.261 ) 0.276 ( 0.261 ) {:[0.276],[(0.261)]:}\begin{gathered} 0.276 \\ (0.261) \end{gathered} 0.012 ( 0.090 ) 0.012 ( 0.090 ) {:[0.012],[(0.090)]:}\begin{gathered} 0.012 \\ (0.090) \end{gathered}
I/K 1.429 ( 0.174 ) 1.429 ( 0.174 ) {:[1.429],[(0.174)]:}\begin{gathered} 1.429 \\ (0.174) \end{gathered} 1.06 ( 0.173 ) 1.06 ( 0.173 ) {:[1.06],[(0.173)]:}\begin{gathered} 1.06 \\ (0.173) \end{gathered} 0.575 ( 0.117 ) 0.575 ( 0.117 ) {:[0.575],[(0.117)]:}\begin{gathered} 0.575 \\ (0.117) \end{gathered} 1.04 ( 0.251 ) 1.04 ( 0.251 ) {:[1.04],[(0.251)]:}\begin{gathered} 1.04 \\ (0.251) \end{gathered} 0.525 ( 0.128 ) 0.525 ( 0.128 ) {:[0.525],[(0.128)]:}\begin{gathered} 0.525 \\ (0.128) \end{gathered}
\# Obs. Adj. R 2  \# Obs.   Adj.  R 2 {:[" \# Obs. "],[" Adj. "R^(2)]:}\begin{aligned} & \text { \# Obs. } \\ & \text { Adj. } R^{2} \end{aligned} 2628 0.300 2628 0.300 {:[2628],[0.300]:}\begin{aligned} & 2628 \\ & 0.300 \end{aligned} 2628 0.446 2628 0.446 {:[2628],[0.446]:}\begin{aligned} & 2628 \\ & 0.446 \end{aligned} 2628 0.818 2628 0.818 {:[2628],[0.818]:}\begin{aligned} & 2628 \\ & 0.818 \end{aligned} 763 0.838 763 0.838 {:[763],[0.838]:}\begin{aligned} & 763 \\ & 0.838 \end{aligned} 1865 0.770 1865 0.770 {:[1865],[0.770]:}\begin{aligned} & 1865 \\ & 0.770 \end{aligned}
Variable All firms (Pooled) All firm (SIC3 effects) All firm (Firm effects) "Fortune 500 (Firm effects)" "Non-500 (Firm effects)" LN(S) "0.334 (0.056)" "0.387 (0.067)" "0.066 (0.112)" "-0.328 (0.742)" "0.053 (0.145)" (LN(S))^(2) "-0.008 (0.005)" "-0.012 (0.006)" "0.030 (0.011)" "0.032 (0.049)" "0.041 (0.018)" K/S "1.044 (0.255)" "1.629 (0.302)" "0.830 (0.300)" "0.510 (0.600)" "0.888 (0.355)" (K//S)^(2) "-0.783 (0.154)" "-0.892 (0.160)" "-0.188 (0.137)" "-0.095 (0.256)" "-0.253 (0.161)" SIGMA "-18.5 (2.14)" "-18.3 (2.12)" "-14.5 (1.68)" "-24.6 (15.3)" "-12.6 (1.67)" SIGDUM "0.089 (0.101)" "0.915 (0.100)" "0.598 (0.115)" "1.63 (0.526)" "0.412 (0.107)" Y//S "1.58 (0.326)" "1.14 (0.335)" "1.80 (0.242)" "4.63 (0.830)" "1.56 (0.235)" (R&D)//K "-0.174 (0.201)" "0.154 (0.223)" "0.380 (0.348)" "3.79 (1.48)" "0.282 (0.353)" RDUM "0.212 (0.065)" "0.003 (0.100)" "0.430 (0.116)" "0.597 (0.315)" "0.379 (0.125)" A/K "-0.139 (0.225)" "0.834 (0.418)" "0.235 (0.658)" "3.34 (1.08)" "0.121 (0.662)" ADUM "0.314 (0.067)" "-0.080 (0.076)" "0.030 (0.082)" "0.276 (0.261)" "0.012 (0.090)" I/K "1.429 (0.174)" "1.06 (0.173)" "0.575 (0.117)" "1.04 (0.251)" "0.525 (0.128)" " \# Obs. Adj. R^(2)" "2628 0.300" "2628 0.446" "2628 0.818" "763 0.838" "1865 0.770"| Variable | All firms (Pooled) | All firm (SIC3 effects) | All firm (Firm effects) | Fortune 500 <br> (Firm effects) | Non-500 <br> (Firm effects) | | :---: | :---: | :---: | :---: | :---: | :---: | | $L N(S)$ | $\begin{gathered} 0.334 \\ (0.056) \end{gathered}$ | $\begin{gathered} 0.387 \\ (0.067) \end{gathered}$ | $\begin{gathered} 0.066 \\ (0.112) \end{gathered}$ | $\begin{gathered} -0.328 \\ (0.742) \end{gathered}$ | $\begin{gathered} 0.053 \\ (0.145) \end{gathered}$ | | $(L N(S))^{2}$ | $\begin{gathered} -0.008 \\ (0.005) \end{gathered}$ | $\begin{gathered} -0.012 \\ (0.006) \end{gathered}$ | $\begin{gathered} 0.030 \\ (0.011) \end{gathered}$ | $\begin{gathered} 0.032 \\ (0.049) \end{gathered}$ | $\begin{gathered} 0.041 \\ (0.018) \end{gathered}$ | | K/S | $\begin{gathered} 1.044 \\ (0.255) \end{gathered}$ | $\begin{gathered} 1.629 \\ (0.302) \end{gathered}$ | $\begin{gathered} 0.830 \\ (0.300) \end{gathered}$ | $\begin{gathered} 0.510 \\ (0.600) \end{gathered}$ | $\begin{gathered} 0.888 \\ (0.355) \end{gathered}$ | | $(K / S)^{2}$ | $\begin{gathered} -0.783 \\ (0.154) \end{gathered}$ | $\begin{gathered} -0.892 \\ (0.160) \end{gathered}$ | $\begin{array}{r} -0.188 \\ (0.137) \end{array}$ | $\begin{gathered} -0.095 \\ (0.256) \end{gathered}$ | $\begin{gathered} -0.253 \\ (0.161) \end{gathered}$ | | SIGMA | $\begin{gathered} -18.5 \\ (2.14) \end{gathered}$ | $\begin{gathered} -18.3 \\ (2.12) \end{gathered}$ | $\begin{gathered} -14.5 \\ (1.68) \end{gathered}$ | $\begin{array}{r} -24.6 \\ (15.3) \end{array}$ | $\begin{gathered} -12.6 \\ (1.67) \end{gathered}$ | | SIGDUM | $\begin{gathered} 0.089 \\ (0.101) \end{gathered}$ | $\begin{gathered} 0.915 \\ (0.100) \end{gathered}$ | $\begin{gathered} 0.598 \\ (0.115) \end{gathered}$ | $\begin{aligned} & 1.63 \\ & (0.526) \end{aligned}$ | $\begin{gathered} 0.412 \\ (0.107) \end{gathered}$ | | $Y / S$ | $\begin{aligned} & 1.58 \\ & (0.326) \end{aligned}$ | $\begin{gathered} 1.14 \\ (0.335) \end{gathered}$ | $\begin{aligned} & 1.80 \\ & (0.242) \end{aligned}$ | $\begin{aligned} & 4.63 \\ & (0.830) \end{aligned}$ | $\begin{gathered} 1.56 \\ (0.235) \end{gathered}$ | | $(R \& D) / K$ | $\begin{array}{r} -0.174 \\ (0.201) \end{array}$ | $\begin{gathered} 0.154 \\ (0.223) \end{gathered}$ | $\begin{gathered} 0.380 \\ (0.348) \end{gathered}$ | $\begin{gathered} 3.79 \\ (1.48) \end{gathered}$ | $\begin{gathered} 0.282 \\ (0.353) \end{gathered}$ | | RDUM | $\begin{gathered} 0.212 \\ (0.065) \end{gathered}$ | $\begin{gathered} 0.003 \\ (0.100) \end{gathered}$ | $\begin{gathered} 0.430 \\ (0.116) \end{gathered}$ | $\begin{gathered} 0.597 \\ (0.315) \end{gathered}$ | $\begin{gathered} 0.379 \\ (0.125) \end{gathered}$ | | A/K | $\begin{gathered} -0.139 \\ (0.225) \end{gathered}$ | $\begin{gathered} 0.834 \\ (0.418) \end{gathered}$ | $\begin{gathered} 0.235 \\ (0.658) \end{gathered}$ | $\begin{gathered} 3.34 \\ (1.08) \end{gathered}$ | $\begin{gathered} 0.121 \\ (0.662) \end{gathered}$ | | ADUM | $\begin{gathered} 0.314 \\ (0.067) \end{gathered}$ | $\begin{gathered} -0.080 \\ (0.076) \end{gathered}$ | $\begin{gathered} 0.030 \\ (0.082) \end{gathered}$ | $\begin{gathered} 0.276 \\ (0.261) \end{gathered}$ | $\begin{gathered} 0.012 \\ (0.090) \end{gathered}$ | | I/K | $\begin{gathered} 1.429 \\ (0.174) \end{gathered}$ | $\begin{gathered} 1.06 \\ (0.173) \end{gathered}$ | $\begin{gathered} 0.575 \\ (0.117) \end{gathered}$ | $\begin{gathered} 1.04 \\ (0.251) \end{gathered}$ | $\begin{gathered} 0.525 \\ (0.128) \end{gathered}$ | | $\begin{aligned} & \text { \# Obs. } \\ & \text { Adj. } R^{2} \end{aligned}$ | $\begin{aligned} & 2628 \\ & 0.300 \end{aligned}$ | $\begin{aligned} & 2628 \\ & 0.446 \end{aligned}$ | $\begin{aligned} & 2628 \\ & 0.818 \end{aligned}$ | $\begin{aligned} & 763 \\ & 0.838 \end{aligned}$ | $\begin{aligned} & 1865 \\ & 0.770 \end{aligned}$ |
Notes: Estimated standard errors (reported in parentheses) are consistent in the presence of heteroskedasticity. The adjusted R 2 R 2 R^(2)R^{2} statistics reflect the inclusion of fixed effects (where included).
注意:估计的标准误差(括号内报告)在异方差存在的情况下是一致的。调整后的 R 2 R 2 R^(2)R^{2} 统计量反映了固定效应的包含(如果包含的话)。
The specifications reported in the second and third columns of Table 4A control for unobserved heterogeneity at the industry level and firm level, respectively. The second column includes fixed three-digit SIC effects; the third column includes fixed firm effects. Demsetz and Lehn (1985) included controls for certain (regulated) industries. By including fixed industry effects (and, in some cases, fixed firm effects), we control for industry influences generally. The inclusion of fixed effects changes the estimated coefficients significantly in some cases. For example, if we do not control for unobserved industry-level or firm-level heterogeneity, the estimated coefficients on size and investment rate are significantly larger in absolute value, and the estimated coefficient of the ratio of R&D spending to capital changes sign. These differences suggest that the unobserved firm characteristics are correlated with the observed characteristics, and therefore bias the estimated coefficients in a cross-sectional or pooled regression. For example, in a univariate regression, if there were a strong positive equilibrium relation between R & D R & D R&D\mathrm{R} \& \mathrm{D} intensity and managerial ownership, excluding firm fixed effects would bias downward the estimated coefficient on R & D / K R & D / K R&D//KR \& D / K in a pooled regression.
表 4A 的第二和第三列报告的规格分别控制了行业层面和公司层面的未观察到的异质性。第二列包括固定的三位数 SIC 效应;第三列包括固定的公司效应。Demsetz 和 Lehn(1985)对某些(受监管的)行业进行了控制。通过包括固定的行业效应(在某些情况下,还包括固定的公司效应),我们一般控制了行业影响。固定效应的包含在某些情况下显著改变了估计系数。例如,如果我们不控制未观察到的行业层面或公司层面的异质性,规模和投资率的估计系数在绝对值上显著更大,研发支出与资本的比率的估计系数则改变了符号。这些差异表明,未观察到的公司特征与观察到的特征相关,因此在横截面或合并回归中偏倚了估计系数。 例如,在单变量回归中,如果 R & D R & D R&D\mathrm{R} \& \mathrm{D} 强度与管理层持股之间存在强正平衡关系,排除公司固定效应将导致在合并回归中对 R & D / K R & D / K R&D//KR \& D / K 的估计系数向下偏差。
The fourth and fifth columns of Table 4A report results from splitting the sample according to whether the firm is in the Fortune 500 in the given year (including firm-level fixed effects). Some subsample differences emerge. The negative effect of idiosyncratic risk (measured by SIGMA) on ownership is traced to non-Fortune 500 firms, consistent with our earlier interpretation. Effects of capital intensity, operating income, R&D intensity, advertising intensity, and the investment rate are larger in absolute value for larger firms.
表 4A 的第四和第五列报告了根据公司在给定年份是否进入《财富》500 强(包括公司层面的固定效应)对样本进行拆分的结果。一些子样本差异出现。特有风险(通过 SIGMA 衡量)对所有权的负面影响归因于非《财富》500 强公司,这与我们之前的解释一致。资本密集度、营业收入、研发强度、广告强度和投资率的影响在大公司中绝对值更大。
Because theoretical models generally emphasize managerial ownership levels relative to the managers’ wealth and not simply the fraction of firm equity held by managers, we present in Table 4B results from the same models presented in Table 4A, but with the dependent variable being the log of managerial equity per manager. (We do not observe managerial wealth, so we focus only on managerial equity.) Broadly speaking, the patterns we identified in Table 4A carry over to the estimates in Table 4B. One difference is that the estimated coefficient on the capital-to-sales ratio is everywhere positive and statistically significantly different from zero. This could reflect the fact that capital-intensive firms employ relatively fewer workers and managers, but have higher levels of value added per worker, and hence derive larger incentive benefits from higher levels of managerial ownership.
因为理论模型通常强调管理者的所有权水平相对于管理者的财富,而不仅仅是管理者持有的公司股权比例,我们在表 4B 中展示了与表 4A 中相同模型的结果,但因变量为每位管理者的管理股权的对数。(我们无法观察管理者的财富,因此我们仅关注管理股权。)总体而言,我们在表 4A 中识别的模式延续到表 4B 中的估计结果。一个不同之处在于,资本与销售比率的估计系数在各处都是正的,并且在统计上显著不同于零。这可能反映了资本密集型公司雇用相对较少的工人和管理者,但每位工人的附加值水平更高,因此从更高水平的管理所有权中获得更大的激励收益。
Taken together, the results presented in Table 4A and B suggest strongly that observable firm characteristics in the contracting environment influence managerial ownership. In addition, unobserved firm characteristics are correlated with observed characteristics, making coefficients estimated using panel data more reliable than those estimated using cross-sectional data. The beneficial ownership data include options exercisable within 60 days, but omit recent awards that are not yet vested. Because we lack data on all of the stock options
综合来看,表 4A 和 B 中呈现的结果强烈表明,合同环境中的可观察公司特征会影响管理层持股。此外,未观察到的公司特征与观察到的特征相关,使得使用面板数据估计的系数比使用横截面数据估计的系数更可靠。受益所有权数据包括在 60 天内可行使的期权,但省略了尚未归属的最近授予的期权。因为我们缺乏所有股票期权的数据。

granted to all top managers, we do not investigate the substitutability of direct ownership stakes and stock options as mechanisms to align incentives for value maximization. In the ExecuComp data over the 1992-1996 period, however, the correlation between the pay-performance sensitivity for managers using the ‘stock’ definition and the pay-performance sensitivity using the stock plus options definition exceeds 0.95 . 1 0.95 . 1 0.95.^(1)0.95 .{ }^{1} Thus our focus on the beneficial ownership data appears warranted.
授予所有高管,我们不研究直接股权和股票期权作为对齐价值最大化激励机制的可替代性。然而,在 1992-1996 年期间的 ExecuComp 数据中,使用“股票”定义的经理薪酬与业绩敏感性之间的相关性与使用股票加期权定义的薪酬与业绩敏感性之间的相关性超过 0.95 . 1 0.95 . 1 0.95.^(1)0.95 .{ }^{1} 。因此,我们对有益所有权数据的关注似乎是合理的。

