Managerial Autonomy, Incentives and Firm Performance: Evidence from Investment Climate Survey in China Xiaoyang Li ♣ October 2007 Abstract This paper studies the relationship between firms’ use of incentive compensation and managerial autonomy, as well as how managerial autonomy affects firm performance. We develop a simple framework in which the principal employs compensation contract and delegation of autonomy to balance the tradeoff between delegation benefits and agency costs. We conduct the empirical analysis using Investment Climate Survey data from China. Our results show that: (i) firm’s use incentive compensation is negatively associated with general manager’s investment decision autonomy but positively associated with labor decision autonomy; (ii) general manager’s investment decision autonomy dampens, while labor decision autonomy boosts firm performance. ♣ I benefit from discussions with Francine Lafontaine, Uday Rajan and Katherine Terrell. I would like to acknowledge the World Bank Enterprise Survey Unit for allowing me to use the Investment Climate Survey data. All remaining errors are my own. 1
27
Embed
Managerial Autonomy, Incentives and Firm Performance ...webuser.bus.umich.edu/jagadees/other/Managerial Autonomy Incentives... · Jensen and Meckling (1976) define agency cost as
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Managerial Autonomy, Incentives and Firm Performance:
Evidence from Investment Climate Survey in China
Xiaoyang Li♣
October 2007
Abstract
This paper studies the relationship between firms’ use of incentive compensation and
managerial autonomy, as well as how managerial autonomy affects firm performance. We
develop a simple framework in which the principal employs compensation contract and
delegation of autonomy to balance the tradeoff between delegation benefits and agency
costs. We conduct the empirical analysis using Investment Climate Survey data from China.
Our results show that: (i) firm’s use incentive compensation is negatively associated with
general manager’s investment decision autonomy but positively associated with labor
decision autonomy; (ii) general manager’s investment decision autonomy dampens, while
labor decision autonomy boosts firm performance.
♣ I benefit from discussions with Francine Lafontaine, Uday Rajan and Katherine Terrell. I would like to acknowledge the World Bank Enterprise Survey Unit for allowing me to use the Investment Climate Survey data. All remaining errors are my own.
1
“Any effect on incentives is contingent upon the degree of autonomy that managers enjoy. After all, even if
incentives were given, if these managers found their decision making powers substantially constrained, one
would not expect any incentive scheme to have much effect.” --- Shahid Yusuf et al. (2006)
1. Introduction
Recently, there has been a significant increase in research interest on the organizational
structure of the firm. Despite numerous theoretical frameworks modeling delegation and
incentive, empirical evidence is relatively limited. This gap is primarily due to the lack of the
kind of detailed data or even of the convincing measures for delegation and incentive that
allow comparisons across firms.
The 2003 Investment Climate Survey conducted by the World Bank in China
spearheaded using some survey instruments trying to gain insights into firm’s organizational
structure. We are fortunate to have information on general manager’s degree of decision
autonomy and the incentive compensation structure in the firm. We use this dataset to
investigate the relationship between incentive and autonomy, as well as their effects on firm.
We aim to provide empirical evidence on how incentive and decision autonomy interact with
each other and how this imply for firm performance. As Mookherjee (2006) observed that
“One hopes that both theory and empirical datasets regarding these organizational attributes
can be developed interactively, permitting better understanding of their productivity
implications, and how they respond to changes in market competition or information
technology.”1 We hope to contribute to the literature along this line by providing empirical
evidence on the determinants and effects of managerial autonomy and incentive.
Previous literature on principal-agent model and contract is replete with what is the
best way to design an incentive compatible contract, therefore often ignoring autonomy.
However, Shahid Yusuf, Dwight H. Perkins, and Kaoru Nabeshima remarked that “Any
effect on incentives is contingent upon the degree of autonomy that managers enjoy. After
all, even if incentives were given, if these managers found their decision making powers
1 Mookherjee (2006) pp.388.
2
substantially constrained, one would not expect any incentive scheme to have much effect.”2
In line with this, we view their relationship as: It is necessary to have managers to enjoy
decision autonomy for any incentive contract to exert effect on firm outcome while the
converse is not necessarily true.
It is no doubt that managers’ autonomy of decisions is extremely important. Gerard
Fairtlough, former CEO of Shell and author of “The Three Ways of Getting Things Done:
Hierarchy, Heterarchy and Responsible Autonomy in Organizations” developed a famous
“Triarchy” theory. Namely, all organizations use a mixture of these three ways, but the
proportions can differ widely. At present, hierarchy is usually considered essential for all
organizations. Heterarchy and responsible autonomy are often misunderstood or neglected,
but they are equally important.
