Excerpt
Table of Contents
List of Figures
List of Tables
List of Appendix
List of Abbreviations
1. Introduction
1.1. Rationale and Motivation
1.2. Project and Layout
2. Literature Review of Empirical Literature and Theoretical Discussion
2.1. Management Practices and Firm Performance
2.2. Organizational Structure and Firm Performance
2.2.1. Difficulties
2.2.2. Empirical Literature
2.2.3. Theoretical Discussion: Contingency Theory and Trade Off Decentralization
2.3. Organizational Structure, Management Practices and Firm Performance
2.3.1. Difficulties
2.3.2. Empirical Literature
2.3.3. Hypothesis based on Principal Agent Theory
2.3.4. Hypotheses based on Change Management Literature
3. Data Collection and Method
3.1. Measuring Management Practices
3.2. Measuring Organizational Structure
3.2.1. Overview
3.2.2. Plant Autonomy
3.2.3. Worker Autonomy
3.2.4. Span of Control
3.2.5. Number of Hierarchical Levels and Number of Levels Changed in the Last Three Years
3.2.6. Firm Slope and Plant Slope
3.3. Measuring Firm Performance
3.4. Measuring Competition
3.5. The Interview Process: Obtaining the Truth
3.6. Noise in the Signal
3.6.1. Control Variables
3.6.2. Second Interviews
4. Data Description
4.1. Sample Description and Criteria
4.2. Country Comparisons
4.2.1. Management Practices
4.2.2. Organizational Structures
4.3. Industry Comparisons
4.4. Explanations and the Role of Regulation
5. Summary of Hypotheses and Econometric Modeling
6. Results
6.1. The Impact of Management Practices on Firm Performance
6.1.1. Hypothesis 1
6.1.2. Discussion and Managerial Implication
6.2. The Impact of Organizational Structure on Firm Performance
6.2.1. Hypotheses 2 - 5
6.2.2. Discussion and Managerial Implication
6.3. The Impact of Management Practices and Organizational Structure on Firm Performance
6.3.1. Hypotheses 6 - 9
6.3.2. Discussion and Managerial Implication
7. Conclusion
7.1. Limitations
7.2. Contribution and Future Research
Literature
List of Figures
Figure 1: Graphical Display of Collected Organizational Structure Variables; (collected variables are in bold)
Figure 2: Scatter Plot First Interviews versus Second Interviews - Considering Different Sites (totaling 127 second interviews)
Figure 3: Summary of the Effects of Noise Controls and the Role of the Second Interview (totaling 3518 firms)
Figure 4: Number of Interviews per Country - 3518 Interviews Total in Twelve Different
Figure 5: Mean of Management Score by Country - Homogenous Company Size (totaling 3076 firms; 50 =< employees <= 2500)
Figure 6: Histogram of Management Scores, by Country (totaling 3518 firms)
Figure 7: Mean of Management Score, by Country - (totaling 2697 firms; 50 =< employees >= 2500 & management score > 2)
Figure 8: Mean of Plant Autonomy Score, by Country (totaling 3076 firms, 50 =< employees >= 2500)
Figure 9: Mean of Worker Autonomy Score, by Country (totaling 3076 firms, 50 =< employees >= 2500)
Figure 10: Mean of Slope, by Country (totaling 3076 firms, 50 =< employees >= 2500)
Figure 11: Mean of Management Scores, by Industry (SIC-2) (totaling 3076 firms, 50 <= employees >= 2500)
Figure 12: Mean and Variation of Management Score by Industry (SIC-3 digits) (totaling 3076 firms, 50 =< employees >= 2500)
Figure 13: Country Relative Lean Management Minus Talent Management Score vs. the Employment Rigidity Index, data from World Bank (2006), p. 110 ff
Figure 14: Scatter Plot Management Score vs. Logarithm Sales including Fitted Values - Lowess Smoother (totaling 3518 firms)
Figure 15: Twoway Scatter Plot Plant Autonomy vs. Logs (Sales) (totaling: 3518 firms)
Figure 16: Mean of Management Score, by Number of Hierarchical Levels Changed within the Last Three Years (totaling 3518 firms)
List of Tables
Table 1: Dimensions and Areas of the Management Interview
Table 2: Regression Coefficient and Significance of All Eighteen Management Questions (totaling 3518 firms)
Table 3: Control Variables
Table 4: Analyzed Countries
Table 5: Distribution of Management Scores
Table 6: Mean and Standard Deviation per Country for Homogenous Company Size (totaling 3076 firms; 50 <= employees <= 2500)