5. Managerial ownership and firm performance
5. 管理层持股与公司绩效

5.1. Evidence on the exogeneity of managerial ownership
5.1. 管理层持股外生性的证据

Thus far, we have emphasized that managerial stakes are part of a larger set of equilibrium contracts undertaken by the firm to align incentives for value maximization, and we have shown that managerial ownership can be explained by observable characteristics of the firm’s contracting environment, such as stock price volatility and the composition of assets, as predicted by the contracting view. These results also show, however, that even when industry dummies are included, many important features of the firm’s contracting environment remain unobserved. Specifically, including firm-level fixed dummy variables raises the adjusted R 2 R 2 R^(2)R^{2} from 0.584 to 0.884 . These results cast doubt on the assumption that managerial ownership is exogenous in regressions that attempt to measure the impact of ownership on performance by regressing variables like Tobin’s Q Q QQ on managerial ownership without controlling for fixed effects.
到目前为止,我们强调管理层的利益是公司为实现价值最大化而采取的一系列均衡合同的一部分,并且我们已经表明,管理层的所有权可以通过公司合同环境的可观察特征来解释,例如股价波动和资产组成,这些都是合同视角所预测的。然而,这些结果也表明,即使包括行业虚拟变量,公司的合同环境的许多重要特征仍然未被观察到。具体而言,包括公司层面的固定虚拟变量将调整后的 R 2 R 2 R^(2)R^{2} 从 0.584 提高到 0.884。这些结果对假设管理层所有权在试图通过回归像托宾的 Q Q QQ 这样的变量与管理层所有权之间的关系时是外生的这一假设提出了质疑,而没有控制固定效应。

In this section, we use panel data techniques to investigate more directly the question of whether managerial ownership can be treated as exogenous in the performance regressions. We use Tobin’s Q Q QQ as our measure of firm performance, but our results are robust to using return on assets as the dependent variable (tables are available upon request). To investigate the impact of managerial ownership on Q Q QQ, we use variants of the reduced-form model in Eq. (3), in which Q Q QQ depends upon managerial ownership, m m mm, observable firm characteristics, x x xx, and unobserved firm characteristics, u u uu. We use two specifications of managerial ownership in the Q Q QQ regression. The first includes m m mm and m 2 m 2 m^(2)m^{2} (see McConnell and Servaes, 1990). The second includes three piecewise-linear terms in m m mm (as in Mørck et al., 1988). Specifically,
在本节中,我们使用面板数据技术更直接地研究管理层持股是否可以在绩效回归中视为外生变量。我们使用托宾的 Q Q QQ 作为公司绩效的衡量标准,但我们的结果在使用资产回报率作为因变量时依然稳健(表格可根据要求提供)。为了研究管理层持股对 Q Q QQ 的影响,我们使用了方程(3)中简化模型的变体,其中 Q Q QQ 依赖于管理层持股、 m m mm 、可观察的公司特征、 x x xx 和不可观察的公司特征 u u uu 。我们在 Q Q QQ 回归中使用了两种管理层持股的规格。第一种包括 m m mm m 2 m 2 m^(2)m^{2} (见 McConnell 和 Servaes,1990)。第二种包括 m m mm 中的三个分段线性项(如 Mørck 等,1988)。具体来说,

m 1 = { managerial ownership level if managerial ownership level < 0.05 , 0.05 if managerial ownership level 0.05 ; m 1 =  managerial ownership level   if managerial ownership level  < 0.05 , 0.05  if managerial ownership level  0.05 ; m1={[" managerial ownership level "," if managerial ownership level " < 0.05","],[0.05," if managerial ownership level " >= 0.05;]:}m 1= \begin{cases}\text { managerial ownership level } & \text { if managerial ownership level }<0.05, \\ 0.05 & \text { if managerial ownership level } \geqslant 0.05 ;\end{cases}
m 2 = { zero if managerial ownership level < 0.05 , managerial ownership if 0.05 managerial ownership level < 0.25 , level minus 0.05 0.20 if managerial ownership level 0.25 ; m 2 =  zero   if managerial ownership level  < 0.05 ,  managerial ownership   if  0.05  managerial ownership level  < 0.25 ,  level minus  0.05 0.20  if managerial ownership level  0.25 ; m2={[" zero "," if managerial ownership level " < 0.05","],[" managerial ownership "," if "0.05 <= " managerial ownership level " < 0.25","],[" level minus "0.05,],[0.20," if managerial ownership level " >= 0.25;]:}m 2= \begin{cases}\text { zero } & \text { if managerial ownership level }<0.05, \\ \text { managerial ownership } & \text { if } 0.05 \leqslant \text { managerial ownership level }<0.25, \\ \text { level minus } 0.05 & \\ 0.20 & \text { if managerial ownership level } \geqslant 0.25 ;\end{cases}
m 3 = { zero if managerial ownership level < 0.25 , managerial ownership if managerial ownership 0.25 . level minus 0.25 m 3 =  zero   if managerial ownership level  < 0.25 ,  managerial ownership   if managerial ownership  0.25 .  level minus  0.25 m3={[" zero "," if managerial ownership level " < 0.25","],[" managerial ownership "," if managerial ownership " >= 0.25.],[" level minus "0.25,]:}m 3= \begin{cases}\text { zero } & \text { if managerial ownership level }<0.25, \\ \text { managerial ownership } & \text { if managerial ownership } \geqslant 0.25 . \\ \text { level minus } 0.25 & \end{cases}
For observable characteristics, we use the same vector of x x xx variables used in the model for managerial ownership. We report results including and excluding arguably endogenous ‘investment’ variables (R&D, advertising, and fixed capital).
对于可观察的特征,我们使用与管理层持股模型中相同的 x x xx 变量向量。我们报告包括和不包括可以说是内生的“投资”变量(研发、广告和固定资本)的结果。
Our empirical analysis of the effects of managerial ownership and firm characteristics on Q Q QQ is summarized in Table 5 A and B . Table 5A reports estimated coefficients for cases in which managerial ownership is represented by m m mm and m 2 m 2 m^(2)m^{2}. Table 5B reports estimated coefficients for cases in which managerial ownership is represented by the piecewise-linear terms, m 1 , m 2 m 1 , m 2 m1,m2m 1, m 2, and m 3 m 3 m3m 3. For both of the above specifications, we report estimated coefficients for (1) regressions with managerial ownership alone (pooled, SIC3 industry effects, and firm effects), (2) the regressions including the full set of x x xx variables (pooled, SIC3 industry effects, and firm effects), and (3) the regressions including the noninvestment set of x x xx variables. All specifications include year effects (not reported).
我们对管理层持股和公司特征对 Q Q QQ 影响的实证分析总结在表 5 A 和 B 中。表 5A 报告了管理层持股由 m m mm m 2 m 2 m^(2)m^{2} 表示的情况下的估计系数。表 5B 报告了管理层持股由分段线性项 m 1 , m 2 m 1 , m 2 m1,m2m 1, m 2 m 3 m 3 m3m 3 表示的情况下的估计系数。对于上述两种规格,我们报告了以下回归的估计系数:(1) 仅包含管理层持股的回归(合并、SIC3 行业效应和公司效应),(2) 包含完整 x x xx 变量集的回归(合并、SIC3 行业效应和公司效应),以及 (3) 包含非投资 x x xx 变量集的回归。所有规格均包括年份效应(未报告)。
Turning first to the quadratic specifications of managerial ownership in Table 5A, we note that the managerial ownership variables are statistically significant only in the pooled model with no other variables and in the model with only industry effects. In other specifications, the managerial ownership coefficients are virtually never statistically significantly different from zero. (The Wald test for the joint significance of m m mm and m 2 m 2 m^(2)m^{2} is reported at the bottom of the table.) Once we control for observed firm characteristics ( x x xx ), or for unobserved firm characteristics (in the firm-fixed-effect version of u u uu ), there is no effect of changes in managerial ownership on Q Q QQ. Though not reported in Table 5A, these results hold for both the Fortune 500 and non-Fortune 500 subsamples considered earlier.
首先看一下表 5A 中管理层持股的二次规格,我们注意到管理层持股变量在没有其他变量的合并模型中以及仅在行业效应模型中是统计显著的。在其他规格中,管理层持股系数几乎从未在统计上显著不同于零。(表底部报告了 m m mm m 2 m 2 m^(2)m^{2} 的联合显著性 Wald 检验。)一旦我们控制了观察到的公司特征( x x xx ),或控制了未观察到的公司特征(在 u u uu 的公司固定效应版本中),管理层持股的变化对 Q Q QQ 没有影响。尽管在表 5A 中未报告,但这些结果在之前考虑的财富 500 强和非财富 500 强子样本中均成立。
Turning to the spline specifications for managerial ownership in Table 5B, the pooled results are consistent with those of Mørck et al. (1988), who find that the impact of m m mm on Q Q QQ increases at a decreasing rate, and thereafter declines. In contrast to the quadratic specification for managerial ownership, the Mørck-Shleifer-Vishny specification is robust to the inclusion of observable
转向表 5B 中关于管理层持股的样条规格,汇总结果与 Mørck 等人(1988)的研究一致,他们发现 m m mm Q Q QQ 的影响以递减的速度增加,然后下降。与管理层持股的二次规格相比,Mørck-Shleifer-Vishny 规格在包含可观察变量时是稳健的。

contracting determinants and industry dummies. Once we control for x x xx variables and for u u uu (via firm fixed effects), however, changes in managerial ownership levels have no statistically significant effect on Q Q QQ. These results hold both for the Fortune 500 and non-Fortune 500 subsamples investigated earlier.
合同决定因素和行业虚拟变量。一旦我们控制了 x x xx 变量和 u u uu (通过公司固定效应),然而,管理层持股水平的变化对 Q Q QQ 没有统计显著影响。这些结果在之前调查的财富 500 强和非财富 500 强子样本中均成立。
The results reported in Table 5A and B confirm the intuition of the contracting example sketched in Section 2. First, the results obtained when observed characteristics ( x x xx ) are included suggest that previously asserted relations between Q Q QQ and m m mm in part reflect equilibrium relations among Q Q QQ and firm characteristics in the firm’s contracting problem. Second, to the extent that firm characteristics unobserved by the econometrician influence the firm’s contracts and the equilibrium level of managerial ownership, the coefficient on m m mm in a Q Q QQ regression (when no attempt is made to incorporate the unobserved heterogeneity) is biased. Third, in keeping with our emphasis on contracting, the relations we estimate suggest that no inference can be made about the effect of ‘exogenous’ local increases in managerial ownership on firm performance.
表 5A 和 B 中报告的结果确认了第 2 节中概述的契约示例的直觉。首先,当观察到的特征( x x xx )被纳入时,获得的结果表明,之前所断言的 Q Q QQ m m mm 之间的关系在一定程度上反映了 Q Q QQ 与公司特征在公司契约问题中的均衡关系。其次,在经济计量学家未观察到的公司特征影响公司的合同和管理层所有权的均衡水平的情况下, Q Q QQ 回归中 m m mm 的系数(当没有尝试纳入未观察到的异质性时)是有偏的。第三,符合我们对契约的强调,我们估计的关系表明,无法对“外生”地方管理层所有权增加对公司绩效的影响做出推断。
One can formalize this evidence against the exogeneity of managerial ownership by testing for a correlation between the fixed effect and managerial ownership. We could use a Hausman (1978) test, but this test would tend to over-reject the null hypothesis of zero correlation because it would tend to reject if any of the explanatory variables were correlated with the fixed effect. To reduce this Type I error, we construct a more precise ‘conditional moment’ test, which is in the spirit of a Hausman test, but tends to reject only if managerial ownership is the source of the specification error (Greene, 1997, p. 534; Newey, 1985).
可以通过测试固定效应与管理层持股之间的相关性来正式化针对管理层持股外生性的证据。我们可以使用 Hausman(1978)检验,但该检验往往会过度拒绝零相关性的原假设,因为如果任何解释变量与固定效应相关,它就会倾向于拒绝。为了减少这种第一类错误,我们构建了一个更精确的“条件矩”检验,这与 Hausman 检验的精神相符,但仅在管理层持股是规范错误的来源时才倾向于拒绝(Greene, 1997, p. 534; Newey, 1985)。
The test is constructed as follows. Let the performance model be
测试的构建如下。设性能模型为
Q i t = β 0 + β z i t + u i + ε i t , Q i t = β 0 + β z i t + u i + ε i t , Q_(it)=beta_(0)+betaz_(it)+u_(i)+epsi_(it),Q_{i t}=\beta_{0}+\beta z_{i t}+u_{i}+\varepsilon_{i t},
where z i t z i t z_(it)z_{i t} includes the managerial ownership variables and the x x xx variables described earlier, and u i u i u_(i)u_{i} is the firm fixed effect. The formal hypothesis we want to test is whether the unobserved fixed effect, u i u i u_(i)u_{i}, is correlated with managerial ownership, an element of z i t z i t z_(it)z_{i t}. That is, H 0 : E ( m i t u i ) = 0 H 0 : E m i t u i = 0 H_(0):E(m_(it)*u_(i))=0H_{0}: \mathrm{E}\left(m_{i t} \cdot u_{i}\right)=0, where m i t m i t m_(it)m_{i t} is an r × 1 r × 1 r xx1r \times 1 vector of variables measuring the effect of managerial ownership. The idea of the test is to construct the simple analogue to the population moment, s = E ( m i t w i t ) s = E m i t w i t s=E(m_(it)w_(it))s=\mathrm{E}\left(m_{i t} w_{i t}\right), and then to test whether it is statistically significantly different from zero.
其中 z i t z i t z_(it)z_{i t} 包括管理层所有权变量和之前描述的 x x xx 变量,而 u i u i u_(i)u_{i} 是公司固定效应。我们想要检验的正式假设是未观察到的固定效应 u i u i u_(i)u_{i} 是否与管理层所有权相关,这是 z i t z i t z_(it)z_{i t} 的一个元素。也就是说, H 0 : E ( m i t u i ) = 0 H 0 : E m i t u i = 0 H_(0):E(m_(it)*u_(i))=0H_{0}: \mathrm{E}\left(m_{i t} \cdot u_{i}\right)=0 ,其中 m i t m i t m_(it)m_{i t} 是一个 r × 1 r × 1 r xx1r \times 1 变量的向量,用于衡量管理层所有权的影响。测试的思路是构建与总体时刻 s = E ( m i t w i t ) s = E m i t w i t s=E(m_(it)w_(it))s=\mathrm{E}\left(m_{i t} w_{i t}\right) 的简单类比,然后检验它是否在统计上显著不同于零。
Using a consistent ‘within’ estimator of β β beta\beta, we can construct consistent estimates of the residual w i t = u i + ε i t w i t = u i + ε i t w_(it)=u_(i)+epsi_(it)w_{i t}=u_{i}+\varepsilon_{i t}. Our test statistic is s ^ = i = 1 N t = 1 T i m i t w ^ i t / N T i s ^ = i = 1 N t = 1 T i m i t w ^ i t / N T i hat(s)=sum_(i=1)^(N)sum_(t=1)^(T_(i))m_(it) hat(w)_(it)//NT_(i)\hat{s}=\sum_{i=1}^{N} \sum_{t=1}^{T_{i}} m_{i t} \hat{w}_{i t} / N T_{i}, where T i T i T_(i)T_{i} is the number of observations for firm i i ii. Under standard regularity conditions and under the null hypothesis that E ( m i t u i ) = 0 , N s ^ E m i t u i = 0 , N s ^ E(m_(it)*u_(i))=0,sqrtN hat(s)\mathrm{E}\left(m_{i t} \cdot u_{i}\right)=0, \sqrt{N} \hat{s} will be asymptotically distributed N ( 0 , Σ ) N ( 0 , Σ ) N(0,Sigma)\mathrm{N}(0, \Sigma). Therefore the statistic k = N s ^ Σ ^ 1 S ^ k = N s ^ Σ ^ 1 S ^ k=N hat(s) hat(Sigma)^(-1) hat(S)k=N \hat{s} \hat{\Sigma}^{-1} \hat{S} is asymptotically chi-squared with r r rr degrees of freedom, where Σ ^ Σ ^ hat(Sigma)\hat{\Sigma} is a consistent estimate of Σ Σ Sigma\Sigma (for more details, see Greene, 1997).
使用一致的“内部”估计量 β β beta\beta ,我们可以构造残差 w i t = u i + ε i t w i t = u i + ε i t w_(it)=u_(i)+epsi_(it)w_{i t}=u_{i}+\varepsilon_{i t} 的一致估计。我们的检验统计量是 s ^ = i = 1 N t = 1 T i m i t w ^ i t / N T i s ^ = i = 1 N t = 1 T i m i t w ^ i t / N T i hat(s)=sum_(i=1)^(N)sum_(t=1)^(T_(i))m_(it) hat(w)_(it)//NT_(i)\hat{s}=\sum_{i=1}^{N} \sum_{t=1}^{T_{i}} m_{i t} \hat{w}_{i t} / N T_{i} ,其中 T i T i T_(i)T_{i} 是公司 i i ii 的观察数量。在标准的正则条件下,以及在零假设下, E ( m i t u i ) = 0 , N s ^ E m i t u i = 0 , N s ^ E(m_(it)*u_(i))=0,sqrtN hat(s)\mathrm{E}\left(m_{i t} \cdot u_{i}\right)=0, \sqrt{N} \hat{s} 将渐近分布为 N ( 0 , Σ ) N ( 0 , Σ ) N(0,Sigma)\mathrm{N}(0, \Sigma) 。因此,统计量 k = N s ^ Σ ^ 1 S ^ k = N s ^ Σ ^ 1 S ^ k=N hat(s) hat(Sigma)^(-1) hat(S)k=N \hat{s} \hat{\Sigma}^{-1} \hat{S} 在渐近情况下服从自由度为 r r rr 的卡方分布,其中 Σ ^ Σ ^ hat(Sigma)\hat{\Sigma} Σ Σ Sigma\Sigma 的一致估计(更多细节,请参见 Greene,1997)。