We all have personal experiences of how autonomy influences our behaviors in
certain ways. But it is hard to define what autonomy is. Boot and Thakor (2003) noted:
“Managers sometimes refer to it (autonomy) as ‘elbow room’, the independence to be able to
alter operating decision and change strategic direction when circumstances change.” In this
paper, we opt for a loosely popular meaning of autonomy instead of a rigorous definition.
The decision autonomy of manager is related to decentralization or delegation; however, it
has more of personal dimension to it.
The first goal of this paper is to analyze the relationship between autonomy and
incentive, especially to see how delegation of autonomy feeds back on incentive contract.
Secondly, we will examine general manager’s decision autonomy on firm performance.
Economists have long documented the huge differences in firm performance within
narrowly defined sectors, (see, for example, Bartelsman and Dhrymes (1998)) but tend to
neglect the effects of organizational characteristics in trying to explain the causes of the
performance differential. We believe addressing these two questions can shed further lights
on our understanding of determinants and consequences of firm’s organization choices.
The rest of the paper is structured as follows: Section 2 discusses the relevant
literature. We set up a simple model in section 3 to motive our empirical work. We describe
our data and some background information in section 4. Regression analysis is performed in
section 5 and section 6 concludes.
2 Yusuf et al (2006) pp.175
3
2. Motivation and Literature Review
The idea of manager’s decision autonomy comes directly from trade-off of costs and
benefits of decentralization. Decentralization is pivotal to the success of an organization. As
Hayek (1945) noted “… the ultimate decisions must be left to the people who are familiar
with these circumstances, who know directly of the relevant changes and of the resources
immediately available to meet them.” Modern firms especially need decentralization to
ensure its rapid adaptation to changes in the particular circumstances of time and place.
Hayek advocated decentralization on the ground that everybody has his or her
specialized knowledge and decentralization can make the best use of the knowledge. This
notion resonates with the idea that board of directors (the principal) hires a general manager
(the agent) because the general possesses management expertise.
Aghion and Tirole (1997) suggest two other benefits of decentralization.
Decentralization can increase agent’s initiative or incentive to acquire information, and
facilitate agent’s participation in the contractual relationship. They model the allocation of
real authority in the context of project choices where principal and agent each has his/her
own preferred project. They show that delegation is more likely for decisions that are
relatively unimportant for the principal. They also find that large span of control, urgency of
decision, and multiple principals tend to increase an agent’s real authority.
On the other hand, decentralization involves a costly loss of control for the
principal. This cost is generally referred to as the “agency cost”. In their seminal paper,
Jensen and Meckling (1976) define agency cost as the sum of the monitoring expenditure by
the principal, the bonding expenditure by the agent and the residual loss. 3 Agency conflict
between owner and manager arises usually from the manager’s tendency to appropriate
perquisites out of the firm’s resources for his own consumption.
As a result of this tradeoff, decentralization does not imply a full transfer to decision
rights from the principal (owner) to the agent (manager). In the language of Aghion and
Tirole (1997), “formal authority, need not confer real authority, that is, an effective control
over decisions, on its holder … a principal who has formal authority over a decision (or
activity) can always reverse her subordinate’s decision…”4
3 Please refer to Jensen and Meckling (1976) for a detailed explanation for this characterization. 4 Aghion Philippe and Jean Tirole (1997), pp.2.
4
Using whether different units of the firm are organized into “profit centers” as
measure of decentralization, Acemoglu et al. (2007) show that firms closer to the
technological frontier, firms in more heterogeneous environments and younger firms are
more likely to choose decentralization. These predictions are confirmed by empirical results
using datasets of French and British firms in the 1990s.
Bloom and Van Reenen (2006) conduct an innovative survey tool, collecting various
management practices data including operations, monitoring, targeting and incentives from
four countries the United States, Germany, UK and France. They establish that measures in
better management practices are strongly associated with superior firm performance in terms
of productivity, profitability, Tobin’s Q, sales growth and survival rates.
Using a detailed database of managerial job descriptions, reporting relationships, and
compensation structures in over 300 large U.S. firms, Rajan and Wulf find that CEO’s span
of control is increasing, authority is pushing down the organizations and long term incentive
is spreading in the organization.