Table 7: Top Five and Bottom Five Industries, Regarding Mean of Management Score
Table 8: Structured Summary of Hypotheses and Matched Research Models
Table 9: Regression Results for Hypothesis 1
Table 10: Regression Results for Hypothesis 2
Table 11: Regression Results for Hypotheses 3 - 5
Table 12: Regression Results for Hypothesis 6
Table 13: Regression Results for Hypothesis 7
Table 14: Regression Results for Hypothesis 8
Table 15: Regression Results for Hypothesis 9
List of Appendix
Appendix 1: Management Interview Guide, Bloom (2006), p. 43ff
Appendix 2: HR and Organizational Structure Interview Guide
Appendix 3: Summary of All Variables (obtained and used in Stata) - Including Definitions, Range and Mean
Appendix 4: GNI per Capita (current US$) in 2005, Data from Worldbank (2006b)
Appendix 5: Number of Interviews per Country (back up data for Figure 4)
Appendix 6: Obtained Management Score vs Selfscore of the Interviewee (totaling
Appendix 7: Mean of Lean Management, by Country (totaling 3076 firms, 50 =< employees >= 2500)
Appendix 8: Mean of Performance Management, by country (totaling 3076 firms, 50 =< employees >= 2500)
Appendix 9: Mean of Talent Management, by Country (totaling 3076 firms, 50 =< employees >= 2500)
Appendix 10: Data for Figure 3 (Waterfall Chart)
Appendix 11: Mean of Employees with a College Degree by Country in Percentage (totaling 3076 firms, 50 =< employees >= 2500)
Appendix 12: Twoway Scatter Plot Plant Autonomy vs. Logs (Sales) (totaling: 3518 firms)
Appendix 13: Twoway Scatter Plot Worker Autonomy vs. Logs (Sales) (totaling: 3518 firms)
Appendix 14: Twoway Scatter Plot Slope vs. Logs (Sales) (totaling: 3518 firms)
List of Abbreviations
illustration not visible in this excerpt
1. Introduction
1.1. Rationale and Motivation
“The most significant factor in America, leading to high production at low cost, is efficient management.”[1] This quotation is not from a longitudinal, empirical study, but it is stated by the British Productivity Council in the context of the Marshal Plan - the Anglo - American productivity mission in 1951. In 2005, fifty four years later, America’s Gross National Income (GNI) per capita is $43740 which is still significantly higher than the GNI per capita of most European countries.[2] See Appendix 4 for GNI per capita comparison of all interviewed countries.[3] Hence, one question might be: Is there indeed a strong link between management practices and firm performance? Further questions might be: How American management practices rank compared with European or Asian management practices? What role plays organizational structure? Are there interacted effects between management practices and organizational structure? Insights into the impact of management practices and organizational structure on firm performance are mainly based on theories.
In addition to theory, business schools around the world teach the ratification of these impacts with case studies. However, there is little evidence for the relationship between management practices, organizational structure and firm performance based on quantitative, longitudinal and international data.
Consequently, there is a strong motivation to measure management practices and organizational structure across countries and cultures to verify theoretical hypotheses with empirical data. This paper sheds some empirical light on these topics as well as on the distribution of management practices and organizational structures across countries and industries.
The lack of quantitative and longitudinal data is not surprising, since it is extremely difficult to evaluate management practices for a large number of companies. Evaluation requires both defining good management and objectively scoring firms against that standard. As a result, most of the existing evidence is from case studies and anecdotes. Case studies help to analyze different analytical questions and pursue different analytical goals.[4] However, case studies lack sufficient evidence to support general business evidence based on a representative sample.