Table 5  表 5
(A) Determinants of firm value (Tobin’s Q Q QQ ), quadratic specification
(A) 企业价值的决定因素 (Tobin’s Q Q QQ ), 二次规格
Variable  变量 Pooled  合并的 SIC3 effects  SIC3 效果 Firm effects  公司效应 Pooled  合并的 SIC3 effects  SIC3 效果 Firm effects  公司效应 Pooled  合并的 SIC3 effects  SIC3 效果 Firm effects  公司效应
m m mm 0.539 ( 0.219 ) 0.539 ( 0.219 ) {:[0.539],[(0.219)]:}\begin{gathered} 0.539 \\ (0.219) \end{gathered} 1.25 ( 0.338 ) 1.25 ( 0.338 ) {:[1.25],[(0.338)]:}\begin{aligned} & 1.25 \\ & (0.338) \end{aligned} 0.573 ( 0.402 ) 0.573 ( 0.402 ) {:[0.573],[(0.402)]:}\begin{gathered} 0.573 \\ (0.402) \end{gathered} 0.460 ( 0.218 ) 0.460 ( 0.218 ) {:[-0.460],[(0.218)]:}\begin{gathered} -0.460 \\ (0.218) \end{gathered} 0.031 ( 0.277 ) 0.031 ( 0.277 ) {:[-0.031],[(0.277)]:}\begin{gathered} -0.031 \\ (0.277) \end{gathered} 0.125 ( 0.395 ) 0.125 ( 0.395 ) {:[0.125],[(0.395)]:}\begin{gathered} 0.125 \\ (0.395) \end{gathered} 0.395 ( 0.234 ) 0.395 ( 0.234 ) {:[-0.395],[(0.234)]:}\begin{gathered} -0.395 \\ (0.234) \end{gathered} 0.061 ( 0.281 ) 0.061 ( 0.281 ) {:[-0.061],[(0.281)]:}\begin{array}{r} -0.061 \\ (0.281) \end{array} 0.293 ( 0.392 ) 0.293 ( 0.392 ) {:[0.293],[(0.392)]:}\begin{gathered} 0.293 \\ (0.392) \end{gathered}
m 2 m 2 m^(2)m^{2} 1.123 ( 0.317 ) 1.123 ( 0.317 ) {:[-1.123],[(0.317)]:}\begin{gathered} -1.123 \\ (0.317) \end{gathered} 1.649 ( 0.457 ) 1.649 ( 0.457 ) {:[-1.649],[(0.457)]:}\begin{gathered} -1.649 \\ (0.457) \end{gathered} 0.582 ( 0.559 ) 0.582 ( 0.559 ) {:[-0.582],[(0.559)]:}\begin{gathered} -0.582 \\ (0.559) \end{gathered} 0.062 ( 0.304 ) 0.062 ( 0.304 ) {:[-0.062],[(0.304)]:}\begin{array}{r} -0.062 \\ (0.304) \end{array} 0.579 ( 0.393 ) 0.579 ( 0.393 ) {:[-0.579],[(0.393)]:}\begin{gathered} -0.579 \\ (0.393) \end{gathered} 0.438 ( 0.507 ) 0.438 ( 0.507 ) {:[-0.438],[(0.507)]:}\begin{gathered} -0.438 \\ (0.507) \end{gathered} 0.235 ( 0.317 ) 0.235 ( 0.317 ) {:[-0.235],[(0.317)]:}\begin{array}{r} -0.235 \\ (0.317) \end{array} 0.571 ( 0.401 ) 0.571 ( 0.401 ) {:[-0.571],[(0.401)]:}\begin{array}{r} -0.571 \\ (0.401) \end{array} 0.577 ( 0.522 ) 0.577 ( 0.522 ) {:[-0.577],[(0.522)]:}\begin{gathered} -0.577 \\ (0.522) \end{gathered}
L N ( S ) L N ( S ) LN(S)L N(S) - - - 0.251 ( 0.052 ) 0.251 ( 0.052 ) {:[-0.251],[(0.052)]:}\begin{array}{r} -0.251 \\ (0.052) \end{array} 0.239 ( 0.062 ) 0.239 ( 0.062 ) {:[-0.239],[(0.062)]:}\begin{array}{r} -0.239 \\ (0.062) \end{array} 0.890 ( 0.147 ) 0.890 ( 0.147 ) {:[-0.890],[(0.147)]:}\begin{gathered} -0.890 \\ (0.147) \end{gathered} 0.329 ( 0.053 ) 0.329 ( 0.053 ) {:[-0.329],[(0.053)]:}\begin{gathered} -0.329 \\ (0.053) \end{gathered} 0.260 ( 0.063 ) 0.260 ( 0.063 ) {:[-0.260],[(0.063)]:}\begin{gathered} -0.260 \\ (0.063) \end{gathered} 0.896 ( 0.152 ) 0.896 ( 0.152 ) {:[-0.896],[(0.152)]:}\begin{array}{r} -0.896 \\ (0.152) \end{array}
( L N ( S ) ) 2 ( L N ( S ) ) 2 (LN(S))^(2)(L N(S))^{2} - - - 0.015 ( 0.004 ) 0.015 ( 0.004 ) {:[0.015],[(0.004)]:}\begin{gathered} 0.015 \\ (0.004) \end{gathered} 0.010 ( 0.005 ) 0.010 ( 0.005 ) {:[0.010],[(0.005)]:}\begin{gathered} 0.010 \\ (0.005) \end{gathered} 0.073 ( 0.012 ) 0.073 ( 0.012 ) {:[0.073],[(0.012)]:}\begin{gathered} 0.073 \\ (0.012) \end{gathered} 0.021 ( 0.004 ) 0.021 ( 0.004 ) {:[0.021],[(0.004)]:}\begin{gathered} 0.021 \\ (0.004) \end{gathered} 0.012 ( 0.005 ) 0.012 ( 0.005 ) {:[0.012],[(0.005)]:}\begin{gathered} 0.012 \\ (0.005) \end{gathered} 0.075 ( 0.012 ) 0.075 ( 0.012 ) {:[0.075],[(0.012)]:}\begin{gathered} 0.075 \\ (0.012) \end{gathered}
K/S - - - 0.621 ( 0.152 ) 0.621 ( 0.152 ) {:[0.621],[(0.152)]:}\begin{gathered} 0.621 \\ (0.152) \end{gathered} 0.277 ( 0.192 ) 0.277 ( 0.192 ) {:[0.277],[(0.192)]:}\begin{gathered} 0.277 \\ (0.192) \end{gathered} 0.482 ( 0.289 ) 0.482 ( 0.289 ) {:[-0.482],[(0.289)]:}\begin{gathered} -0.482 \\ (0.289) \end{gathered} 0.469 ( 0.155 ) 0.469 ( 0.155 ) {:[0.469],[(0.155)]:}\begin{gathered} 0.469 \\ (0.155) \end{gathered} 0.342 ( 0.208 ) 0.342 ( 0.208 ) {:[0.342],[(0.208)]:}\begin{gathered} 0.342 \\ (0.208) \end{gathered} 0.504 ( 0.303 ) 0.504 ( 0.303 ) {:[-0.504],[(0.303)]:}\begin{gathered} -0.504 \\ (0.303) \end{gathered}
( K / S ) 2 ( K / S ) 2 (K//S)^(2)(K / S)^{2} - - - 0.391 ( 0.084 ) 0.391 ( 0.084 ) {:[-0.391],[(0.084)]:}\begin{array}{r} -0.391 \\ (0.084) \end{array} 0.420 ( 0.120 ) 0.420 ( 0.120 ) {:[-0.420],[(0.120)]:}\begin{array}{r} -0.420 \\ (0.120) \end{array} 0.040 ( 0.123 ) 0.040 ( 0.123 ) {:[0.040],[(0.123)]:}\begin{gathered} 0.040 \\ (0.123) \end{gathered} 0.403 ( 0.087 ) 0.403 ( 0.087 ) {:[-0.403],[(0.087)]:}\begin{gathered} -0.403 \\ (0.087) \end{gathered} 0.454 ( 0.131 ) 0.454 ( 0.131 ) {:[-0.454],[(0.131)]:}\begin{array}{r} -0.454 \\ (0.131) \end{array} 0.048 ( 0.124 ) 0.048 ( 0.124 ) {:[0.048],[(0.124)]:}\begin{gathered} 0.048 \\ (0.124) \end{gathered}
SIGMA - - - 5.06 ( 1.25 ) 5.06 ( 1.25 ) {:[-5.06],[(1.25)]:}\begin{array}{r} -5.06 \\ (1.25) \end{array} 4.82 ( 1.26 ) 4.82 ( 1.26 ) {:[-4.82],[(1.26)]:}\begin{array}{r} -4.82 \\ (1.26) \end{array} 4.26 ( 1.28 ) 4.26 ( 1.28 ) {:[-4.26],[(1.28)]:}\begin{array}{r} -4.26 \\ (1.28) \end{array} 4.47 ( 1.25 ) 4.47 ( 1.25 ) {:[-4.47],[(1.25)]:}\begin{array}{r} -4.47 \\ (1.25) \end{array} 5.40 ( 1.27 ) 5.40 ( 1.27 ) {:[-5.40],[(1.27)]:}\begin{array}{r} -5.40 \\ (1.27) \end{array} 4.62 ( 1.29 ) 4.62 ( 1.29 ) {:[-4.62],[(1.29)]:}\begin{array}{r} -4.62 \\ (1.29) \end{array}
Variable Pooled SIC3 effects Firm effects Pooled SIC3 effects Firm effects Pooled SIC3 effects Firm effects m "0.539 (0.219)" "1.25 (0.338)" "0.573 (0.402)" "-0.460 (0.218)" "-0.031 (0.277)" "0.125 (0.395)" "-0.395 (0.234)" "-0.061 (0.281)" "0.293 (0.392)" m^(2) "-1.123 (0.317)" "-1.649 (0.457)" "-0.582 (0.559)" "-0.062 (0.304)" "-0.579 (0.393)" "-0.438 (0.507)" "-0.235 (0.317)" "-0.571 (0.401)" "-0.577 (0.522)" LN(S) - - - "-0.251 (0.052)" "-0.239 (0.062)" "-0.890 (0.147)" "-0.329 (0.053)" "-0.260 (0.063)" "-0.896 (0.152)" (LN(S))^(2) - - - "0.015 (0.004)" "0.010 (0.005)" "0.073 (0.012)" "0.021 (0.004)" "0.012 (0.005)" "0.075 (0.012)" K/S - - - "0.621 (0.152)" "0.277 (0.192)" "-0.482 (0.289)" "0.469 (0.155)" "0.342 (0.208)" "-0.504 (0.303)" (K//S)^(2) - - - "-0.391 (0.084)" "-0.420 (0.120)" "0.040 (0.123)" "-0.403 (0.087)" "-0.454 (0.131)" "0.048 (0.124)" SIGMA - - - "-5.06 (1.25)" "-4.82 (1.26)" "-4.26 (1.28)" "-4.47 (1.25)" "-5.40 (1.27)" "-4.62 (1.29)"| Variable | Pooled | SIC3 effects | Firm effects | Pooled | SIC3 effects | Firm effects | Pooled | SIC3 effects | Firm effects | | :---: | :---: | :---: | :---: | :---: | :---: | :---: | :---: | :---: | :---: | | $m$ | $\begin{gathered} 0.539 \\ (0.219) \end{gathered}$ | $\begin{aligned} & 1.25 \\ & (0.338) \end{aligned}$ | $\begin{gathered} 0.573 \\ (0.402) \end{gathered}$ | $\begin{gathered} -0.460 \\ (0.218) \end{gathered}$ | $\begin{gathered} -0.031 \\ (0.277) \end{gathered}$ | $\begin{gathered} 0.125 \\ (0.395) \end{gathered}$ | $\begin{gathered} -0.395 \\ (0.234) \end{gathered}$ | $\begin{array}{r} -0.061 \\ (0.281) \end{array}$ | $\begin{gathered} 0.293 \\ (0.392) \end{gathered}$ | | $m^{2}$ | $\begin{gathered} -1.123 \\ (0.317) \end{gathered}$ | $\begin{gathered} -1.649 \\ (0.457) \end{gathered}$ | $\begin{gathered} -0.582 \\ (0.559) \end{gathered}$ | $\begin{array}{r} -0.062 \\ (0.304) \end{array}$ | $\begin{gathered} -0.579 \\ (0.393) \end{gathered}$ | $\begin{gathered} -0.438 \\ (0.507) \end{gathered}$ | $\begin{array}{r} -0.235 \\ (0.317) \end{array}$ | $\begin{array}{r} -0.571 \\ (0.401) \end{array}$ | $\begin{gathered} -0.577 \\ (0.522) \end{gathered}$ | | $L N(S)$ | - | - | - | $\begin{array}{r} -0.251 \\ (0.052) \end{array}$ | $\begin{array}{r} -0.239 \\ (0.062) \end{array}$ | $\begin{gathered} -0.890 \\ (0.147) \end{gathered}$ | $\begin{gathered} -0.329 \\ (0.053) \end{gathered}$ | $\begin{gathered} -0.260 \\ (0.063) \end{gathered}$ | $\begin{array}{r} -0.896 \\ (0.152) \end{array}$ | | $(L N(S))^{2}$ | - | - | - | $\begin{gathered} 0.015 \\ (0.004) \end{gathered}$ | $\begin{gathered} 0.010 \\ (0.005) \end{gathered}$ | $\begin{gathered} 0.073 \\ (0.012) \end{gathered}$ | $\begin{gathered} 0.021 \\ (0.004) \end{gathered}$ | $\begin{gathered} 0.012 \\ (0.005) \end{gathered}$ | $\begin{gathered} 0.075 \\ (0.012) \end{gathered}$ | | K/S | - | - | - | $\begin{gathered} 0.621 \\ (0.152) \end{gathered}$ | $\begin{gathered} 0.277 \\ (0.192) \end{gathered}$ | $\begin{gathered} -0.482 \\ (0.289) \end{gathered}$ | $\begin{gathered} 0.469 \\ (0.155) \end{gathered}$ | $\begin{gathered} 0.342 \\ (0.208) \end{gathered}$ | $\begin{gathered} -0.504 \\ (0.303) \end{gathered}$ | | $(K / S)^{2}$ | - | - | - | $\begin{array}{r} -0.391 \\ (0.084) \end{array}$ | $\begin{array}{r} -0.420 \\ (0.120) \end{array}$ | $\begin{gathered} 0.040 \\ (0.123) \end{gathered}$ | $\begin{gathered} -0.403 \\ (0.087) \end{gathered}$ | $\begin{array}{r} -0.454 \\ (0.131) \end{array}$ | $\begin{gathered} 0.048 \\ (0.124) \end{gathered}$ | | SIGMA | - | - | - | $\begin{array}{r} -5.06 \\ (1.25) \end{array}$ | $\begin{array}{r} -4.82 \\ (1.26) \end{array}$ | $\begin{array}{r} -4.26 \\ (1.28) \end{array}$ | $\begin{array}{r} -4.47 \\ (1.25) \end{array}$ | $\begin{array}{r} -5.40 \\ (1.27) \end{array}$ | $\begin{array}{r} -4.62 \\ (1.29) \end{array}$ |
1 1 1 1 1
E y m y 0 . E y m y 0 . {:[E_(y)],[m^(y)],[0.]:}\begin{aligned} & \underset{y}{E} \\ & \stackrel{y}{m} \\ & 0 . \end{aligned} 1 1 1 1 1
1 1 1 1 1
± 0 0 0 0 ± 0 0 0 0 [+-0],[0],[0],[0]\begin{array}{ll} \pm 0 \\ 0 \\ 0 \\ 0 \\ \hline \end{array}
0 σ σ 0 σ σ {:[0∼],[sigma _(sigma)]:}\begin{array}{ll} \underset{\sim}{0} \\ \underset{\sigma}{\sigma} \end{array} π 6 π 6 π 6 π 6 (pi)/(6)(pi)/(6)\frac{\pi}{6} \frac{\pi}{6}
y ^ N ^ y ^ N ^ widehat(y)_({: hat(N):})\underset{\substack{\hat{N}}}{\widehat{y}}
± 0 0 0 ± 0 0 0 {:[+-],[(0)/(0)],[0]:}\begin{aligned} & \pm \\ & \frac{0}{0} \\ & 0 \end{aligned} N π π N π π N^(pi)vdots^(pi)\stackrel{\pi}{N} \stackrel{\pi}{\vdots}
n 0 0 0 0 ˙ 0 0 n 0 0 0 0 ˙ 0 0 {:[n^(0)],[0^(0)0^(˙)],[0^(0)]:}\begin{aligned} & \stackrel{0}{n} \\ & \stackrel{0}{0} \dot{0} \\ & \stackrel{0}{0} \end{aligned} 9 9 9^(⇀)^(⇀^(⏜))\stackrel{\overparen{\rightharpoonup}}{\stackrel{\rightharpoonup}{9}}
1 1 1 1
1 1 1
1 1 1 1
1 1 1
1 1 1 1
1 1 1
J U W N  J   U   W   N  {:[" J "],[" U "],[" W "],[" N "]:}\begin{aligned} & \text { J } \\ & \text { U } \\ & \text { W } \\ & \text { N } \end{aligned} \lesssim
S 0 2 S 0 2 {:[S],[0],[2]:}\begin{aligned} & \mathbb{S} \\ & 0 \\ & 2 \end{aligned} (AA)/(≷)\frac{\forall}{\gtrless} Δ Δ ⇆^(Delta)\stackrel{\Delta}{\leftrightarrows}
1 1 1 1 1 https://cdn.mathpix.com/cropped/2025_01_10_cc71e97ea52bbc49cb95g-25.jpg?height=78&width=177&top_left_y=285&top_left_x=723 https://cdn.mathpix.com/cropped/2025_01_10_cc71e97ea52bbc49cb95g-25.jpg?height=88&width=81&top_left_y=516&top_left_x=93 "E_(y) m^(y) 0." 1 1 1 1 1 https://cdn.mathpix.com/cropped/2025_01_10_cc71e97ea52bbc49cb95g-25.jpg?height=115&width=177&top_left_y=489&top_left_x=723 https://cdn.mathpix.com/cropped/2025_01_10_cc71e97ea52bbc49cb95g-25.jpg?height=90&width=81&top_left_y=655&top_left_x=93 https://cdn.mathpix.com/cropped/2025_01_10_cc71e97ea52bbc49cb95g-25.jpg?height=92&width=81&top_left_y=653&top_left_x=186 1 1 1 1 1 https://cdn.mathpix.com/cropped/2025_01_10_cc71e97ea52bbc49cb95g-25.jpg?height=118&width=179&top_left_y=627&top_left_x=723 "+-0 0 0 0" https://cdn.mathpix.com/cropped/2025_01_10_cc71e97ea52bbc49cb95g-25.jpg?height=94&width=81&top_left_y=827&top_left_x=186 "0∼ sigma _(sigma)" (pi)/(6)(pi)/(6) https://cdn.mathpix.com/cropped/2025_01_10_cc71e97ea52bbc49cb95g-25.jpg?height=90&width=82&top_left_y=827&top_left_x=545 https://cdn.mathpix.com/cropped/2025_01_10_cc71e97ea52bbc49cb95g-25.jpg?height=78&width=179&top_left_y=837&top_left_x=723 https://cdn.mathpix.com/cropped/2025_01_10_cc71e97ea52bbc49cb95g-25.jpg?height=88&width=81&top_left_y=1070&top_left_x=93 widehat(y)_({: hat(N):}) https://cdn.mathpix.com/cropped/2025_01_10_cc71e97ea52bbc49cb95g-25.jpg?height=114&width=82&top_left_y=1044&top_left_x=363 "+- (0)/(0) 0" N^(pi)vdots^(pi) https://cdn.mathpix.com/cropped/2025_01_10_cc71e97ea52bbc49cb95g-25.jpg?height=114&width=179&top_left_y=1044&top_left_x=723 https://cdn.mathpix.com/cropped/2025_01_10_cc71e97ea52bbc49cb95g-25.jpg?height=88&width=81&top_left_y=1207&top_left_x=186 "n^(0) 0^(0)0^(˙) 0^(0)" 9^(⇀)^(⇀^(⏜)) https://cdn.mathpix.com/cropped/2025_01_10_cc71e97ea52bbc49cb95g-25.jpg?height=106&width=179&top_left_y=1191&top_left_x=723 1 1 1 1 https://cdn.mathpix.com/cropped/2025_01_10_cc71e97ea52bbc49cb95g-25.jpg?height=86&width=179&top_left_y=1387&top_left_x=723 1 1 1 1 1 1 1 https://cdn.mathpix.com/cropped/2025_01_10_cc71e97ea52bbc49cb95g-25.jpg?height=85&width=179&top_left_y=1596&top_left_x=723 1 1 1 1 1 1 1 1 1 1 " J U W N " ≲ https://cdn.mathpix.com/cropped/2025_01_10_cc71e97ea52bbc49cb95g-25.jpg?height=123&width=48&top_left_y=1891&top_left_x=272 "S 0 2" (AA)/(≷) ⇆^(Delta) https://cdn.mathpix.com/cropped/2025_01_10_cc71e97ea52bbc49cb95g-25.jpg?height=99&width=179&top_left_y=1915&top_left_x=723| $$ | $$ | 1 | 1 | 1 | 1 | 1 | ![](https://cdn.mathpix.com/cropped/2025_01_10_cc71e97ea52bbc49cb95g-25.jpg?height=78&width=177&top_left_y=285&top_left_x=723) | | :---: | :---: | :---: | :---: | :---: | :---: | :---: | :---: | | ![](https://cdn.mathpix.com/cropped/2025_01_10_cc71e97ea52bbc49cb95g-25.jpg?height=88&width=81&top_left_y=516&top_left_x=93) | $\begin{aligned} & \underset{y}{E} \\ & \stackrel{y}{m} \\ & 0 . \end{aligned}$ | 1 | 1 | 1 | 1 | 1 | ![](https://cdn.mathpix.com/cropped/2025_01_10_cc71e97ea52bbc49cb95g-25.jpg?height=115&width=177&top_left_y=489&top_left_x=723) | | ![](https://cdn.mathpix.com/cropped/2025_01_10_cc71e97ea52bbc49cb95g-25.jpg?height=90&width=81&top_left_y=655&top_left_x=93) | ![](https://cdn.mathpix.com/cropped/2025_01_10_cc71e97ea52bbc49cb95g-25.jpg?height=92&width=81&top_left_y=653&top_left_x=186) | 1 | 1 | 1 | 1 | 1 | ![](https://cdn.mathpix.com/cropped/2025_01_10_cc71e97ea52bbc49cb95g-25.jpg?height=118&width=179&top_left_y=627&top_left_x=723) | | $\begin{array}{ll} \pm 0 \\ 0 \\ 0 \\ 0 \\ \hline \end{array}$ | ![](https://cdn.mathpix.com/cropped/2025_01_10_cc71e97ea52bbc49cb95g-25.jpg?height=94&width=81&top_left_y=827&top_left_x=186) | $\begin{array}{ll} \underset{\sim}{0} \\ \underset{\sigma}{\sigma} \end{array}$ | $\frac{\pi}{6} \frac{\pi}{6}$ | $$ | ![](https://cdn.mathpix.com/cropped/2025_01_10_cc71e97ea52bbc49cb95g-25.jpg?height=90&width=82&top_left_y=827&top_left_x=545) | $$ | ![](https://cdn.mathpix.com/cropped/2025_01_10_cc71e97ea52bbc49cb95g-25.jpg?height=78&width=179&top_left_y=837&top_left_x=723) | | ![](https://cdn.mathpix.com/cropped/2025_01_10_cc71e97ea52bbc49cb95g-25.jpg?height=88&width=81&top_left_y=1070&top_left_x=93) | $$ | $\underset{\substack{\hat{N}}}{\widehat{y}}$ | ![](https://cdn.mathpix.com/cropped/2025_01_10_cc71e97ea52bbc49cb95g-25.jpg?height=114&width=82&top_left_y=1044&top_left_x=363) | $$ | $\begin{aligned} & \pm \\ & \frac{0}{0} \\ & 0 \end{aligned}$ | $\stackrel{\pi}{N} \stackrel{\pi}{\vdots}$ | ![](https://cdn.mathpix.com/cropped/2025_01_10_cc71e97ea52bbc49cb95g-25.jpg?height=114&width=179&top_left_y=1044&top_left_x=723) | | $$ | ![](https://cdn.mathpix.com/cropped/2025_01_10_cc71e97ea52bbc49cb95g-25.jpg?height=88&width=81&top_left_y=1207&top_left_x=186) | $$ | $\begin{aligned} & \stackrel{0}{n} \\ & \stackrel{0}{0} \dot{0} \\ & \stackrel{0}{0} \end{aligned}$ | $$ | $$ | $\stackrel{\overparen{\rightharpoonup}}{\stackrel{\rightharpoonup}{9}}$ | ![](https://cdn.mathpix.com/cropped/2025_01_10_cc71e97ea52bbc49cb95g-25.jpg?height=106&width=179&top_left_y=1191&top_left_x=723) | | | | | 1 | 1 | 1 | 1 | ![](https://cdn.mathpix.com/cropped/2025_01_10_cc71e97ea52bbc49cb95g-25.jpg?height=86&width=179&top_left_y=1387&top_left_x=723) | | 1 | 1 | 1 | | | | | | | | | | 1 | 1 | 1 | 1 | ![](https://cdn.mathpix.com/cropped/2025_01_10_cc71e97ea52bbc49cb95g-25.jpg?height=85&width=179&top_left_y=1596&top_left_x=723) | | 1 | 1 | 1 | | | | | | | | | | 1 | 1 | 1 | 1 | $$ | | 1 | 1 | 1 | | | | | | | $\begin{aligned} & \text { J } \\ & \text { U } \\ & \text { W } \\ & \text { N } \end{aligned}$ | $\lesssim$ | ![](https://cdn.mathpix.com/cropped/2025_01_10_cc71e97ea52bbc49cb95g-25.jpg?height=123&width=48&top_left_y=1891&top_left_x=272) | $\begin{aligned} & \mathbb{S} \\ & 0 \\ & 2 \end{aligned}$ | $\frac{\forall}{\gtrless}$ | $$ | $\stackrel{\Delta}{\leftrightarrows}$ | ![](https://cdn.mathpix.com/cropped/2025_01_10_cc71e97ea52bbc49cb95g-25.jpg?height=99&width=179&top_left_y=1915&top_left_x=723) |
Table 5. Continued.  表 5. 续。
(B) Determinants of firm value (Tobin’s Q Q QQ ), spline specifications
(B) 企业价值的决定因素 (Tobin’s Q Q QQ ), 样条规格