As we can see, most empirical literature looks at multidivisional firms where
incentives and decentralization are commonly observed. Therefore, the fact that our dataset
contains many observations of small and medium enterprise can complement these studies.
China is a great case in studying the relationship between governance and
performance. Since the late 1990s, China has deepened its economic reform to “emphasize
the institutional innovation of enterprises”. The strategy is to establish a modern corporation
system featuring “clearly established property rights, well-defined power and responsibility,
separation of enterprise from government, and scientific management”.5 In 1999, the
Decision on Several Important Issues Regarding Reform and Development of State-Owned
Enterprise (SOEs), adopted at the Fourth Plenary Session of the 15th Central Committee of
Communist Party of China (CCCPC) in 1999 emphasized that “corporate governance
structure, which can establish checks and balances between the owner and the manager, is
the core of the corporate system and required all corporatized SOEs to establish effective
corporate governance. Meanwhile, the non-state sector also gained momentum in
development through favorable market environment. Many private firms improved
governance structure by forming partnership or alliance with Foreign Invested Enterprises
(FIEs) or benefiting from the spillovers of the presence of FIEs. The reform has greatly 5 See Wu (2005), pp154.
5
improved the status of corporate governance in China, enabling us to apply principal-agent
model to analyze its costs and benefits.
3. Tradeoff between Costs and Benefits of Autonomy
We outline a principal agent model that introduces a trade-off between benefits of delegating
decision autonomy to a better informed agent with expertise and the costs of having that
agent being able to use autonomy to extract resources from the principal. In this paper, we
view autonomy as the optimal degree of decentralization to balance the benefits and costs of
the agency relationship between firm owner and general manager.
The population is composed of two types of agents, of which s share are virtuous
and 1-s share are egoist6. We assume that virtuous agents always behave, i.e. serve the firm in
the best interest of the principal. Egoist agent has a tendency to extract resources from the
firm.
The principal decides whether to control the decisions or to delegate to the agent. If
the principal controls the decision, he produces less than when fully delegating to the agent.
However, if he controls, he can find out whether the egoist agent is extracting the resources
or not; if he finds that the agent is extracting, he can save these resources and further punish
the agent. The compensation contract is either a fixed wage part or a fixed wage plus an
incentive pay linked to the performance of the firm.
We use the following notations and make some assumptions:
- Y is the output when the manager has full decision autonomy;
- y is the output when the owner gives no autonomy to the general manager, i.e. the owner
totally controls all decisions.
- Y > y since manager has superior information and knowledge, so Yy
measures the
delegation benefits;
- b is the share of sales revenue given to manager as the incentive pay;
- a is the share of sales revenue extracted by the egoist manager if he gets full autonomy, so a
measures the potential loss of delegation;
6 This setup is closely related to Carlin and Gervais (2007)
6
- w is the basic wage, which will be given to the manager irrespective of whether the
principal gives how much autonomy to the agent and is equal to agent’s reservation wage;
- The principal maximizes expected payoffs.
In the full information case when the principal can distinguish virtuous agents and
egoist agents, he will give a fixed wage contract equal to his reservation wage and delegate
fully to the agent when facing the virtuous agent.
In case the agent is an egoist agent, we can write out the model in the form of
normal form game as follows:
Principal
Delegate (γ ) Control (1- γ )
Behave ( ) Δ w+bY, (1-b)Y – w w+by, (1-b)y - w Egoist Agent
(1-s) Extract(1- ) Δ w+(a+b)Y, (1-a-b)Y - w w, y - w
If we assume that y > (1-a-b)Y, there is no pure strategy Nash Equilibrium for this
game. If the principal expects the agent to be virtuous, he would fully delegate the decisions
to the agent. But if the principal expects the agent to be egoist, he would choose to control
the decisions to save the resources that would otherwise be extracted by the agent.
One way to understand the mixed strategy is that principal cannot commit to how
much autonomy to delegate to the agent ex ante, but the principal always has the privilege to
overrule the agent when he thinks that the agent is not virtuous. In a mixed strategy Nash
Equilibrium, agent chooses be virtuous with probability Δ and the principal chooses to
delegate with probabilityγ .solving for the mixed strategy Nash Equilibrium, we get:
* (1 )y a b YaY by
− − −Δ =
+
1 (1 ) Ya by
Ya by
− − −=
+
* byaY by
γ =+
bYa by
=+
We can perform simple comparative statics on the equilibrium strategies.