Most of the papers about organizational structure and firm performance are from the 1970s and 1980s. Hence this offers an additional strong motivation to review this relationship in the more recent communication and information age.
This paper brings a unique data set with more than 3500 management interviews to document and analyze the influence of management practices and organizational structure on firm performance (Bloom 2010). Furthermore, the role and impact of organizational change can be obtained and compared to a strand of literature.
This paper relates to a number of strands in the literature. It builds on and extends the paper of Nick Bloom and John van Reenen (2007) “Measuring and explaining management practices across countries and industries.”[5] Furthermore, this paper is related to the paper of Marianne Bertrand and Antoinette Schoar (2003), “Managing with style: the effect of managers on firm policies“ in which the effect of individual managers on firm performance is analyzed.[6] This paper also builds on major organizational theories from Chandler, Lawrence and Lorsch and Picot.[7]
1.2. Project and Layout
The data has been obtained in a McKinsey & Company, London School of Economics and Stanford University joint project. The author of the paper was part of a team of 35 international MBA, Ph.D students and professionals from top business schools around the world. The role of the author of the paper as a management research analyst within the project was to:
- Successfully complete extensive McKinsey Training sessions on interviewing, problem solving, consulting and lean management.
- Conduct 201 in - depth interviews with top executives around the world to assess and evaluate management practices and organizational structures of interviewed firms.
- Provide input as an ongoing process on research approach, survey improvements and interview technique improvements.
The layout of this paper is as follows. Section two documents an extensive
literature review to obtain intriguing hypotheses which either verify current literature or expand current literature. Section three documents, explains and compares the data collection process and method with literature, as well as it describes all obtained variables to test the derived hypotheses of section two. In section four, this paper describes the data set on a country and industry perspective. In section five the econometric modeling is derived and explained. Section six tests the hypotheses with multivariate regressions according to the derived equations of section five. Further, section six discusses the results of the hypotheses with the literature and offers some managerial implications. Finally, section seven summarizes all of the findings and contributions of the paper in a conclusion and discusses limitations as well as interesting future research areas.
2. Literature Review of Empirical Literature and Theoretical Discussion
The following section summarizes current empirical papers related to this study on management practices, organizational structure and firm performance. It also discusses organizational theories with the goal of obtaining intriguing research hypotheses. The obtained hypotheses should either verify current literature or expand the current literature. One major contribution of this paper is that it not only analyzes the relationship between organizational structure on firm performance and management practices on firm performance, but it also analyzes interaction effects of organizational structure combined with management practices on firm performance for various hypotheses using cross country data.
2.1. Management Practices and Firm Performance
The belief that management matters for firm performance may seem obviously true. However, there is not yet much empirical evidence published on the subject. This seems to be a gap in social science. So far, only a small number of quantitative studies have been published. However, these focus on related topics, rather than on management practices directly. One of the most important papers in this field is Nick Bloom and John van Reenen’s (2007) recently published paper “Measuring and explaining management practices across countries and industries”.[8] Their paper discusses the distribution of management practices across countries and industries. The data was gathered through interviews, using the same innovative management survey tool this paper used too.[9] This approach to obtain management data is similar to the approach in this paper. Further, explanations for variations of management practices are discussed. However, the basis for the analysis is limited to only 732 companies in four countries and it does neither include organizational aspects, nor does it analyze interacted effects between management practices and organizational structure.