The specifications reported in this table all model firm value, Q Q QQ, a linear function of the explanatory variables indicated below. The influence of m m mm enters as a spline function. Intercept terms and year dummies are included for all regressions, but not reported. Fixed effects at the industry or firm level are included where indicated, but not reported. Variable definitions for the acronyms are given in Table 3.
本表中报告的规格均为模型公司价值 Q Q QQ 的线性函数,解释变量如下所示。 m m mm 的影响以样条函数的形式进入。所有回归均包含截距项和年份虚拟变量,但未报告。行业或公司层面的固定效应在指示的地方包含,但未报告。缩略词的变量定义见表 3。
Variable  变量 Pooled  合并的 SIC3 effects  SIC3 效果 Firm effects  公司效应 Pooled  合并的 SIC3 effects  SIC3 效果 Firm effects  公司效应 Pooled  合并的 SIC3 effects  SIC3 效果 Firm effects  公司效应
m 1 m 1 m1m 1 4.678 ( 1.127 ) 4.678 ( 1.127 ) {:[4.678],[(1.127)]:}\begin{gathered} 4.678 \\ (1.127) \end{gathered} 7.379 ( 1.727 ) 7.379 ( 1.727 ) {:[7.379],[(1.727)]:}\begin{gathered} 7.379 \\ (1.727) \end{gathered} 0.772 ( 1.820 ) 0.772 ( 1.820 ) {:[0.772],[(1.820)]:}\begin{gathered} 0.772 \\ (1.820) \end{gathered} 2.88 ( 1.34 ) 2.88 ( 1.34 ) {:[2.88],[(1.34)]:}\begin{gathered} 2.88 \\ (1.34) \end{gathered} 3.75 ( 1.53 ) 3.75 ( 1.53 ) {:[3.75],[(1.53)]:}\begin{gathered} 3.75 \\ (1.53) \end{gathered} 1.62 ( 1.73 ) 1.62 ( 1.73 ) {:[1.62],[(1.73)]:}\begin{gathered} 1.62 \\ (1.73) \end{gathered} 3.691 ( 1.334 ) 3.691 ( 1.334 ) {:[3.691],[(1.334)]:}\begin{gathered} 3.691 \\ (1.334) \end{gathered} 3.724 ( 1.555 ) 3.724 ( 1.555 ) {:[3.724],[(1.555)]:}\begin{gathered} 3.724 \\ (1.555) \end{gathered} 2.097 ( 1.736 ) 2.097 ( 1.736 ) {:[2.097],[(1.736)]:}\begin{gathered} 2.097 \\ (1.736) \end{gathered}
m 2 m 2 m2m 2 0.070 ( 0.312 ) 0.070 ( 0.312 ) {:[-0.070],[(0.312)]:}\begin{gathered} -0.070 \\ (0.312) \end{gathered} 0.428 ( 0.365 ) 0.428 ( 0.365 ) {:[0.428],[(0.365)]:}\begin{gathered} 0.428 \\ (0.365) \end{gathered} 0.122 ( 0.395 ) 0.122 ( 0.395 ) {:[0.122],[(0.395)]:}\begin{gathered} 0.122 \\ (0.395) \end{gathered} 0.587 ( 0.295 ) 0.587 ( 0.295 ) {:[-0.587],[(0.295)]:}\begin{gathered} -0.587 \\ (0.295) \end{gathered} 0.150 ( 0.338 ) 0.150 ( 0.338 ) {:[-0.150],[(0.338)]:}\begin{gathered} -0.150 \\ (0.338) \end{gathered} 0.214 ( 0.387 ) 0.214 ( 0.387 ) {:[-0.214],[(0.387)]:}\begin{gathered} -0.214 \\ (0.387) \end{gathered} 0.689 ( 0.314 ) 0.689 ( 0.314 ) {:[-0.689],[(0.314)]:}\begin{gathered} -0.689 \\ (0.314) \end{gathered} 0.201 ( 0.342 ) 0.201 ( 0.342 ) {:[-0.201],[(0.342)]:}\begin{gathered} -0.201 \\ (0.342) \end{gathered} 0.167 ( 0.386 ) 0.167 ( 0.386 ) {:[-0.167],[(0.386)]:}\begin{array}{r} -0.167 \\ (0.386) \end{array}
m3 0.567 ( 0.167 ) 0.567 ( 0.167 ) {:[-0.567],[(0.167)]:}\begin{gathered} -0.567 \\ (0.167) \end{gathered} 0.546 ( 0.201 ) 0.546 ( 0.201 ) {:[-0.546],[(0.201)]:}\begin{array}{r} -0.546 \\ (0.201) \end{array} 0.171 ( 0.257 ) 0.171 ( 0.257 ) {:[0.171],[(0.257)]:}\begin{gathered} 0.171 \\ (0.257) \end{gathered} 0.446 ( 0.168 ) 0.446 ( 0.168 ) {:[-0.446],[(0.168)]:}\begin{array}{r} -0.446 \\ (0.168) \end{array} 0.703 ( 0.218 ) 0.703 ( 0.218 ) {:[-0.703],[(0.218)]:}\begin{gathered} -0.703 \\ (0.218) \end{gathered} 0.225 ( 0.247 ) 0.225 ( 0.247 ) {:[-0.225],[(0.247)]:}\begin{gathered} -0.225 \\ (0.247) \end{gathered} 0.636 ( 0.173 ) 0.636 ( 0.173 ) {:[-0.636],[(0.173)]:}\begin{array}{r} -0.636 \\ (0.173) \end{array} 0.712 ( 0.228 ) 0.712 ( 0.228 ) {:[-0.712],[(0.228)]:}\begin{gathered} -0.712 \\ (0.228) \end{gathered} 0.152 ( 0.250 ) 0.152 ( 0.250 ) {:[-0.152],[(0.250)]:}\begin{gathered} -0.152 \\ (0.250) \end{gathered}
L N ( S ) L N ( S ) LN(S)L N(S) - - - 0.263 ( 0.054 ) 0.263 ( 0.054 ) {:[-0.263],[(0.054)]:}\begin{array}{r} -0.263 \\ (0.054) \end{array} 0.247 ( 0.062 ) 0.247 ( 0.062 ) {:[-0.247],[(0.062)]:}\begin{gathered} -0.247 \\ (0.062) \end{gathered} 0.896 ( 0.147 ) 0.896 ( 0.147 ) {:[-0.896],[(0.147)]:}\begin{gathered} -0.896 \\ (0.147) \end{gathered} 0.345 ( 0.055 ) 0.345 ( 0.055 ) {:[-0.345],[(0.055)]:}\begin{gathered} -0.345 \\ (0.055) \end{gathered} 0.268 ( 0.064 ) 0.268 ( 0.064 ) {:[-0.268],[(0.064)]:}\begin{gathered} -0.268 \\ (0.064) \end{gathered} 0.903 ( 0.152 ) 0.903 ( 0.152 ) {:[-0.903],[(0.152)]:}\begin{gathered} -0.903 \\ (0.152) \end{gathered}
( L N ( S ) ) 2 ( L N ( S ) ) 2 (LN(S))^(2)(L N(S))^{2} - - - 0.017 ( 0.005 ) 0.017 ( 0.005 ) {:[0.017],[(0.005)]:}\begin{gathered} 0.017 \\ (0.005) \end{gathered} 0.012 ( 0.005 ) 0.012 ( 0.005 ) {:[0.012],[(0.005)]:}\begin{gathered} 0.012 \\ (0.005) \end{gathered} 0.074 ( 0.012 ) 0.074 ( 0.012 ) {:[0.074],[(0.012)]:}\begin{gathered} 0.074 \\ (0.012) \end{gathered} 0.023 ( 0.005 ) 0.023 ( 0.005 ) {:[0.023],[(0.005)]:}\begin{gathered} 0.023 \\ (0.005) \end{gathered} 0.013 ( 0.005 ) 0.013 ( 0.005 ) {:[0.013],[(0.005)]:}\begin{gathered} 0.013 \\ (0.005) \end{gathered} 0.075 ( 0.012 ) 0.075 ( 0.012 ) {:[0.075],[(0.012)]:}\begin{gathered} 0.075 \\ (0.012) \end{gathered}
K/S - - - 0.648 ( 0.154 ) 0.648 ( 0.154 ) {:[0.648],[(0.154)]:}\begin{gathered} 0.648 \\ (0.154) \end{gathered} 0.297 ( 0.193 ) 0.297 ( 0.193 ) {:[0.297],[(0.193)]:}\begin{gathered} 0.297 \\ (0.193) \end{gathered} 0.475 ( 0.289 ) 0.475 ( 0.289 ) {:[-0.475],[(0.289)]:}\begin{gathered} -0.475 \\ (0.289) \end{gathered} 0.492 ( 0.156 ) 0.492 ( 0.156 ) {:[0.492],[(0.156)]:}\begin{gathered} 0.492 \\ (0.156) \end{gathered} 0.359 ( 0.208 ) 0.359 ( 0.208 ) {:[0.359],[(0.208)]:}\begin{gathered} 0.359 \\ (0.208) \end{gathered} 0.493 ( 0.304 ) 0.493 ( 0.304 ) {:[-0.493],[(0.304)]:}\begin{gathered} -0.493 \\ (0.304) \end{gathered}
( K / S ) 2 ( K / S ) 2 (K//S)^(2)(K / S)^{2} - - - 0.395 ( 0.084 ) 0.395 ( 0.084 ) {:[-0.395],[(0.084)]:}\begin{gathered} -0.395 \\ (0.084) \end{gathered} 0.427 ( 0.120 ) 0.427 ( 0.120 ) {:[-0.427],[(0.120)]:}\begin{gathered} -0.427 \\ (0.120) \end{gathered} 0.036 ( 0.123 ) 0.036 ( 0.123 ) {:[0.036],[(0.123)]:}\begin{gathered} 0.036 \\ (0.123) \end{gathered} 0.404 ( 0.087 ) 0.404 ( 0.087 ) {:[-0.404],[(0.087)]:}\begin{gathered} -0.404 \\ (0.087) \end{gathered} 0.459 ( 0.130 ) 0.459 ( 0.130 ) {:[-0.459],[(0.130)]:}\begin{gathered} -0.459 \\ (0.130) \end{gathered} 0.042 ( 0.125 ) 0.042 ( 0.125 ) {:[0.042],[(0.125)]:}\begin{gathered} 0.042 \\ (0.125) \end{gathered}
SIGMA - - - 4.83 ( 1.25 ) 4.83 ( 1.25 ) {:[-4.83],[(1.25)]:}\begin{array}{r} -4.83 \\ (1.25) \end{array} 4.66 ( 1.24 ) 4.66 ( 1.24 ) {:[-4.66],[(1.24)]:}\begin{array}{r} -4.66 \\ (1.24) \end{array} 4.24 ( 1.27 ) 4.24 ( 1.27 ) {:[-4.24],[(1.27)]:}\begin{array}{r} -4.24 \\ (1.27) \end{array} 4.23 ( 1.25 ) 4.23 ( 1.25 ) {:[-4.23],[(1.25)]:}\begin{gathered} -4.23 \\ (1.25) \end{gathered} 5.24 ( 1.25 ) 5.24 ( 1.25 ) {:[-5.24],[(1.25)]:}\begin{array}{r} -5.24 \\ (1.25) \end{array} 4.61 ( 1.28 ) 4.61 ( 1.28 ) {:[-4.61],[(1.28)]:}\begin{gathered} -4.61 \\ (1.28) \end{gathered}
Variable Pooled SIC3 effects Firm effects Pooled SIC3 effects Firm effects Pooled SIC3 effects Firm effects m1 "4.678 (1.127)" "7.379 (1.727)" "0.772 (1.820)" "2.88 (1.34)" "3.75 (1.53)" "1.62 (1.73)" "3.691 (1.334)" "3.724 (1.555)" "2.097 (1.736)" m2 "-0.070 (0.312)" "0.428 (0.365)" "0.122 (0.395)" "-0.587 (0.295)" "-0.150 (0.338)" "-0.214 (0.387)" "-0.689 (0.314)" "-0.201 (0.342)" "-0.167 (0.386)" m3 "-0.567 (0.167)" "-0.546 (0.201)" "0.171 (0.257)" "-0.446 (0.168)" "-0.703 (0.218)" "-0.225 (0.247)" "-0.636 (0.173)" "-0.712 (0.228)" "-0.152 (0.250)" LN(S) - - - "-0.263 (0.054)" "-0.247 (0.062)" "-0.896 (0.147)" "-0.345 (0.055)" "-0.268 (0.064)" "-0.903 (0.152)" (LN(S))^(2) - - - "0.017 (0.005)" "0.012 (0.005)" "0.074 (0.012)" "0.023 (0.005)" "0.013 (0.005)" "0.075 (0.012)" K/S - - - "0.648 (0.154)" "0.297 (0.193)" "-0.475 (0.289)" "0.492 (0.156)" "0.359 (0.208)" "-0.493 (0.304)" (K//S)^(2) - - - "-0.395 (0.084)" "-0.427 (0.120)" "0.036 (0.123)" "-0.404 (0.087)" "-0.459 (0.130)" "0.042 (0.125)" SIGMA - - - "-4.83 (1.25)" "-4.66 (1.24)" "-4.24 (1.27)" "-4.23 (1.25)" "-5.24 (1.25)" "-4.61 (1.28)"| Variable | Pooled | SIC3 effects | Firm effects | Pooled | SIC3 effects | Firm effects | Pooled | SIC3 effects | Firm effects | | :---: | :---: | :---: | :---: | :---: | :---: | :---: | :---: | :---: | :---: | | $m 1$ | $\begin{gathered} 4.678 \\ (1.127) \end{gathered}$ | $\begin{gathered} 7.379 \\ (1.727) \end{gathered}$ | $\begin{gathered} 0.772 \\ (1.820) \end{gathered}$ | $\begin{gathered} 2.88 \\ (1.34) \end{gathered}$ | $\begin{gathered} 3.75 \\ (1.53) \end{gathered}$ | $\begin{gathered} 1.62 \\ (1.73) \end{gathered}$ | $\begin{gathered} 3.691 \\ (1.334) \end{gathered}$ | $\begin{gathered} 3.724 \\ (1.555) \end{gathered}$ | $\begin{gathered} 2.097 \\ (1.736) \end{gathered}$ | | $m 2$ | $\begin{gathered} -0.070 \\ (0.312) \end{gathered}$ | $\begin{gathered} 0.428 \\ (0.365) \end{gathered}$ | $\begin{gathered} 0.122 \\ (0.395) \end{gathered}$ | $\begin{gathered} -0.587 \\ (0.295) \end{gathered}$ | $\begin{gathered} -0.150 \\ (0.338) \end{gathered}$ | $\begin{gathered} -0.214 \\ (0.387) \end{gathered}$ | $\begin{gathered} -0.689 \\ (0.314) \end{gathered}$ | $\begin{gathered} -0.201 \\ (0.342) \end{gathered}$ | $\begin{array}{r} -0.167 \\ (0.386) \end{array}$ | | m3 | $\begin{gathered} -0.567 \\ (0.167) \end{gathered}$ | $\begin{array}{r} -0.546 \\ (0.201) \end{array}$ | $\begin{gathered} 0.171 \\ (0.257) \end{gathered}$ | $\begin{array}{r} -0.446 \\ (0.168) \end{array}$ | $\begin{gathered} -0.703 \\ (0.218) \end{gathered}$ | $\begin{gathered} -0.225 \\ (0.247) \end{gathered}$ | $\begin{array}{r} -0.636 \\ (0.173) \end{array}$ | $\begin{gathered} -0.712 \\ (0.228) \end{gathered}$ | $\begin{gathered} -0.152 \\ (0.250) \end{gathered}$ | | $L N(S)$ | - | - | - | $\begin{array}{r} -0.263 \\ (0.054) \end{array}$ | $\begin{gathered} -0.247 \\ (0.062) \end{gathered}$ | $\begin{gathered} -0.896 \\ (0.147) \end{gathered}$ | $\begin{gathered} -0.345 \\ (0.055) \end{gathered}$ | $\begin{gathered} -0.268 \\ (0.064) \end{gathered}$ | $\begin{gathered} -0.903 \\ (0.152) \end{gathered}$ | | $(L N(S))^{2}$ | - | - | - | $\begin{gathered} 0.017 \\ (0.005) \end{gathered}$ | $\begin{gathered} 0.012 \\ (0.005) \end{gathered}$ | $\begin{gathered} 0.074 \\ (0.012) \end{gathered}$ | $\begin{gathered} 0.023 \\ (0.005) \end{gathered}$ | $\begin{gathered} 0.013 \\ (0.005) \end{gathered}$ | $\begin{gathered} 0.075 \\ (0.012) \end{gathered}$ | | K/S | - | - | - | $\begin{gathered} 0.648 \\ (0.154) \end{gathered}$ | $\begin{gathered} 0.297 \\ (0.193) \end{gathered}$ | $\begin{gathered} -0.475 \\ (0.289) \end{gathered}$ | $\begin{gathered} 0.492 \\ (0.156) \end{gathered}$ | $\begin{gathered} 0.359 \\ (0.208) \end{gathered}$ | $\begin{gathered} -0.493 \\ (0.304) \end{gathered}$ | | $(K / S)^{2}$ | - | - | - | $\begin{gathered} -0.395 \\ (0.084) \end{gathered}$ | $\begin{gathered} -0.427 \\ (0.120) \end{gathered}$ | $\begin{gathered} 0.036 \\ (0.123) \end{gathered}$ | $\begin{gathered} -0.404 \\ (0.087) \end{gathered}$ | $\begin{gathered} -0.459 \\ (0.130) \end{gathered}$ | $\begin{gathered} 0.042 \\ (0.125) \end{gathered}$ | | SIGMA | - | - | - | $\begin{array}{r} -4.83 \\ (1.25) \end{array}$ | $\begin{array}{r} -4.66 \\ (1.24) \end{array}$ | $\begin{array}{r} -4.24 \\ (1.27) \end{array}$ | $\begin{gathered} -4.23 \\ (1.25) \end{gathered}$ | $\begin{array}{r} -5.24 \\ (1.25) \end{array}$ | $\begin{gathered} -4.61 \\ (1.28) \end{gathered}$ |
SIGDUM - - - 0.243 ( 0.051 ) 0.243 ( 0.051 ) {:[0.243],[(0.051)]:}\begin{aligned} & 0.243 \\ & (0.051) \end{aligned} 0.219 ( 0.054 ) 0.219 ( 0.054 ) {:[0.219],[(0.054)]:}\begin{aligned} & 0.219 \\ & (0.054) \end{aligned} 0.044 ( 0.066 ) 0.044 ( 0.066 ) {:[-0.044],[(0.066)]:}\begin{array}{r} -0.044 \\ (0.066) \end{array} 0.242 ( 0.054 ) 0.242 ( 0.054 ) {:[0.242],[(0.054)]:}\begin{aligned} & 0.242 \\ & (0.054) \end{aligned} 0.254 ( 0.056 ) 0.254 ( 0.056 ) {:[0.254],[(0.056)]:}\begin{aligned} & 0.254 \\ & (0.056) \end{aligned} 0.035 ( 0.067 ) 0.035 ( 0.067 ) {:[-0.035],[(0.067)]:}\begin{array}{r} -0.035 \\ (0.067) \end{array}
Y / S Y / S Y//SY / S - - - 0.661 ( 0.279 ) 0.661 ( 0.279 ) {:[0.661],[(0.279)]:}\begin{aligned} & 0.661 \\ & (0.279) \end{aligned} 0.725 ( 0.303 ) 0.725 ( 0.303 ) {:[0.725],[(0.303)]:}\begin{aligned} & 0.725 \\ & (0.303) \end{aligned} 1.45 ( 0.269 ) 1.45 ( 0.269 ) {:[1.45],[(0.269)]:}\begin{aligned} & 1.45 \\ & (0.269) \end{aligned} 0.790 ( 0.292 ) 0.790 ( 0.292 ) {:[0.790],[(0.292)]:}\begin{aligned} & 0.790 \\ & (0.292) \end{aligned} 0.675 ( 0.315 ) 0.675 ( 0.315 ) {:[0.675],[(0.315)]:}\begin{aligned} & 0.675 \\ & (0.315) \end{aligned} 1.44 ( 0.264 ) 1.44 ( 0.264 ) {:[1.44],[(0.264)]:}\begin{aligned} & 1.44 \\ & (0.264) \end{aligned}
( R & D ) / K ( R & D ) / K (R&D)//K(R \& D) / K - - - 0.551 ( 0.225 ) 0.551 ( 0.225 ) {:[0.551],[(0.225)]:}\begin{aligned} & 0.551 \\ & (0.225) \end{aligned} 0.499 ( 0.268 ) 0.499 ( 0.268 ) {:[0.499],[(0.268)]:}\begin{aligned} & 0.499 \\ & (0.268) \end{aligned} 0.384 ( 0.409 ) 0.384 ( 0.409 ) {:[0.384],[(0.409)]:}\begin{gathered} 0.384 \\ (0.409) \end{gathered} - - -
RDUM - - - 0.158 ( 0.046 ) 0.158 ( 0.046 ) {:[0.158],[(0.046)]:}\begin{gathered} 0.158 \\ (0.046) \end{gathered} 0.273 ( 0.072 ) 0.273 ( 0.072 ) {:[0.273],[(0.072)]:}\begin{gathered} 0.273 \\ (0.072) \end{gathered} 0.202 ( 0.125 ) 0.202 ( 0.125 ) {:[0.202],[(0.125)]:}\begin{gathered} 0.202 \\ (0.125) \end{gathered} - - -
A/K - - - 0.046 ( 0.125 ) 0.046 ( 0.125 ) {:[-0.046],[(0.125)]:}\begin{aligned} & -0.046 \\ & (0.125) \end{aligned} 0.402 ( 0.205 ) 0.402 ( 0.205 ) {:[-0.402],[(0.205)]:}\begin{aligned} & -0.402 \\ & (0.205) \end{aligned} 0.083 ( 0.451 ) 0.083 ( 0.451 ) {:[0.083],[(0.451)]:}\begin{gathered} 0.083 \\ (0.451) \end{gathered} - - -
ADUM - - - 0.159 ( 0.041 ) 0.159 ( 0.041 ) {:[0.159],[(0.041)]:}\begin{aligned} & 0.159 \\ & (0.041) \end{aligned} 0.142 ( 0.055 ) 0.142 ( 0.055 ) {:[0.142],[(0.055)]:}\begin{aligned} & 0.142 \\ & (0.055) \end{aligned} 0.144 ( 0.095 ) 0.144 ( 0.095 ) {:[0.144],[(0.095)]:}\begin{gathered} 0.144 \\ (0.095) \end{gathered} - - -
I / K I / K I//KI / K - - - 0.776 ( 0.123 ) 0.776 ( 0.123 ) {:[0.776],[(0.123)]:}\begin{aligned} & 0.776 \\ & (0.123) \end{aligned} 0.713 ( 0.115 ) 0.713 ( 0.115 ) {:[0.713],[(0.115)]:}\begin{aligned} & 0.713 \\ & (0.115) \end{aligned} 0.339 ( 0.103 ) 0.339 ( 0.103 ) {:[0.339],[(0.103)]:}\begin{gathered} 0.339 \\ (0.103) \end{gathered} - - -
# Obs. 2630 2630 2630 2630 2630 2630 2630 2630 2630
Adj. R 2 R 2 R^(2)R^{2}  形容词 R 2 R 2 R^(2)R^{2} 0.016 0.135 0.584 0.131 0.215 0.630 0.075 0.181 0.626
p p pp-value - 0.004 0.126 - 0.001 0.037 - 0.002 0.018
Wald  瓦尔德 34.641 23.729 1.379 25.396 20.928 1.984 41.12 20.133 1.937
pwald - - 0.71 - - 0.576 - - 0.586
SIGDUM - - - "0.243 (0.051)" "0.219 (0.054)" "-0.044 (0.066)" "0.242 (0.054)" "0.254 (0.056)" "-0.035 (0.067)" Y//S - - - "0.661 (0.279)" "0.725 (0.303)" "1.45 (0.269)" "0.790 (0.292)" "0.675 (0.315)" "1.44 (0.264)" (R&D)//K - - - "0.551 (0.225)" "0.499 (0.268)" "0.384 (0.409)" - - - RDUM - - - "0.158 (0.046)" "0.273 (0.072)" "0.202 (0.125)" - - - A/K - - - "-0.046 (0.125)" "-0.402 (0.205)" "0.083 (0.451)" - - - ADUM - - - "0.159 (0.041)" "0.142 (0.055)" "0.144 (0.095)" - - - I//K - - - "0.776 (0.123)" "0.713 (0.115)" "0.339 (0.103)" - - - # Obs. 2630 2630 2630 2630 2630 2630 2630 2630 2630 Adj. R^(2) 0.016 0.135 0.584 0.131 0.215 0.630 0.075 0.181 0.626 p-value - 0.004 0.126 - 0.001 0.037 - 0.002 0.018 Wald 34.641 23.729 1.379 25.396 20.928 1.984 41.12 20.133 1.937 pwald - - 0.71 - - 0.576 - - 0.586| SIGDUM | - | | - | | - | | $\begin{aligned} & 0.243 \\ & (0.051) \end{aligned}$ | $\begin{aligned} & 0.219 \\ & (0.054) \end{aligned}$ | $\begin{array}{r} -0.044 \\ (0.066) \end{array}$ | $\begin{aligned} & 0.242 \\ & (0.054) \end{aligned}$ | $\begin{aligned} & 0.254 \\ & (0.056) \end{aligned}$ | $\begin{array}{r} -0.035 \\ (0.067) \end{array}$ | | :---: | :---: | :---: | :---: | :---: | :---: | :---: | :---: | :---: | :---: | :---: | :---: | :---: | | $Y / S$ | - | | - | | - | | $\begin{aligned} & 0.661 \\ & (0.279) \end{aligned}$ | $\begin{aligned} & 0.725 \\ & (0.303) \end{aligned}$ | $\begin{aligned} & 1.45 \\ & (0.269) \end{aligned}$ | $\begin{aligned} & 0.790 \\ & (0.292) \end{aligned}$ | $\begin{aligned} & 0.675 \\ & (0.315) \end{aligned}$ | $\begin{aligned} & 1.44 \\ & (0.264) \end{aligned}$ | | $(R \& D) / K$ | - | | - | | - | | $\begin{aligned} & 0.551 \\ & (0.225) \end{aligned}$ | $\begin{aligned} & 0.499 \\ & (0.268) \end{aligned}$ | $\begin{gathered} 0.384 \\ (0.409) \end{gathered}$ | - | - | - | | RDUM | | - | | - | | - | $\begin{gathered} 0.158 \\ (0.046) \end{gathered}$ | $\begin{gathered} 0.273 \\ (0.072) \end{gathered}$ | $\begin{gathered} 0.202 \\ (0.125) \end{gathered}$ | - | - | - | | A/K | | - | | - | | - | $\begin{aligned} & -0.046 \\ & (0.125) \end{aligned}$ | $\begin{aligned} & -0.402 \\ & (0.205) \end{aligned}$ | $\begin{gathered} 0.083 \\ (0.451) \end{gathered}$ | - | - | - | | ADUM | | - | | - | | - | $\begin{aligned} & 0.159 \\ & (0.041) \end{aligned}$ | $\begin{aligned} & 0.142 \\ & (0.055) \end{aligned}$ | $\begin{gathered} 0.144 \\ (0.095) \end{gathered}$ | - | - | - | | $I / K$ | | - | | - | | - | $\begin{aligned} & 0.776 \\ & (0.123) \end{aligned}$ | $\begin{aligned} & 0.713 \\ & (0.115) \end{aligned}$ | $\begin{gathered} 0.339 \\ (0.103) \end{gathered}$ | - | - | - | | # Obs. | | 2630 | | 2630 | | 2630 | 2630 | 2630 | 2630 | 2630 | 2630 | 2630 | | Adj. $R^{2}$ | | 0.016 | | 0.135 | | 0.584 | 0.131 | 0.215 | 0.630 | 0.075 | 0.181 | 0.626 | | $p$-value | | - | | 0.004 | | 0.126 | - | 0.001 | 0.037 | - | 0.002 | 0.018 | | Wald | | 34.641 | | 23.729 | | 1.379 | 25.396 | 20.928 | 1.984 | 41.12 | 20.133 | 1.937 | | pwald | | - | | - | | 0.71 | - | - | 0.576 | - | - | 0.586 |
The p p pp-values for this test statistic are reported in Table 5A and B. In both tables, the p p pp-values tend to be lower for the tests based on industry-level fixed-effects estimator. This presumably reflects the higher test power generally implied by the greater efficiency of the slope estimates. The rejection of the null hypothesis of exogeneity of managerial ownership is particularly strong for the spline specification reported in Table 5B. These results strongly suggest that reported results using such a specification are subject to endogeneity bias.
该测试统计量的 p p pp -值在表 5A 和 B 中报告。在这两个表中,基于行业层面固定效应估计的 p p pp -值往往较低。这可能反映了斜率估计更高效率所暗示的更高检验能力。对于表 5B 中报告的样条规范,管理所有权的外生性零假设被拒绝的证据特别强。这些结果强烈表明,使用这种规范报告的结果可能受到内生性偏差的影响。