7
*
0bγ∂
>∂
indicates that more incentive pay is associated with more decision autonomy
of the agent; *
0aγ∂
<∂
indicates that more loss from extraction is associated with less autonomy;
*
0b
∂Δ>
∂indicates that agent is more likely to behave given more incentive;
*
0Yy
∂Δ<
∂and
*
0Yy
γ∂<
∂ indicate that agent is more likely to choose to extract if he can
bring more benefits than the principal himself, however, the autonomy the agent gets is
negatively associated with the delegation benefits.
When the principal cannot differentiate the virtuous agent from an egoist agent, he
will choose between contract 1: ( 1, w)γ = and contract 2: *( ,by w baY by
γ =+
, )
w
Y* w
.
Under contract 1, the expected payoff for the principal is: 1 ( ) (1 )[(1 ) ] (1 )E s Y w s a Y w a sa Yπ = − + − − − = − + −
Under contract 2, the expected payoff is: 2 * * *[ (1 ) (1 ) (1 ) (1 )(1 )(1 ) ]E s b Y s b Y s a bπ γ= − + − Δ − + − −Δ − −
* *(1 )[ (1 ) (1 ) (1 ) (1 )(1 ) ]s b y s b y s yγ+ − − + − Δ − + − −Δ −
Clearly, if s is sufficiently big, 1Eπ > 2Eπ , we should expect to see contract 1
more often, i.e. full autonomy with fixed wage. Here comes our proposition 1:
Proposition 1 If s is big enough, we should expect more autonomy to be
associated with fixed wage without incentives.
Intuitively, if the population is mostly composed of virtuous people, then the
principal should give full autonomy but no incentive contract to the agent.
If s is small, then 1Eπ < 2Eπ we should expect principal to offer contract 2. Then
the principal’s optimization becomes: 2
bMax Eπ * * * * *(1 ) (1 )(1 )Y a b a sa a s y sb b sγ γ= − − + Δ + + Δ + − − −Δ + Δ*b
The first order condition is:
8
2 ** * *[ (1 ) (1 )]E Y a b a sa y sb b s
b bπ γ∂ ∂
= − − + Δ + Δ − − −Δ +∂ ∂
*Δ
* * *
* * *( 1 ) (1 ) ( )Y a as y s b s sbb b b
γ γ∂Δ ∂Δ ∂Δ ∂Δ+ − + + + − − −Δ − + Δ +
∂ ∂ ∂ ∂
**
b
We can solve for the optimal b from the above equation, but we are more
concerned with the relationship between b and *γ . We can use the Implicit Function to
see how b varies with *γ 7. This exercise gives us the following prediction:
Proposition 2 If the delegation benefit is sufficiently big (i.e. Yy
is big) and the
cost of extraction is small (i.e. a is small), more decision autonomy is negatively
associated with incentive compensation.
Proposition 2 states that, in case of big delegation benefits but little agency cost, the
principal trust the agent more, but gives less incentive. In other words, the principal runs the
loss extracted by the agent but may reap big delegation benefits.
As for the effect of autonomy on the principal’s expected payoff, we have
Proposition 3 The effect of autonomy on the firm performance is
indetermined: Full autonomy can be associated with either higher outcome (case 1)
or lower outcome (case 2).
In the next section, we will take the above three predictions as well as other
hypotheses identified in the literature to the data.
4. Data and Descriptive Statistics
We are fortunate to have the Investment Climate Survey conducted by the World Bank in
China in 2003. It surveys 2,400 firms in 17 cities from 14 provinces and one municipality
directly under the jurisdiction of central government (Chongqing)8. 12 cities are capital cities
of each province and the remaining cities are also urban centers of its own province.
This survey is a stratified random survey on both manufacturing and service sectors
including “Garments and leather products”, “Electronic equipment”, “Electronic parts”, 7 The result can be obtained upon request. 8 The sample is composed of 100 firms from Benxi, 150 from Changsha, 100 from Dalian, 150 from Harbin, 100 from Jiangmen, 150 from Changchun, 100 from Guiyang, 100 from Hangzhou, 150 from Kunming, 150 from Nanchang, 100 from Shenzhen, 100 from Wenzhou, 150 from Wuhan, 150 from Xi’an, 150 from Lanzhou, 150 from Zhengzhou and 150 from Nanning.