Another relevant longitudinal quantitative research paper is from Bertrand and Schoar (2003). Bertrand and Schoar investigate the movement of top individual managers across US firms and find that this movement significantly influences performance. They identified particular patterns in managerial decision making that suggest differences in style between managers. Their work concludes that this style is correlated to performance and that managers with higher performance receive higher compensation. However cause and effect can not be clarified in this paper, i.e. it does not state if the more successful managers led their companies to success or if successful companies attracted proven and thriving executives. In addition, they only focus on the executive level and do not consider middle management. Bertrand and Schoar (2003) also observe how aggressive managers apply financial strategies for their firms. They find that executives from earlier birth cohorts appear to be more conservative than those holding an MBA degree.[10]
Bresnahan et al. (2002) investigates the reasons for and the results of the relative demand in skilled labor. They link their analysis with data about companies’ investments in information technology. They conclude that a cluster exists in complementary changes involving workplace organization and that services and investments in IT are important components of a skill based technical change. However their primarily focus is not on the impact of management practices on firm performance.[11]
Ichniowski et al. (1997) show that human resource management practices influence productivity in seventeen US steel companies. Ichniowski’s productivity function suggests that certain modern work practices like incentive pay, teamwork, training and flexible job assignments have a positive impact on company’s productivity. The major limitation of the study is that it only takes into account work practices and performance outcomes in one country and of one very specific type of production process.[12]
Chiang Kao and HSI-Tai Hung (2005) measure management performance in their paper and correlate it to performance measures. Their process is to divide management up into production management, marketing management, financial management and HR management. The data is obtained through surveys, which are filled out by the companies on their own. This is a major critic point, since self judgment generally is not very accurate. See Appendix 6 for the data obtained for this paper, regarding the self score results for overall management. Overall a positive correlation of management practices to organizational performance is documented. However, due to the small sample of only 41 companies in only one country (Taiwan) and one industry, it is suggested by Kao and Hung to verify their outcome with cross country and cross industry data of a larger sample.[13]
None of the studies mentioned above uses data of 3391 different companies[14] in twelve countries to verify the following hypothesis:
Hypothesis 1: Firms with better management (higher management scores) perform better.[15]
What is better management? To measure management and to define good management are difficult tasks. These processes are explained in detail under section 3.1.
If Hypothesis 1 is true, it shows that the obtained management data has the correct and expected correlation with firm performance as an external validity test. This would give further credibility to the descriptive management data of the thesis and makes section 4 data description very interesting.
2.2. Organizational Structure and Firm Performance
A central theoretical question is how organizational structure affects firm performance. Organizational structure can be seen as the internal pattern of relationships, authority and communication.[16] The optimal organizational structure would be the structure that minimizes both, the sum of the production costs and the agency costs.[17] Chandler’s (1962)[18] and Williamson’s (1975) analysis of transaction costs have shown the high importance of organizational structure on firm performance.[19] Transaction cost economics belongs to the neo - institutional economics. This contradicts the neo-classical view in which the firm is seen as a “black box” and the cost of economic exchange within the organization are neglected.[20] In this neo-classical view, a firm is a production function. All coordination is done by the market through market prices. This paper is based on the neo - institutional economics view, as a majority of recent studies are as well.[21] Further reason is that costs of economic exchange within an organization seem to be obviously relevant for firm performance.
However, there are several difficulties in analyzing the effect of organizational structure on firm performance. These are discussed under section 2.2.1. Under Section 2.2.2., a strand of literature will be reviewed which analyzes the impact of organizational structure on firm performance. Under section 2.2.3., the contingency theory, as a major organizational theory which considers the environment, and the trade off of decentralization will be discussed to explain the results of the empirical literature review.
2.2.1. Difficulties
Analyzing correlations between organizational structure and firm performance with empirical data is a difficult task, due to various reasons discussed in the following. First, the specific environment for companies varies across industries and countries, which leads consequently to different needs in the organizational structure. This is discussed in more detail under section 2.2.3. Second, companies are in different life cycle stages, for which different organizational structures may be optimal and different profits are typical.[22] Therefore, for instance, a company with a certain type of organizational structure could perform better simply because they are at a more mature stage within the organizational life cycle.
In addition, to see the extent of the influence of organizational structure on firm performance these effects would need to be isolated from other extraneous influences.[23] This paper takes this into consideration. In all regressions including organizational structure variables, it is controlled for obvious drivers of firm performance, such as capital and the number of employees.
All these factors make it difficult to derive conclusions out of regressions between organizational structure and firm performance.
2.2.2. Empirical Literature
Despite the difficulties summarized above, many papers, especially in the 1970s and 1980s, analyzed the relationship between organizational structure and firm performance econometrically. A major point of interest was: Should decisions concerning distribution, pricing, marketing, sales and new product introductions be centralized, or is it favorable to decentralize decisions about product introductions, marketing, sales and hiring?