An important caveat to all empirical work using fixed-effect estimators on panel data is that the ‘within’ estimator can, under a range of certain circumstances identified by Griliches and Hausman (1986), exacerbate the bias toward zero caused by measurement error. If our ownership variable were measured with classical error, then this would reduce the power of our Wald test for the joint significance of the ownership variables, and would invalidate the distributional assumptions for our test statistic due to the inconsistency of the residual estimates. While it is always possible to make an a priori case for measurement error, there is little empirical evidence that measurement error is a serious problem in our data. Table 4 A and B show, for example, that the within variation in managerial ownership is significantly correlated with the explanatory variables, a result that does not square with serious measurement error. In addition, the within-firm point estimates of the ownership coefficients in Table 5A and B are not obviously biased toward zero, as measurement error would suggest. Finally, the conditions identified by Griliches and Hausman might not hold. If the variance of the measurement error were primarily cross-sectional rather than within, then the within estimator would actually tend to reduce the bias effects of measurement error. We nevertheless recognize the limitations of the within estimator, and in the next section, we report instrumental variables estimated as an alternative approach to deal with the endogeneity of ownership variables.
使用固定效应估计量对面板数据进行的所有实证工作有一个重要的警告,即在 Griliches 和 Hausman(1986)确定的一系列特定情况下,“内部”估计量可能会加剧由测量误差引起的向零偏差。如果我们的所有权变量是用经典误差测量的,那么这将降低我们对所有权变量联合显著性的 Wald 检验的效能,并由于残差估计的不一致性而使我们的检验统计量的分布假设失效。虽然总是可以事先提出测量误差的理由,但在我们的数据中,几乎没有实证证据表明测量误差是一个严重的问题。例如,表 4 A 和 B 显示,管理所有权的内部变异与解释变量显著相关,这一结果与严重的测量误差不符。此外,表 5A 和 B 中所有权系数的内部公司点估计并没有明显向零偏差,正如测量误差所暗示的那样。最后,Griliches 和 Hausman 确定的条件可能并不成立。 如果测量误差的方差主要是横截面的而不是内部的,那么内部估计量实际上会倾向于减少测量误差的偏差效应。尽管如此,我们仍然认识到内部估计量的局限性,在下一节中,我们报告作为处理所有权变量内生性的一种替代方法的工具变量估计。