9
“Household electronics”, “Auto & auto parts”, “Information technology”, “Accounting-
banking and financial services”, “Advertisement and marketing”, “Business services”, “Food
processing”, “Chemical products and medicine”, “Biotech products and Chinese medicine”,
“Metallurgical products”, and “Transportation equipment”.
It collects information on various kinds of firm information including general
characteristics, innovation and technology, certification of products or services, relations
with clients, suppliers and governments, sales and supplies, labor, infrastructure, trade,
finance and taxes etc. One section of particular interest is “Information about the general
manager and board of directors” provide us with information on organizational
characteristics of the firm. Questions in this section provide us with information on general
managers and board of directors. Information on general manager includes the education,
the tenure, the party position and how the general manager was appointed etc. In addition,
we know the ratio of wage of general manager to that of middle-level manager and whether
manager has any incentive compensation plans linking to firm performance.
We obtain the measure of manager autonomy from three questions of autonomy of
general manager on production decisions (output, quantity, quality, investment, and so on),
autonomy of investment decision and investment on labor flexibility (hiring, firing and
wage). The answers to this question is structured on a percentage basis, specifically, eight
category of 100%, 90-99%, 80-89%, 70-79%, 60-69%, 40-59%, 20-39% and 0-19% with
each corresponding to a score from 8 to 19, thus higher scores imply greater degree of
autonomy. In particular, we focus on the autonomy of investment and labor decisions since
many service firms do not answer the production autonomy the same way service firms
answer this question.
We also have general manager’s autonomy directly measured in different aspects,
which allows us to conduct a breakdown analysis. Previous literature often uses proxies for
manager’s autonomy. For example, Acemoglu et al. (2007) uses “whether your firm is
organized into different profit centers” to measure decentralization as they argue that when a
firm organizes into profit centers a manager is responsible for the profits of the unit she
manages. In general the profit center manager is given considerable autonomy to make
decisions on the purchase of assets, hiring of personnel, setting salary and promotion
schedules and managing inventories. Their measure depends crucially on the scale of the 9 In the original survey, each category corresponds to a score from 1 to 8.
10
firm. Small firms’ manager autonomy will be excluded from their measures. Our measures
do not have this problem, however, this is a very subjective measure of manager’s autonomy
and the category is not very well defined.
Table 1 presents summary statistics of the two measures of autonomy. We can see
that, on average, general managers obtain more autonomy in labor decisions than in
investment decisions.
In general, firms give to its general manager more autonomy on labor than on
investment decisions. It is interesting to compare the means of manager’s decision autonomy
for the group of firms with incentive compensation and the group without it. Table 1c
shows that general manager with an incentive compensation contract have more autonomy
in investment decision but less autonomy in labor decisions.
Table 1 Here
5. Empirical Investigation
5.1 Baseline Specification
5.1.1 Effects of Managerial Autonomy and Incentives
We assume that firm performance is a function of firm level, industry level variables,
augmented by general manager’s incentive and autonomy.
ijc i i j j ijcY X X aut incβ β ϕ η= + + + +ε (1)
Where is the output measure for firm i in industry j at city c. We primarily use
sales revenue but also use sales growth and sales per employee.
ijcY
iX is a vector of firm
characteristics, including labor and capital, jX is a vector of industry characteristics, aut
includes investment autonomy and labor autonomy of the general manager, inc is the
incentive compensation for the general manager.
In the baseline regression, the dependent variable is for year 2002 and the logarithm
of labor and capital are for year 2000, which is an attempt to prevent the most obvious form
of reverse causality. All omitted factors are captured by the error term, which we assume to
be normally distributed.
5.1.2 The Relationship between Managerial Autonomy and Incentive
11
In this part, we will document a number of correlations motivated by related literature and
by the propositions presented in section 3.
ijc i i j j ijcinc aut X Xα φ β β ε= + + + + (2)
5.1.3 The Determinants of Managerial Autonomy
In searching for explanatory variables for managerial autonomy, we include three sets of
covariates, industry level, firm level and personal level.
Consider the following model for general manager’s autonomy determination:
ijc i i j j ijcaut per X Xα φ β β ε= + + + + (3)
Where i denotes firm, j denotes industry and c denotes city. iX is a vector of firm
level variables, include firm size, firm’s capital intensity, firm’s age, and firm’s distance to
frontiers and its competition environment. jX is a vector of industry characteristics
including the heterogeneity of environment measured by the sales growth rate differential in
the industry. Acemoglue et al. (2007) document that younger firms, firms in more
heterogeneous environment and firms far away from the technological frontier use more
decentralization than others. The set of personal characteristics include general manager’s
education, how long the general manager has held this position, the tenure of the general
manager and whether the general manager’s wage is paid annually.