Comparisons between centralization and decentralization as the question about the locus of decision making still continues to be a field of ongoing research within the analysis of the relationship between organizational structure and firm performance.[24] The results are mixed throughout the studies. For instance, especially Williamson’s (1970) so called multi - division form hypothesis has been tested many times. This hypothesis basically states that multi - divisional organizations perform better than the unitary - form organization.[25] One of the most important differences between the two is that, the multi - divisional organization has a higher degree of decentralization of corporate decision making than the unitary - form organization. The multi - divisional organization is functionally decentralized, in contrast to the unitary - form organization, which is functionally specialized[26]. Within fourteen empirical studies, which have been reviewed for this section in this paper the results regarding the multi - divisional hypothesis and whether decentralizing is favorable vary significantly.
Steer and Cable (1978), Burton and Obel (1980), Thompson (1980), Hill (1985) and Hill and Pickering (1986) find empirical evidence that companies with high degrees of decentralized decision making perform better.[27] In contrast, Teece (1981), Harris (1983), and Cable and Yasuki (1985) report negative results for the multi - divisional hypothesis.[28] Armour and Teece (1978), Roberts and Viscione (1981), Cable and Dirrheimer (1983), Thompson (1983) and Hoskisson and Galbraith (1985) report mixed and insignificant results.[29]
Most of the studies summarized above focused on large companies, rather than on medium size companies, as this paper does. Despite this disparity, this literature review shows that the impact of organizational variables on firm performance offers mixed results in current literature. No clear and easy conclusion can be drawn out of empirical literature. The cause of the unclear empirical evidence in this area could be either that the results of data collection methods for organizational variables are not reliable and not interchangeable[30] or that there is a theoretical reason for the mixed outcome. The following section focuses on explaining this phenomenon through theoretically reasoning.
2.2.3. Theoretical Discussion: Contingency Theory and Trade Off Decentralization
Since empirical literature offers mixed results for the impact of decentralization, at both the plant and worker levels, on firm performance, as summarized above, this paper discusses the big picture of contingency theory and discusses the trade off effects of decentralization on firm performance to explain these mixed outcomes and to predict hypotheses. The contingency theory has also been used by Gupta (1997)[31] and is a major theory in organizational theory. Decentralization itself offers advantages and disadvantages; hence it is valuable to analyze it to better understand the possible impact of decentralization on firm performance. According to the contingency model by Lawrence and Lorsch (1967), there is not one best way of organizing.[32] The right organizational structure for a specific company depends on - is contingent on many internal and external determents like the specific environment for a company and the tasks the company wants to perform.[33] Consequently, organizational structure is affected by the firm’s external environment.[34]
For instance, research suggests that firms organized to deal with reliable and stable markets may not be as effective in a complex, rapidly changing environment.[35] In a more certain environment, companies tend to be more hierarchical, with centralized decision making.[36] According to Rueckert (1985), companies tend to decentralize decision making in an uncertain environment.[37]
How can an “uncertain environment” be quantified? The argument in this paper is that companies which face higher competition are in a more uncertain environment, because more competitors support a changing and uncertain environment. With higher degrees of competition, planning is more difficult, since more players on the market need to be considered. This suggests that companies with higher degrees of competition tend to use higher degrees of decentralization in decision making as shown below:
Hypothesis 2: Firms with higher degrees of competition (self reported) tend to use higher degrees of decentralization in decision making.[38]
The contingency theory sees organizational structure as endogenous to the environment. This means that the environment matters for the organizational structure. As discussed above, considering different specific environments and internal capabilities for each company in the sample data, a certain kind of organizational structure is not good or bad in general. Each structure has its specific advantages and disadvantages.