5.2. Toward a more structural interpretation of contracting relations
5.2. 朝着对合同关系更具结构性的解释

The strength of the empirical evidence against the exogeneity of managerial ownership suggests that more model structure is required to identify the impact of managerial ownership on firm value. A standard remedy would be to use an instrumental variable in the regression for firm value. In a related paper, Hermalin and Weisbach (1991) recognize a similar endogeneity problem and use lagged explanatory variables as instruments for managerial ownership. They find that the instrumental variable estimator increases the magnitude of the ownership effect on firm value. Hermalin and Weisbach report that a Hausman specification test rejects the exogeneity assumption. While this rejection provides evidence of endogenous ownership, it does not validate their choice of instruments. If omitted firm characteristics are the source of the endogeneity (as we have argued above), and if these unobserved firm characteristics change
关于管理层持股的外生性反对的实证证据的强度表明,需要更多的模型结构来识别管理层持股对公司价值的影响。一个标准的补救措施是在公司价值的回归中使用工具变量。在一篇相关论文中,Hermalin 和 Weisbach (1991) 认识到类似的内生性问题,并使用滞后解释变量作为管理层持股的工具。他们发现,工具变量估计量增加了持股对公司价值的影响程度。Hermalin 和 Weisbach 报告称,Hausman 规格检验拒绝了外生性假设。虽然这一拒绝提供了内生性持股的证据,但并未验证他们选择的工具。如果被遗漏的公司特征是内生性的来源(正如我们上面所论述的),并且如果这些未观察到的公司特征发生变化

slowly over time (as we have also argued above), then lagged explanatory variables will suffer as much from the endogeneity problem as do contemporaneous ones.
随着时间的推移(正如我们上面所论述的),滞后解释变量将与同时变量一样遭受内生性问题。
Instrumental variables for managerial ownership are difficult to find. The basic problem is that for any variable that plausibly determines the optimal level of managerial ownership, it is also possible to argue that the same variable might plausibly affect Tobin’s Q Q QQ. For example, our results in Table 4A and B showed that market power (as measured by operating margins) is a candidate instrument. However, even though it is correlated with managerial ownership, it also determines the equilibrium value of Tobin’s Q Q QQ, and therefore must appear independently in this regression. Additional candidates suggested by these results, such as the capital-to-sales ratio, advertising, R&D, and fixed investment, are also invalid because of links between investment and Q Q QQ and because intangible assets are conservatively valued on the balance sheet, therefore influencing the level of Tobin’s Q Q QQ.
管理层持股的工具变量很难找到。基本问题是,对于任何可能决定管理层持股最佳水平的变量,也可以辩称该变量可能会合理地影响托宾的 Q Q QQ 。例如,我们在表 4A 和 B 中的结果显示,市场力量(以营业利润率衡量)是一个候选工具。然而,尽管它与管理层持股相关,但它也决定了托宾的 Q Q QQ 的均衡价值,因此必须在这个回归中独立出现。这些结果建议的其他候选变量,如资本销售比、广告、研发和固定投资,也因投资与 Q Q QQ 之间的联系以及无形资产在资产负债表上被保守估值而无效,因此影响托宾的 Q Q QQ 水平。
A more plausible case can be made for using firm size and stock price volatility as instruments. It is possible to construct arguments under which either variable could be correlated with Tobin’s Q Q QQ. For example, suppose that high Q Q QQ values reflect future growth opportunities. Such firms might generally be smaller (or larger), and might also have more volatile stock prices due to the greater uncertainty about future growth prospects. However, these arguments seem weaker than the arguments against operating margins, the capital-to-sales ratio, advertising, R & D R & D R&D\mathrm{R} \& \mathrm{D}, and investment. Moreover, in studies of fixed investment, it is generally argued that deviations of Tobin’s Q Q QQ from its equilibrium value are explained by the costs of adjusting the capital stock, and that these adjustment costs are proportional to the rate of investment. Therefore, the inclusion of advertising and R & D R & D R&D\mathrm{R} \& \mathrm{D} intensity and the investment rate should control for future growth opportunities. This argument eliminates the a priori case for including the size and volatility variables in the Q Q QQ equation, and thus provides an argument for omitting these variables from the Q Q QQ equation and using them as instruments for managerial ownership instead.
更有说服力的案例可以提出使用公司规模和股票价格波动作为工具。可以构建论点,使得任一变量可能与托宾的 Q Q QQ 相关。例如,假设高 Q Q QQ 值反映未来的增长机会。这类公司可能通常较小(或较大),并且由于对未来增长前景的不确定性,股票价格也可能更为波动。然而,这些论点似乎比反对营业利润率、资本与销售比率、广告、 R & D R & D R&D\mathrm{R} \& \mathrm{D} 和投资的论点要弱。此外,在固定投资的研究中,通常认为托宾的 Q Q QQ 偏离其均衡值是由调整资本存量的成本所解释的,而这些调整成本与投资率成正比。因此,广告和 R & D R & D R&D\mathrm{R} \& \mathrm{D} 强度以及投资率的纳入应该能够控制未来的增长机会。 这个论点消除了在 Q Q QQ 方程中包含规模和波动性变量的先验理由,因此为从 Q Q QQ 方程中省略这些变量并将其作为管理层持股的工具提供了论据。
The results using L N ( S ) , ( L N ( S ) ) 2 , S I G M A L N ( S ) , ( L N ( S ) ) 2 , S I G M A LN(S),(LN(S))^(2),SIGMAL N(S),(L N(S))^{2}, S I G M A, and SIGDUM as instruments are reported in Table 6. We use the more parsimonious quadratic specification for managerial ownership to reduce the number of instruments required for identification. The first column of Table 6 reports the results of pooling without controlling for industry or firm effects. In contrast to Table 5A, these results confirm a large and statistically significant inverse-U relation between ownership and firm value. The coefficients of 6.29 and -10.8 on m m mm and m 2 m 2 m^(2)m^{2}, respectively, imply an inflection point of about 0.58 . Given the distribution of managerial ownership shown in Fig. 1, Tobin’s Q Q QQ is generally an increasing, concave function of m m mm.
使用 L N ( S ) , ( L N ( S ) ) 2 , S I G M A L N ( S ) , ( L N ( S ) ) 2 , S I G M A LN(S),(LN(S))^(2),SIGMAL N(S),(L N(S))^{2}, S I G M A 和 SIGDUM 作为工具的结果见表 6。我们使用更简约的二次规范来减少识别所需的工具数量。表 6 的第一列报告了不控制行业或公司效应的汇总结果。与表 5A 相比,这些结果确认了所有权与公司价值之间存在显著的逆 U 型关系。 m m mm m 2 m 2 m^(2)m^{2} 上的系数分别为 6.29 和-10.8,意味着拐点约为 0.58。根据图 1 所示的管理所有权分布,托宾的 Q Q QQ 通常是 m m mm 的一个递增的凹函数。
The second and third columns of Table 6 show that these results are robust to the inclusion of three-digit industry effects, but not to firm effects. In both
表 6 的第二和第三列显示,这些结果对三位数行业效应的包含是稳健的,但对公司效应则不是。在两者中
Table 6  表 6
Ownership-performance model with instrumental variables
使用工具变量的所有权-绩效模型