5.2 Empirical Strategy
We aim to estimate equations (1) (2) and (3) together. This of course argues for the use of
some kind of simultaneous equations system estimator. Because the three variables of
interest, manager’s incentive and two measures of manager’s autonomy are potentially
endogenous, we will opt for Three Stage Least Squares (3SLS) estimation. In the 3SLS
estimation, all the personal, firm industry and city level variables will be treated as
instruments. We also use system Ordinary Least Squares (OLS) estimation for comparisons.
All regression control of industry and city fixed effects.
5.3 Regression Results
12
Table 2 presents our baseline regression results using log of sales revenue as dependent
variable. The first four columns estimate using system OLS and the last four columns
estimate using 3SLS.
When we look at the performance estimation, we find that that general manager’s
investment autonomy is negatively associated with sales revenue, but labor autonomy is
positively associated with sales revenue. Comparing OLS results with 3SLS results, we find
that the magnitude the coefficients on the two autonomy measures increase significantly. In
terms of economic significance, on average, increasing general manager’s investment
autonomy by 10% is associated with a decrease of firm’s sales revenue by 0.8 standard
deviations; increasing general manager’s labor autonomy by 10% is associated with an
increase of firm’s sales by 1.5 standard deviations. Besides, the coefficient on incentive
switch signs from positive to significantly negative.
The second and sixth columns estimate the effects of managerial autonomy on
incentive compensation. We find that manager’s investment autonomy is negatively
associated with incentive, but manager’s labor autonomy is positively associated with
incentive compensation. This partly confirms our proposition 1 and 2.
Estimation on the determinants of managerial autonomy tells us that general
mangers with a longer tenure, with more managing experiences tend to have more autonomy
in both dimensions. Younger firms generally give more autonomy especially labor autonomy
to their managers. General managers who are hired outside the firm usually get more
autonomy. If the firm is further away from the industry productivity frontier, general
managers are expected to have more autonomy on investment decisions.
Table 2 Here
Table 3 reports 3SLS results using sales per employee and sales growth rate as
dependent variables. In general, these two regressions display similar patterns of results: the
coefficient on incentive is negative, the coefficient on investment autonomy is also negative,
but the coefficient on labor autonomy is positive. All of them are statistically significant at
1% level. In terms of economic significance, on average, increasing general manager’s
investment autonomy by 10% is associated with a decrease of 1500 yuan of sales per
employee (the sample mean is 4300 yuan) and with a decrease of annual sales growth rate by
about 20%; increasing general manager’s labor autonomy by 10% is associated with an
13
increase of sales revenue per employee by 2500 yuan and with an annual sales growth rate by
about 40%.
Table 3 Here
The huge increase in the magnitude of the coefficients on two autonomy measures
calls into question of the selection of instruments: the instruments are potentially correlated
with our dependent variables. The effects of instruments on firm performance are absorbed
by the two autonomy measures. Finding better instrument will be our future work.
5. 4 Robustness Checks
To check the robustness of our results, we first run separate regressions on firms with or
without incentive structure. The estimation results are reported in table 4. For all three
dependent variables, the overall pattern remains the same, however, we find that the
magnitude of the coefficients on the two autonomy measures is bigger for firms without
incentive compensation for the general manager. This result is consistent with the
implication from the previous story: in case of fixed wage contract, if the virtuous agents
bring better outcome to the firm, but egoist agents extract more resources in case of more
autonomy. The two contrary effects magnify the impact of autonomy.
Table 4 Here
Previous related literature mostly conducts empirical analysis based on firm from
manufacturing sector. In our sample, about one fourth of firms are from service sector, thus
enabling us to check whether the results differ across sectors. Table 5 presents the
breakdown estimation results. Compared with our baseline results, we lose significance levels
on investment autonomy coefficients on service sector regressions with the dependent
variables being sales growth rate and sales per employee and on labor autonomy coefficients
on service sector with the dependent variable being sales growth rate. The results from
regressions on manufacturing sectors are robust throughout.