The specifications reported in this table all model firm value, Q Q QQ, as a linear function of the explanatory variables indicated below. Intercept terms and year dummies are included. Instruments are L N ( S ) , ( L N ( S ) ) 2 , S I G M A L N ( S ) , ( L N ( S ) ) 2 , S I G M A LN(S),(LN(S))^(2),SIGMAL N(S),(L N(S))^{2}, S I G M A, and S I G D U M S I G D U M SIGDUMS I G D U M. Fixed effects at the industry or firm level are included where indicated, but not reported. Variable definitions for the acronyms are given in Table 3.
本表中报告的规格将公司价值 Q Q QQ 作为下面所示解释变量的线性函数。包括截距项和年份虚拟变量。工具变量为 L N ( S ) , ( L N ( S ) ) 2 , S I G M A L N ( S ) , ( L N ( S ) ) 2 , S I G M A LN(S),(LN(S))^(2),SIGMAL N(S),(L N(S))^{2}, S I G M A S I G D U M S I G D U M SIGDUMS I G D U M 。在指示的地方包括行业或公司层面的固定效应,但未报告。缩略词的变量定义见表 3。
Variable  变量 Pooled  合并的 SIC3 effects  SIC3 效果 Firm effects  公司效应
m m mm 6.29 ( 1.27 ) 6.29 ( 1.27 ) {:[6.29],[(1.27)]:}\begin{gathered} 6.29 \\ (1.27) \end{gathered} 8.38 ( 1.78 ) 8.38 ( 1.78 ) {:[8.38],[(1.78)]:}\begin{gathered} 8.38 \\ (1.78) \end{gathered} 10.7 ( 10.6 ) 10.7 ( 10.6 ) {:[-10.7],[(10.6)]:}\begin{array}{r} -10.7 \\ (10.6) \end{array}
m 2 m 2 m^(2)m^{2} 10.8 ( 2.61 ) 10.8 ( 2.61 ) {:[-10.8],[(2.61)]:}\begin{gathered} -10.8 \\ (2.61) \end{gathered} 12.3 ( 3.48 ) 12.3 ( 3.48 ) {:[-12.3],[(3.48)]:}\begin{gathered} -12.3 \\ (3.48) \end{gathered} 4.88 ( 10.8 ) 4.88 ( 10.8 ) {:[-4.88],[(10.8)]:}\begin{gathered} -4.88 \\ (10.8) \end{gathered}
K / S K / S K//SK / S 1.45 ( 0.220 ) 1.45 ( 0.220 ) {:[1.45],[(0.220)]:}\begin{aligned} & 1.45 \\ & (0.220) \end{aligned} 0.736 ( 0.376 ) 0.736 ( 0.376 ) {:[0.736],[(0.376)]:}\begin{gathered} 0.736 \\ (0.376) \end{gathered} 1.25 ( 1.01 ) 1.25 ( 1.01 ) {:[1.25],[(1.01)]:}\begin{gathered} 1.25 \\ (1.01) \end{gathered}
( K / S ) 2 ( K / S ) 2 (K//S)^(2)(K / S)^{2} 0.738 ( 0.119 ) 0.738 ( 0.119 ) {:[-0.738],[(0.119)]:}\begin{array}{r} -0.738 \\ (0.119) \end{array} 0.587 ( 0.171 ) 0.587 ( 0.171 ) {:[-0.587],[(0.171)]:}\begin{array}{r} -0.587 \\ (0.171) \end{array} 0.611 ( 0.396 ) 0.611 ( 0.396 ) {:[-0.611],[(0.396)]:}\begin{array}{r} -0.611 \\ (0.396) \end{array}
Y / S Y / S Y//SY / S 0.414 ( 0.273 ) 0.414 ( 0.273 ) {:[0.414],[(0.273)]:}\begin{gathered} 0.414 \\ (0.273) \end{gathered} 0.603 ( 0.300 ) 0.603 ( 0.300 ) {:[0.603],[(0.300)]:}\begin{gathered} 0.603 \\ (0.300) \end{gathered} 1.003 ( 0.637 ) 1.003 ( 0.637 ) {:[1.003],[(0.637)]:}\begin{gathered} 1.003 \\ (0.637) \end{gathered}
( R & D ) / K ( R & D ) / K (R&D)//K(R \& D) / K 0.739 ( 0.242 ) 0.739 ( 0.242 ) {:[0.739],[(0.242)]:}\begin{gathered} 0.739 \\ (0.242) \end{gathered} 0.718 ( 0.339 ) 0.718 ( 0.339 ) {:[0.718],[(0.339)]:}\begin{gathered} 0.718 \\ (0.339) \end{gathered} 0.687 ( 1.11 ) 0.687 ( 1.11 ) {:[0.687],[(1.11)]:}\begin{gathered} 0.687 \\ (1.11) \end{gathered}
RDUM 0.178 ( 0.049 ) 0.178 ( 0.049 ) {:[0.178],[(0.049)]:}\begin{gathered} 0.178 \\ (0.049) \end{gathered} 0.256 ( 0.115 ) 0.256 ( 0.115 ) {:[0.256],[(0.115)]:}\begin{gathered} 0.256 \\ (0.115) \end{gathered} 0.230 ( 0.805 ) 0.230 ( 0.805 ) {:[-0.230],[(0.805)]:}\begin{array}{r} -0.230 \\ (0.805) \end{array}
A/K 0.169 ( 0.164 ) 0.169 ( 0.164 ) {:[0.169],[(0.164)]:}\begin{gathered} 0.169 \\ (0.164) \end{gathered} 0.735 ( 0.290 ) 0.735 ( 0.290 ) {:[-0.735],[(0.290)]:}\begin{array}{r} -0.735 \\ (0.290) \end{array} 0.007 ( 1.262 ) 0.007 ( 1.262 ) {:[0.007],[(1.262)]:}\begin{gathered} 0.007 \\ (1.262) \end{gathered}
ADUM 0.140 ( 0.043 ) 0.140 ( 0.043 ) {:[0.140],[(0.043)]:}\begin{gathered} 0.140 \\ (0.043) \end{gathered} 0.251 ( 0.076 ) 0.251 ( 0.076 ) {:[0.251],[(0.076)]:}\begin{gathered} 0.251 \\ (0.076) \end{gathered} 0.301 ( 0.234 ) 0.301 ( 0.234 ) {:[0.301],[(0.234)]:}\begin{gathered} 0.301 \\ (0.234) \end{gathered}
I / K I / K I//KI / K 0.990 ( 0.140 ) 0.990 ( 0.140 ) {:[0.990],[(0.140)]:}\begin{gathered} 0.990 \\ (0.140) \end{gathered} 0.937 ( 0.151 ) 0.937 ( 0.151 ) {:[0.937],[(0.151)]:}\begin{gathered} 0.937 \\ (0.151) \end{gathered} 0.478 ( 0.202 ) 0.478 ( 0.202 ) {:[0.478],[(0.202)]:}\begin{gathered} 0.478 \\ (0.202) \end{gathered}
# Obs.
Adj. R 2 R 2 R^(2)R^{2}
Wald
pwald
# Obs. Adj. R^(2) Wald pwald| # Obs. | | :--- | | Adj. $R^{2}$ | | Wald | | pwald |
2630 0.197 67.576 2630 0.197 67.576 {:[2630],[-0.197],[67.576]:}\begin{aligned} & 2630 \\ & -0.197 \\ & 67.576 \end{aligned} 2630 0.057 55.659 2630 0.057 55.659 {:[2630],[-0.057],[55.659]:}\begin{aligned} & 2630 \\ & -0.057 \\ & 55.659 \end{aligned} 2630 0.192 1.310 0.519 2630 0.192 1.310 0.519 {:[2630],[0.192],[1.310],[0.519]:}\begin{array}{r} 2630 \\ 0.192 \\ 1.310 \\ 0.519 \end{array}
Variable Pooled SIC3 effects Firm effects m "6.29 (1.27)" "8.38 (1.78)" "-10.7 (10.6)" m^(2) "-10.8 (2.61)" "-12.3 (3.48)" "-4.88 (10.8)" K//S "1.45 (0.220)" "0.736 (0.376)" "1.25 (1.01)" (K//S)^(2) "-0.738 (0.119)" "-0.587 (0.171)" "-0.611 (0.396)" Y//S "0.414 (0.273)" "0.603 (0.300)" "1.003 (0.637)" (R&D)//K "0.739 (0.242)" "0.718 (0.339)" "0.687 (1.11)" RDUM "0.178 (0.049)" "0.256 (0.115)" "-0.230 (0.805)" A/K "0.169 (0.164)" "-0.735 (0.290)" "0.007 (1.262)" ADUM "0.140 (0.043)" "0.251 (0.076)" "0.301 (0.234)" I//K "0.990 (0.140)" "0.937 (0.151)" "0.478 (0.202)" "# Obs. Adj. R^(2) Wald pwald" "2630 -0.197 67.576" "2630 -0.057 55.659" "2630 0.192 1.310 0.519"| Variable | Pooled | SIC3 effects | Firm effects | | :---: | :---: | :---: | :---: | | $m$ | $\begin{gathered} 6.29 \\ (1.27) \end{gathered}$ | $\begin{gathered} 8.38 \\ (1.78) \end{gathered}$ | $\begin{array}{r} -10.7 \\ (10.6) \end{array}$ | | $m^{2}$ | $\begin{gathered} -10.8 \\ (2.61) \end{gathered}$ | $\begin{gathered} -12.3 \\ (3.48) \end{gathered}$ | $\begin{gathered} -4.88 \\ (10.8) \end{gathered}$ | | $K / S$ | $\begin{aligned} & 1.45 \\ & (0.220) \end{aligned}$ | $\begin{gathered} 0.736 \\ (0.376) \end{gathered}$ | $\begin{gathered} 1.25 \\ (1.01) \end{gathered}$ | | $(K / S)^{2}$ | $\begin{array}{r} -0.738 \\ (0.119) \end{array}$ | $\begin{array}{r} -0.587 \\ (0.171) \end{array}$ | $\begin{array}{r} -0.611 \\ (0.396) \end{array}$ | | $Y / S$ | $\begin{gathered} 0.414 \\ (0.273) \end{gathered}$ | $\begin{gathered} 0.603 \\ (0.300) \end{gathered}$ | $\begin{gathered} 1.003 \\ (0.637) \end{gathered}$ | | $(R \& D) / K$ | $\begin{gathered} 0.739 \\ (0.242) \end{gathered}$ | $\begin{gathered} 0.718 \\ (0.339) \end{gathered}$ | $\begin{gathered} 0.687 \\ (1.11) \end{gathered}$ | | RDUM | $\begin{gathered} 0.178 \\ (0.049) \end{gathered}$ | $\begin{gathered} 0.256 \\ (0.115) \end{gathered}$ | $\begin{array}{r} -0.230 \\ (0.805) \end{array}$ | | A/K | $\begin{gathered} 0.169 \\ (0.164) \end{gathered}$ | $\begin{array}{r} -0.735 \\ (0.290) \end{array}$ | $\begin{gathered} 0.007 \\ (1.262) \end{gathered}$ | | ADUM | $\begin{gathered} 0.140 \\ (0.043) \end{gathered}$ | $\begin{gathered} 0.251 \\ (0.076) \end{gathered}$ | $\begin{gathered} 0.301 \\ (0.234) \end{gathered}$ | | $I / K$ | $\begin{gathered} 0.990 \\ (0.140) \end{gathered}$ | $\begin{gathered} 0.937 \\ (0.151) \end{gathered}$ | $\begin{gathered} 0.478 \\ (0.202) \end{gathered}$ | | # Obs. <br> Adj. $R^{2}$ <br> Wald <br> pwald | $\begin{aligned} & 2630 \\ & -0.197 \\ & 67.576 \end{aligned}$ | $\begin{aligned} & 2630 \\ & -0.057 \\ & 55.659 \end{aligned}$ | $\begin{array}{r} 2630 \\ 0.192 \\ 1.310 \\ 0.519 \end{array}$ |
Notes: Standard errors are in parentheses. Wald and pwald report, respectively, the Wald statistic and associated p p pp-value for a test that the ownership variables are jointly zero.
注意:标准误差在括号中。Wald 和 pwald 分别报告 Wald 统计量和与所有权变量联合为零的检验相关的 p p pp -值。

specifications, the standard errors rise substantially, rendering the coefficients statistically indistinguishable from zero. However, this need not be interpreted as bad news for the results reported in the first column. One cannot reject the hypothesis that the firm effects are jointly zero, though one can reject that the industry effects are jointly zero (the p p pp-values on the associated Hausman tests
规范,标准误差大幅上升,使得系数在统计上与零无法区分。然而,这不必被解读为对第一列报告结果的坏消息。虽然不能拒绝公司效应共同为零的假设,但可以拒绝行业效应共同为零的假设(与相关 Hausman 检验的 p p pp -值)。

are 0.208 for the test of pooled versus the inclusion of firm effects and 0.00002 for the test of pooled versus the inclusion of industry effects). This conclusion has intuitive appeal because, by using instrumental variables, we have presumably controlled for the endogeneity that was the motivation for including firm fixed effects. However, it is more likely that the combined effect of using instrumental variables and controlling for fixed effects has reduced the precision of estimates to the point at which such a test would have little power. We believe that the results in Table 6 represent a promising step toward the construction of more complete models of the relation between managerial ownership and firm performance.
对于合并与包含公司效应的测试,值为 0.208;对于合并与包含行业效应的测试,值为 0.00002。这个结论具有直观的吸引力,因为通过使用工具变量,我们大概控制了包含公司固定效应的动机所导致的内生性。然而,更可能的是,使用工具变量和控制固定效应的综合效果降低了估计的精确度,以至于这样的测试几乎没有效力。我们相信,表 6 中的结果代表了朝着构建管理层持股与公司绩效之间关系的更完整模型迈出的有希望的一步。

6. Conclusions  6. 结论

Firms are governed by a network of relations representing contracts for financing, capital structure, and managerial ownership and compensation, among others. For any of these contractual arrangements, it is difficult to identify the correspondence between the contractual choice and firm performance (e.g., measured by accounting rates of return or Tobin’s Q Q QQ ) because contractual choices and performance outcomes are endogenously determined by exogenous and only partly observed features of the firm’s contracting environment.
公司由一系列关系构成,这些关系代表了融资、资本结构以及管理层所有权和薪酬等方面的合同。对于这些合同安排中的任何一种,识别合同选择与公司绩效之间的对应关系是困难的(例如,通过会计收益率或托宾的 Q Q QQ 来衡量),因为合同选择和绩效结果是由外生因素和公司合同环境中仅部分可观察的特征内生决定的。
We confront this endogeneity problem in the context of the firm’s compensation contract with managers. Because managerial equity stakes are an important and well-known mechanism used to align the incentives of managers and owners, we examine the determinants of managerial ownership as a function of the contracting environment. We extend the cross-sectional results of Demsetz and Lehn (1985) and use panel data to show that managerial ownership is explained by variables describing the contracting environment in ways consistent with the predictions of principal-agent models.
我们在公司的管理者薪酬合同的背景下面对这个内生性问题。由于管理者的股权持有是用来对齐管理者和所有者激励的重要且众所周知的机制,我们考察了管理者所有权作为合同环境的一个函数的决定因素。我们扩展了 Demsetz 和 Lehn(1985)的横截面结果,并使用面板数据表明,管理者所有权可以通过描述合同环境的变量来解释,这与委托-代理模型的预测是一致的。
We find that a large fraction of the cross-sectional variation in managerial ownership is explained by unobserved firm heterogeneity. This unobserved heterogeneity in the contracting environment has important implications for econometric models designed to estimate the effect of managerial ownership on firm performance. Our empirical analysis shows that existing results are not robust to controls for endogeneity induced by time-invariant unobserved heterogeneity. Moreover, once we control both for observed firm characteristics and firm fixed effects, it becomes difficult to conclude that changes in firm managerial ownership affect performance. Our instrumental-variables results, however, suggest a promising step toward the construction of more complete models of the relation between managerial ownership and firm performance.
我们发现,管理层持股的横截面变异中有很大一部分可以用未观察到的公司异质性来解释。这种在契约环境中的未观察到的异质性对旨在估计管理层持股对公司绩效影响的计量经济学模型具有重要意义。我们的实证分析表明,现有结果对因时间不变的未观察到的异质性引起的内生性控制并不稳健。此外,一旦我们同时控制观察到的公司特征和公司固定效应,就很难得出管理层持股的变化影响绩效的结论。然而,我们的工具变量结果表明,朝着构建管理层持股与公司绩效之间关系的更完整模型迈出了有希望的一步。
To take these observations one step further, we believe that the Q Q QQ model results reported in Table 5A and B can be interpreted as supporting more
为了进一步推进这些观察,我们认为表 5A 和 B 中报告的 Q Q QQ 模型结果可以被解读为支持更多

generally the notion that the firm chooses among alternative mechanisms for minimizing agency costs. This is, of course, the concept articulated in Alchian (1969); Fama (1980); Fama and Jensen (1983) and Demsetz and Lehn (1985); more recently, see Crutchley and Hansen (1989) and Agrawal and Knoeber (1996).
一般来说,公司在不同机制中选择以最小化代理成本的概念。当然,这是在 Alchian (1969)、Fama (1980)、Fama 和 Jensen (1983)以及 Demsetz 和 Lehn (1985)中阐述的概念;最近,参见 Crutchley 和 Hansen (1989)以及 Agrawal 和 Knoeber (1996)。
Suppose, for example, that Q Q QQ capitalizes the market’s expectation of the effect of agency costs on firm value. The loss in value reflects residual agency costs, or agency costs remaining after corporate control mechanisms are chosen. In addition to managerial ownership choices emphasized here, alternative means of reducing agency costs include leverage (Jensen, 1986; Gertler and Hubbard, 1993), increased reliance on outside directors (American Law Institute, 1982; Baysinger and Butler, 1985; Millstein and MacAvoy, 1998), large shareholders (Shleifer and Vishny, 1986; Zeckhauser and Pound, 1990), institutional investors, dividend policy (Easterbrook, 1984), and radical changes in corporate control (Kaplan, 1989).
假设,例如, Q Q QQ 体现了市场对代理成本对公司价值影响的预期。价值的损失反映了剩余的代理成本,即在选择公司控制机制后仍然存在的代理成本。除了这里强调的管理层持股选择,减少代理成本的其他方式包括杠杆(Jensen, 1986;Gertler 和 Hubbard, 1993)、对外部董事的依赖增加(American Law Institute, 1982;Baysinger 和 Butler, 1985;Millstein 和 MacAvoy, 1998)、大股东(Shleifer 和 Vishny, 1986;Zeckhauser 和 Pound, 1990)、机构投资者、股息政策(Easterbrook, 1984)以及公司控制的激进变革(Kaplan, 1989)。
One can interpret the results in Table 5A, B, and 6 as reduced-form exercises in which the x x xx variables and the firm fixed effects are determinants of the use of these mechanisms. For example, Gertler and Hubbard (1993) relate leverage in this context to the relative importance of firm-specific and aggregate risk and proxies for the scope for moral hazard (variables captured by R&D and advertising intensity, year dummies, and firm effects). Benefits from large shareholders likely depend on size or the relative importance of R&D (Zeckhauser and Pound, 1990); these channels are proxied through size, R&D intensity, and firm effects. Institutional shareholdings likely depend on firm size and whether the firm is listed on the New York Stock Exchange (variables accounted for in part by firm effects). The degree to which dividend policy can reduce agency costs depends on the importance of the scope for moral hazard (perhaps measured by idiosyncratic risk, R & D R & D R&D\mathrm{R} \& D, or advertising) and the tax costs of paying dividends (measured in part by year and firm effects). Net benefits of a major restructuring also are reflected in proxies for moral hazard and firm effects.
可以将表 5A、B 和 6 中的结果解释为简化形式的练习,其中 x x xx 变量和公司固定效应是使用这些机制的决定因素。例如,Gertler 和 Hubbard(1993)在此背景下将杠杆与公司特定风险和整体风险的相对重要性以及道德风险的范围(通过研发和广告强度、年份虚拟变量和公司效应捕捉的变量)相关联。大股东的利益可能依赖于规模或研发的相对重要性(Zeckhauser 和 Pound,1990);这些渠道通过规模、研发强度和公司效应进行代理。机构持股可能依赖于公司规模以及公司是否在纽约证券交易所上市(部分由公司效应考虑的变量)。股息政策在多大程度上可以减少代理成本取决于道德风险范围的重要性(可能通过特异风险 R & D R & D R&D\mathrm{R} \& D 或广告来衡量)以及支付股息的税收成本(部分通过年份和公司效应来衡量)。重大重组的净收益也反映在道德风险和公司效应的代理中。
Two other possible strategies are tasks for future research. The first involves identifying large, arguably exogenous changes in ownership levels arising from shifts in tax policy, regulation, or fixed costs in the market for corporate control (Kaplan, 1989; Hubbard and Palia, 1995; Cole and Mehran, 1997), though care must be taken because even certain ‘natural experiments’ are endogenous in that they affect performance directly. The second involves designing a dynamic structural model of firm contracting decisions, possibly permitting identification from economically reasonable assumptions about functional form (Margiotta and Miller, 1991). This strategy is particularly desirable given the lack of easily identified instrumental variables.
另外两种可能的策略是未来研究的任务。第一种涉及识别由于税收政策、监管或企业控制市场中的固定成本变化而产生的大规模、可以说是外生的所有权水平变化(Kaplan, 1989; Hubbard and Palia, 1995; Cole and Mehran, 1997),但必须小心,因为即使某些“自然实验”也是内生的,因为它们直接影响绩效。第二种涉及设计一个动态结构模型,以研究公司合同决策,可能允许从对功能形式的经济合理假设中进行识别(Margiotta and Miller, 1991)。考虑到缺乏易于识别的工具变量,这种策略特别可取。
While our findings are consistent with the proposition that firms choose strategies to reduce agency costs optimally over the long run, at least two issues remain. First, the simultaneous choice of individual mechanisms or some subset
虽然我们的研究结果与公司选择策略以在长期内最佳地减少代理成本的提议一致,但仍然存在至少两个问题。首先,个别机制或某些子集的同时选择