Table 5 Here
On average, general managers in SOE receive less autonomy than managers in other
types of firms10 as they are to some extent constrained by the government. Thus it is
interesting to see if autonomy for managers in SOE has different effects from managers in
10 The sample average for investment autonomy is 5.7 and the sample average for SOE manager is 5.2; The sample average for labor autonomy is 6.5 and the sample average for SOE manager is 5.8.
14
other types of firms. We can see from table 6, consistent with our conjecture that autonomy
measures indeed have a bigger impact on firm performance.
Table 6 Here
Overall, as we can see, our results therefore are fairly robust to different sectors and
different ownership forms, although the baseline results are a little bit more driven by firms
from manufacturing sectors and Non-SOE firms.
6 Making Sense of the Regression Results
The previous analysis seems to tell a consistent story that general manager’s investment
autonomy does harm to the firm, but labor autonomy benefits the firm. In this section, we
try to provide some intuition for this result.
It is now cliché to say that managers tend to over invest. As Dessein (2002) note “It
is well accepted, for instance, that managers have a propensity to cause their department,
division or firm to grow beyond the optimal size, i.e. they are empire builders and undertake
too many investments. They further seldom take externalities on future managers into
account and, hence, are excessively oriented towards short-term profitability and results.”
Hennessey and Levy (2002) develop a unified model to test various hypotheses of manager
investment distortions. They find strong empirical evidence in favor of empire building
hypothesis, with investment being highly correlated with CEO entrenchment. Shleifer and
Vishny (1989) describe entrenchment in terms of manager-specific investments, where
manager entrenches himself by investing excessively in assets that are complementary to
manager’s skills thus reducing the probability of being replaced. The two papers do not link
manager’s investment decisions to firm performance; however, it should not be surprising
that if left unchecked, manager’s tendency to over invest can jeopardize the firm
significantly.
Bodmer (2002) studies the relationship between the manager’s autonomy and
productivity, using a survey of 769 SOEs during 1980–94 in China. He uses four dummy
variables: manager’s performance contract, the share of contract workers, manager’s output
autonomy, and manager’s autonomy in hiring/firing workers. The empirical results show
that the first three variables have significant and positive effects on productivity, although
employment reform yields a negative but insignificant outcome, suggesting that labor
decision has little effect on productivity.
15
Compared with investment decisions, general managers do not have too much
incentive to distort firing/hiring decisions as these do not bring as big personal benefits as
investment decisions.
7 Concluding Remarks
This paper tries to achieve two goals: First is to establish the relationship between firm’s use
of incentive compensation contract and its general manager’s decision autonomy and the
second is to examine the effects of managerial autonomy on firm performance.
We develop a simple principal agent model where the principal makes use of
incentive compensation and delegation of autonomy to maximize the benefits of delegation
and minimize the agency costs. We show that, firms may use less incentive compensation
combined with more decision autonomy to best exploit the agent.
We use Investment Climate Survey data from China to conduct the empirical
analysis. This survey contains direct measures of general manager’s decision autonomy and
has information on general manager’s compensation contract. We estimate a system of
equations in which we allow manager’s decision autonomy and firm’s use of incentive
contract to be endogenous. Our results show that firm’s use of incentive compensation is
negatively associated with general manager’s investment autonomy but positively associated
with labor autonomy. General manager’s personal characteristics can explain much variation
in his decision autonomy. We also find that general manager’s investment autonomy hurts
while labor autonomy benefits the firm across different performance measures. This result is
much driven by firms in manufacturing sectors and Non-SOE firms.
In the future, I plan to better tie the theoretical framework with empirical analysis,
especially how to incorporate both dimensions of decision autonomy into the principal agent
model. Besides, more carefully work needs to be done in finding exogenous instruments for
managerial autonomy.
16
References
Acemoglu, Daron, Philippe Aghion, Claire Lelarge, John Van Reenan and Fabrizio Zilibotti
(2007) “Technology, Information and the Decentralization of the Firm”, forthcoming,
Quarterly Journal of Economics.
Aghion, Phillippe and Jean Tirole (1997) “Formal and Real Authority in Organizations”,
Journal of Political Economy, vol. 105, no. 1 pp. 1-29.
Bartelsman, E. and Dhrymes, P. (1998), “Productivity Dynamics: US Manufacturing plants,
1972-1986”, Journal of Productivity Analysis, Vol. 9, pp.5-34.
Bertrand, M. and Schoar, A. (2003) “Managing with Style: The Effect of Managers on Firm
Policies”, Quarterly Journal of Economics, pp.1169-1208.