needs to be modeled; subsets of these choices have been considered in a re-duced-form setting, as in Hermalin and Weisbach (1991); Jensen et al. (1992); Moyer et al. (1992); Holthausen and Larcker (1993) and Agrawal and Knoeber (1996). Second, the choice of mechanisms likely involves some fixed costs or ‘costs of adjustment’ so that firms are not always at their long-run contractual optimum. Exploring these costs and how they might have changed over time for different agency-cost-reducing mechanisms is a particularly interesting task for future research.
需要建模;这些选择的子集已在简化形式的环境中考虑,如 Hermalin 和 Weisbach(1991);Jensen 等(1992);Moyer 等(1992);Holthausen 和 Larcker(1993)以及 Agrawal 和 Knoeber(1996)。其次,机制的选择可能涉及一些固定成本或“调整成本”,因此公司并不总是在其长期合同最优状态。探索这些成本以及它们如何随着时间的推移在不同的降低代理成本机制中发生变化,是未来研究中特别有趣的任务。

References  参考文献

Agrawal, A., Knoeber, C., 1996. Firm performance and mechanisms to control agency problems between managers and shareholders. Journal of Financial and Quantitative Analysis 31, 377-397.
Agrawal, A., Knoeber, C., 1996. 公司绩效与控制管理者与股东之间代理问题的机制。《金融与定量分析杂志》31, 377-397.

Alchian, A., 1969. Corporate management and property rights. In: Henry Manne, (Ed.), Economic Policy and Regulation of Corporate Securities. American Enterprise Institute for Public Policy Research, Washington, DC.
阿尔基安,A.,1969 年。企业管理与产权。在:亨利·曼恩,(编),《经济政策与企业证券的监管》。美国企业研究所,华盛顿特区。

American Law Institute, 1982. Principles of Corporate Governance and Structure: Restatement and Recommendations. ALI, New York.
美国法律研究所,1982 年。《公司治理与结构原则:重述与建议》。ALI,纽约。

Baysinger, B., Butler, H., 1985. Corporate governance and the board of directors: performance effects of changes in board composition. Journal of Law, Economics, and Organization 1, 102-123.
Baysinger, B., Butler, H., 1985. 公司治理与董事会:董事会组成变化的绩效影响。《法律、经济与组织杂志》1, 102-123.

Berle, A., Means, G., 1932. The Modern Corporation and Private Property. Macmillan, New York.
伯尔,A.,米恩斯,G.,1932 年。《现代公司与私人财产》。麦克米伦出版社,纽约。

Cole, R., Mehran, H., 1997. The effect of changes in ownership structure on performance: evidence from the thrift industry. Mimeograph, Board of Governors of the Federal Reserve System.
科尔,R.,梅赫兰,H.,1997 年。所有权结构变化对绩效的影响:来自储蓄行业的证据。复印件,美国联邦储备系统董事会。

Crutchley, C., Hansen, R., 1989. A test of the agency theory of managerial ownership, corporate leverage, and corporate dividends. Financial Management 18, 36-46.
Crutchley, C., Hansen, R., 1989. 一项关于管理层持股、公司杠杆和公司分红的代理理论测试。金融管理 18, 36-46。

Demsetz, H., Lehn, K., 1985. The structure of corporate ownership: causes and consequences. Journal of Political Economy 93, 1155-1177.
Demsetz, H., Lehn, K., 1985. 企业所有权的结构:原因与后果。政治经济学杂志 93, 1155-1177.

Dimson, E., 1979. Risk measurement when shares are subject to infrequent trading. Journal of Financial Economics 7, 197-226.
Dimson, E., 1979. 当股票交易不频繁时的风险测量。金融经济学杂志 7, 197-226.

Easterbrook, F., 1984. Two agency-cost explanations of dividends. American Economic Review 78, 650-659.
Easterbrook, F., 1984. 两种代理成本对股息的解释。美国经济评论 78, 650-659.

Fama, E., 1980. Agency problems and the theory of the firm. Journal of Political Economy 88, 288-307.
法马,E.,1980 年。代理问题与公司理论。《政治经济学杂志》88,288-307。

Fama, E., Jensen, M., 1983. Separation of ownership and control. Journal of Law and Economics 26, 301-325.
法马,E.,詹森,M.,1983 年。所有权与控制的分离。《法律与经济学杂志》26,301-325。

Fershtman, C., Judd, K., 1987. Equilibrium incentives in oligopoly. American Economic Review 77, 927-940.
Fershtman, C., Judd, K., 1987. 寡头垄断中的均衡激励. 美国经济评论 77, 927-940.

Gertler, M., Hubbard, R.G., 1988. Financial factors in business fluctuations. Financial Market Volatility: Causes and Consequences. Federal Reserve Bank, Kansas City.
Gertler, M., Hubbard, R.G., 1988. 商业波动中的金融因素。金融市场波动:原因与后果。堪萨斯城联邦储备银行。

Gertler, M., Hubbard, R.G., 1993. Corporate financial policy, taxation, and macroeconomic risk. RAND Journal of Economics 24, 286-303.
Gertler, M., Hubbard, R.G., 1993. 企业财务政策、税收与宏观经济风险。RAND 经济学杂志 24, 286-303。

Greene, W., 1997. Econometric Analysis. Prentice Hall, Upper Saddle River, NJ.
格林,W.,1997 年。《计量经济学分析》。普伦蒂斯霍尔,上萨德尔河,新泽西州。

Griliches, Z., Hausman, J., 1986. Errors in variables in panel data. Journal of Econometrics 31, 93-188.
Griliches, Z., Hausman, J., 1986. 面板数据中的变量误差。计量经济学杂志 31, 93-188.

Hausman, J., 1978. Specification tests in econometrics. Econometrica 46, 1251-1271.
Hausman, J., 1978. 计量经济学中的规范性检验。经济计量学 46, 1251-1271。
Hermalin, B., Weisbach, M., 1991. The effects of board compensation and direct incentives on firm performance. Financial Management 20, 101-112.
Hermalin, B., Weisbach, M., 1991. 董事会薪酬和直接激励对公司绩效的影响。金融管理 20, 101-112.

Holderness, C., Kroszner, R., Sheehan, D., 1999. Were the good old days that good?: evolution of managerial stock ownership and corporate governance since the great depression. Journal of Finance 54, 435-469.
霍尔德尼斯,C.,克罗斯纳,R.,希汉,D.,1999 年。过去的美好时光真的那么美好吗?:自大萧条以来管理层股票持有和公司治理的演变。《金融学杂志》54,435-469。

Holthausen, R., Larcker, D., 1993. Organizational structure and financial performance. Unpublished working paper. The Wharton School.
霍尔特豪森,R.,拉克,D.,1993 年。组织结构与财务绩效。未发表的工作论文。沃顿商学院。

Hubbard, R.G., Palia, D., 1995. Executive pay and performance: evidence from the U.S. banking industry. Journal of Financial Economics 39, 105-130.
哈伯德, R.G., 帕利亚, D., 1995. 高管薪酬与绩效:来自美国银行业的证据. 财务经济学杂志 39, 105-130.

Jensen, G., Solberg, D., Zorn, T., 1992. Simultaneous determination of insider ownership, debt, and dividend policies. Journal of Financial and Quantitative Analysis 27, 247-263.
詹森,G.,索尔伯格,D.,佐恩,T.,1992 年。内部持股、债务和分红政策的同时确定。《金融与定量分析杂志》27,247-263。

Jensen, M., 1986. Agency costs of free cash flow, corporate finance, and takeovers. American Economic Review 76, 323-329.
詹森,M.,1986 年。自由现金流、公司财务与收购的代理成本。《美国经济评论》76,323-329。

Jensen, M., Meckling, W., 1976. Theory of the firm: managerial behavior, agency costs and ownership structure. Journal of Financial Economics 3, 305-360.
詹森,M.,梅克林,W.,1976 年。《公司的理论:管理行为、代理成本和所有权结构》。金融经济学杂志 3,305-360。

Jensen, M., Warner, J., 1988. The distribution of power among corporate managers, shareholders, and directors. Journal of Financial Economics 20, 3-24.
詹森,M.,华纳,J.,1988 年。公司经理、股东和董事之间的权力分配。《金融经济学杂志》20,3-24。

Kaplan, S., 1989. Effects of management buyouts on operating performance and value. Journal of Financial Economics 24, 217-254.
Kaplan, S., 1989. 管理层收购对经营绩效和价值的影响。金融经济学杂志 24, 217-254.

Kole, S., 1995. Measuring managerial equity ownership: a comparison of sources of ownership data. Journal of Corporate Finance 1, 413-435.
Kole, S., 1995. 测量管理层股权拥有情况:所有权数据来源的比较。《公司财务杂志》1, 413-435.

Kole, S., 1996. Managerial ownership and firm performance: incentives or rewards? Advances in Financial Economics 2, 119-149.
Kole, S., 1996. 管理层持股与公司绩效:激励还是奖励?金融经济学进展 2, 119-149.

Margiotta, M., Miller, R., 1991. Managerial compensation and the cost of moral hazard. Unpublished working paper. Carnegie Mellon University.
Margiotta, M., Miller, R., 1991. 管理人员薪酬与道德风险成本。未发表的工作论文。卡内基梅隆大学。

McConnell, J., Servaes, H., 1990. Additional evidence on equity ownership and corporate value. Journal of Financial Economics 27, 595-612.
麦康奈尔, J., 塞尔瓦斯, H., 1990. 关于股权所有权与公司价值的额外证据. 财务经济学杂志 27, 595-612.

Millstein, I., MacAvoy, P., 1998. The active board of directors and improved performance of the large publicly-traded corporation. Columbia Law Review 98, 1283-1322.
Millstein, I., MacAvoy, P., 1998. 活跃的董事会与大型上市公司的业绩提升。哥伦比亚法律评论 98, 1283-1322.

Mørck, R., Shleifer, A., Vishny, R., 1988. Management ownership and market valuation. Journal of Financial Economics 20, 293-315.
Mørck, R., Shleifer, A., Vishny, R., 1988. 管理层持股与市场估值。金融经济学杂志 20, 293-315.

Moyer, R., Rao, R., Sisneros, P., 1992. Substitutes for voting rights: evidence from dual class recapitalization. Financial Management 21, 35-47.
Moyer, R., Rao, R., Sisneros, P., 1992. 投票权的替代品:来自双重类别资本重组的证据。金融管理 21, 35-47。

Newey, W., 1985. Generalized method of moments specification testing. Journal of Econometrics 29, 229-256.
Newey, W., 1985. 广义矩方法的规范检验. 计量经济学杂志 29, 229-256.

Scholes, M., Williams, J., 1976. Estimating betas from nonsynchronous data. Journal of Financial Economics 5, 309-327.
Scholes, M., Williams, J., 1976. 从非同步数据估计贝塔值。金融经济学杂志 5, 309-327.

Shleifer, A., Vishny, R., 1986. Large shareholders and corporate control. Journal of Political Economy 94, 461-488.
Shleifer, A., Vishny, R., 1986. 大股东与公司控制。政治经济学杂志 94, 461-488.

Zeckhauser, R., Pound, J., 1990. Are large shareholders effective monitors?: an investigation of share ownership and corporate performance. In: Hubbard, R.G. (Ed.), Asymmetric Information, Corporate Finance, and Investment. University of Chicago Press, Chicago.
Zeckhauser, R., Pound, J., 1990. 大股东是否有效监控?:对股权和公司绩效的调查。载于:Hubbard, R.G. (编),不对称信息、公司财务与投资。芝加哥大学出版社,芝加哥。

    • Corresponding author. Tel.: + 1-212-854-3493; fax: + 1-212-864-6184.
      通讯作者。电话:+ 1-212-854-3493;传真:+ 1-212-864-6184。
    E-mail addresses: cph15@columbia.edu (C.P. Himmelberg), rgh1@columbia.edu (R.G. Hubbard), dnp1@columbia.edu (D. Palia)
    电子邮件地址:cph15@columbia.edu (C.P. Himmelberg),rgh1@columbia.edu (R.G. Hubbard),dnp1@columbia.edu (D. Palia)

    ² ²  ^("² "){ }^{\text {² }}² We are grateful for helpful comments and suggestions from two anonymous referees and from Anup Agrawal, George Baker, Sudipto Bhattacharya, Steve Bond, Charles Calomiris, Harold Demsetz, Rob Hansen, Laurie Hodrick, Randy Kroszner, Mark Mitchell, Andrew Samwick, Bill Schwert (the editor), Scott Stern, Rob Vishny, and Karen Wruck, as well as participants in seminars at Boston College, Columbia University, University of Florida, Harvard University, London School of Economics, Massachusetts Institute of Technology, Virginia Tech, the 1998 Western Finance Association meetings, and the National Bureau of Economic Research.
    我们感谢两位匿名评审以及 Anup Agrawal、George Baker、Sudipto Bhattacharya、Steve Bond、Charles Calomiris、Harold Demsetz、Rob Hansen、Laurie Hodrick、Randy Kroszner、Mark Mitchell、Andrew Samwick、Bill Schwert(编辑)、Scott Stern、Rob Vishny 和 Karen Wruck 的有益评论和建议,以及波士顿学院、哥伦比亚大学、佛罗里达大学、哈佛大学、伦敦经济学院、麻省理工学院、弗吉尼亚理工大学、1998 年西部金融协会会议和国家经济研究局的研讨会参与者。
  1. 1 1 ^(1){ }^{1} We are grateful to Andrew Samwick for this calculation.
    我们感谢安德鲁·萨姆威克对此计算的贡献。
  2. Note: Estimated standard errors (reported in parentheses) are consistent in the presence of heteroskedasticity. The adjusted R 2 R 2 R^(2)R^{2} statistics reflect the inclusion of fixed effects (where included). (The ’ p p pp-value’ is the probability of observing the test statistic for endogeneity described in the text. Low p p pp-values suggest that ownership is endogenous.) Wald and pwald, report, respectively, the Wald statistic and associated p-value for a test that the managerial ownership variables are jointly zero.
    注意:估计的标准误差(括号内报告)在异方差存在的情况下是一致的。调整后的 R 2 R 2 R^(2)R^{2} 统计量反映了固定效应的包含(如果包含的话)。(‘ p p pp -值’是观察到文本中描述的内生性检验统计量的概率。低 p p pp -值表明所有权是内生的。)Wald 和 pwald 分别报告管理所有权变量共同为零的检验的 Wald 统计量和相关的 p 值。