Bloom, Nick and John Van Reenen (2006) “Measuring and Explaining Management
Practices across Firms and Countries,” NBER Working Paper 12216.
Bodmer, Frank (2002) “The Effect of Reform on Employment Flexibility in Chinese SOEs,
1980-1994”, Economics of Transition, vol. 12, no. 3 pp. 637-658.
Boot, Arnoud W. A. and Anjan Thakor (2003) “The Economic Value of Autonomy”,
mimeo.
Carlin, Bruce, and Simon Gervais (2007) “Work Ethic, Employment Contracts, and Firm
Value,” Journalof Finance, forthcoming.
Dessein, Wouter (2002) “Authority and Communication in Organizations”, Review of
Economic Studies, Vol. 69, No. 4. pp. 811-838.
Fairtlough, Gerard (2005) The Three Ways of Getting Things Done: Hierarchy, Heterarchy and
Responsible Autonomy in Organizations, Triarchy Press.
Fudenberg, Drew and Jean Tirole (1991) Game Theory, Boston: MIT press.
Mookherjee,Dilip (2006) “Decentralization, Hierarchies, and Incentives: A Mechanism
Design Perspective”, Journal of Economic Literature, Vol. XLIV, pp. 367-390.
Hayek, F.A. (1945) “The Use of Knowledge in Society,” American Economic Review, Vol. 35
No. 4, pp. 519–530.
Hennessy, Chris and Amnon Levy (2002) “A Unified Model of Distorted Investment:
Theory and Evidence”, mimeo.
Jensen, Michael C. and William H. Meckling (1976) “Theory of the Firm: Managerial
Behavior, Agency Costs and Ownership Structure”, Journal of Financial Economics, Vol. 3, No.
4 (1976) pp.305-360.
17
McGoldrick ,Peter and Patrick Paul Walsh (2004) “Reforms and Productivity Dynamics in
Chinese State-Owned Enterprises”, IZA discussion paper no. 1201.
Rajan, Raghuram G.and Julie Wulf (2006) "The Flattening Firm: Evidence from Panel Data
on the Changing Nature of Corporate Hierarchies." Review of Economics and Statistics Vol. 88,
No. 4: 759-773
Shleifer, Andrei and Robert Vishny (1989) “Management Entrenchment: The Case of
Manager-specific Investments,” Journal of Financial Economics, Vol. 25, No. 1 pp. 123-139.
Wagner, F. Alexander, Nolan H. Miller and Richard J. Zeckhauser (2006) “Screening
Budgets,” Journal of Economic Behavior and Organization, Vol. 61, No. 3 pp. 351-374.
Wu, Jinglian (2005) Understanding and Interpreting Chinese Economic Reform, Mason, Ohio:
Thomson/South-Western.
Yusuf, Shahid, Kaoru Nabeshima and Dwight H. Perkins (2006) Under New Ownership:
Privatizing China's State-owned Enterprises, Stanford University Press.
Observations 389 1281 389 1281 389 1281 Standard errors in parentheses
* significant at 10%; ** significant at 5%; *** significant at 1%
26
Appendix A: Definition of variables
LnL00 Logarithm of average number of worker in 2000 LnK00 Logarithm of total fixed asset value for year 2000 LnY02 Logarithm of value of total sales Sales growth rate
Ln(Y2002/Y2000)
Investment autonomy
Score 1 to 8 correspond to 0-19%; 20-39%; 40-59%; 60-69%; 70-79%; 80-89%; 90-99% and100%.
Labor autonomy
Labor flexibility (hiring, firing and wage). Score 1 to 8 correspond to 0-19%; 20-39%; 40-59%; 60-69%; 70-79%; 80-89%; 90-99% and100%.
Incentive Dummy variable to question “Does the General Manager has any incentive plans linking his/her income to firm performance, Yes/No”
GM tenure What is the tenure for the general manager? (Measured in years) GM inside Dummy variables whether the General Manager is from the firm or hired from outside GM Annual Pay
Dummy variable to question “Is the General Manager’s wage paid annually? Yes/No”
GM experience
“How many years had the General Manager held this position?”
GM education
Scores 1 to 6 corresponding to “no education; primary school education; secondary education; high-school education; undergraduate education and postgraduate education
Firm age Logarithm of firm’s age Distance to
frontier Firm’s solow residual to the solow residual of the best firm in its industry
Industry heterogeneity
The range of sales growth rate between 2000 and 2002