Decentralization itself offers a trade off between benefits and disadvantages.[39] A major disadvantage is the risk of duplication of information without centralized management.[40] A second related disadvantage is that the costs of mistakes may increase in the absence of specialized monitoring and in the presence of lower direct control.[41] In addition, a possible disadvantage is that decentralization reduces the benefits of economies of scale.[42] The logic behind that statement is that decentralized multitasking decreases specialization returns and makes mass production less attractive. Furthermore, from a stress level point of view, not every worker may be appropriate for higher degrees of responsibility. In this case decentralization may lead to reduced worker productivity in certain circumstances.[43]
On the other hand, there are also a range of advantages of decentralization. Decentralizing reduces costs of communication. The argument is that, once decentralized, the point where the information is processed is closer to where it is used.[44] This leads to reduced communication costs.[45] Second, monitoring costs can be reduced by decentralizing, as a result of delayering.[46] Third, decentralization accelerates the firms’ reaction to market changes. This argument supports Hypothesis 2.[47] One reason for that is that centralized companies consist of highly specialized workers. Once the environment changes, any response involves the coordination of many activities, which makes an adequate reaction too complex and slow. While decentralizing, the power of decision making is shifted downwards. This allows a quicker response, given that the coordination of multitasking employees requires less effort. In addition, it can be argued that delegating responsibility leads to greater involvement and job enrichment, leaving the worker happier and more satisfied.[48]
This leads to the conclusion that a firm’s organizational choice, regarding the distribution of authority, will be guided by the trade off considerations discussed above, with the external environment as a decisive factor. The theoretical plausibility consideration with many pros and cons for decentralization depending on the specific environment, which is based on the contingency theory and the neo - institutional economics, as well as the mixed results indicated in the literature review above, suggest that there is no direct and significant effect of organizational variables on firm performance, considering the whole sample.
Another reason for the following hypotheses is that the organizational structure can be seen as a freely choose able variable. It might be true that companies get the organizational structure generally right.
Therefore analyzed for the whole cross country and cross industry sample:
Hypothesis 3: The impact of firm slope on firm performance is not significant.[49]
Hypothesis 4: The impact of plant autonomy on firm performance is not significant.[50]
Hypothesis 5: The impact of worker autonomy on firm performance is not significant.[51]
2.3. Organizational Structure, Management Practices and Firm Performance
The following section derives hypotheses that consider interaction effects between management practices and organizational structure on firm performance. Some research hypotheses of past literature will be extended and verified with the new data in the information and communication age, as well as a new hypothesis will be derived by drawing a conclusion out of the principal agent theory.
2.3.1. Difficulties
While analyzing the impact of management practices and organizational structure on firm performance one major econometric challenge makes it difficult to prove hypotheses in that section. As explained under section 5, the econometric modeling requires a product in the equations, which models the interacted effects between these terms on firm performance. As a natural conclusion, this product leads to the fact that also the measurement errors[52] of both terms get multiplied. This means that, for instance, if we have a true signal of 0.59 for the management score as discussed in section 3.6.2., and for the organizational structure variable of 0.3, as discussed in section 3.6.2., the true signal of the product will be 0.177. This makes it difficult to show significance of theses terms in this section.
2.3.2. Empirical Literature
Gupta et al. (1997) analyze the interdependencies between advanced manufacturing technology management, organizational structure and organizational performance.[53] Modern manufacturing methods, like lean management, can provide competitive advantage for a company.[54] A strand of literature points out that the organizational structure is key to successfully leveraging the benefits of modern manufacturing methods.[55] This leads Gupta et al. (1997) to the conclusion that an organizational structure complimentary to advanced manufacturing methods should have a positive impact on firm performance.[56] Various authors suggest that firms which use modern manufacturing methods should use a less mechanistic and more decentralized organizational structure, in order to realize the benefits of modern manufacturing techniques.[57] This consideration leads to various research hypotheses, which are tested by Gupta et al. (1997). The researchers introduce a measurement for the use of modem manufacturing methods, which they called “Intensity for Advanced Manufacturing Techniques” (IAMT). This measurement is based on the following four criteria:
1. Ratio of capital outlay in computer automation to total manufacturing outlay.
2. Time advanced manufacturing technology used.
3. Percentage of employees in an organization using advanced manufacturing techniques.
4. Average educational level of employees using Advanced Manufacturing Techniques (AMT).[58]
One of the major outcomes is the finding that companies with a high intensity of advanced manufacturing techniques need a higher degree in decentralization in decision making in order to achieve high performance.[59] As suggested by Gupta et al. (1997), it would be worth verifying this intriguing outcome with a cross country data set. Major limitations of their paper include the limited sample to only 101 companies in only one country, the US.[60] To verify this intriguing outcome, the following hypothesis is obtained and tested with cross country data. The IAMT score of Gupta (1997) is approximated with the lean management score.[61] They both have the same goal of quantifying the degree to which advanced/modern manufacturing practices are used. The two scores measure similar factors, such as:
- the degree of use of modern manufacturing
- the amount of time the modern manufacturing is used
Major difference between the measures is that the lean management score is collected by an interview and the IAMT is collected by a written survey.
Hypothesis 6: Where lean management scores are high, performance is increased by higher degrees of decentralization (worker level).[62]
2.3.3. Hypothesis based on Principal Agent Theory
As a major theory within neo - institutional economics which helps to determine efficient principal agent organizational structures, the principal agent theory will be discussed under this section to obtain an intriguing hypothesis.[63]
Organizations can be seen as a network of principal agent relationships.[64] The principal is the client. An agency relationship involves delegating some decision making authority to the agent. The principal(s) engage(s) another person, the agent. In the case that both parties, the principal and the agent, are utility maximizers, it is expected that the agent’s actions may differ from the principal’s interest.[65] The following three problem types can be generally distinguished:[66]
- Hidden characteristics
- Hidden action
- Hidden intention
The principal agent relationship is characterized by asymmetrical information and different utility functions between the principal and the agent. Agency costs consist of the following three components:[67]
- Monitoring and tracking costs of the principal.
- Signaling, guarantee and bonding costs of the agent.
- Residual loss of the agency relationship.[68]
Interest alignment is one of the major methods to reduce principal - agent problems.[69] Another method is monitoring. However monitoring is not the focus of this paper, since this research did not obtain enough data regarding monitoring.[70] Decentralized assignment of duties increases principal agent problems. Hidden action problems especially arise with physical decentralized task allocation.[71] One theoretical conclusion is that it especially benefits decentralized organizations to align interests between the agent and the principal. This increases the positive effects of decentralization, which were discussed in section 2.2.3. One possible way to align the interests of the principal and the agent is to develop an incentive and payment system based on the individuals’ performance.
This consideration suggests that especially firms with high worker autonomy should tend to perform better, in general, with an aligned performance and incentive system. The hypothesis is that, in highly decentralized companies, the agency costs, as mentioned above, are higher than the costs of an aligned incentive system. This means, that the benefits of an aligned appraisal and incentive system offset the costs of such a system for companies with high worker autonomy. The following hypothesis is derived from this theoretical consideration. None of the papers reviewed and, to the best of the author’s knowledge, no paper in the literature has tested this interacted effect with empirical data:
Hypothesis 7: Where the payment and incentive system is aligned, firm performance is increased by higher degrees of worker autonomy (management dimension 14 score interacted with worker autonomy).[72]
2.3.4. Hypotheses based on Change Management Literature
It seems to be a generally accepted statement that the environment in which organizations operate is constantly changing.[73] Accelerated technological progress, especially in the information and communication technology sector, the effects of globalization, mature markets, as well as the fall of socialist and communist regimes are common reasons for the continuous change, even perceived at an increasing rate.[74] Knowledge management and organizational learning literature recommends building learning organizations to successfully compete in the changing environment.[75] Organizational learning has been accepted, by a strand of literature, as holding the key to financial success and competitive advantage in a changing environment.[76] In learning organizations, change is seen as an ongoing process.[77] Aside from theory, case studies report that companies which change their organizational structure according to the changing environment perform better. For instance, General Motors, which recently reported record losses in 2005 of $10.6 billion[78], had already been criticized much earlier by Garvin (1993) for its inability to change.[79] In contrast, GE’s former CEO, Jack Welch, pronounced that the ability to change is one of GE’s success factors.[80]
This paper seeks to empirically test whether companies which changed the number of hierarchical levels in the last three years - as one quantifiable measure for organizational change - perform better. The assumption is that companies that changed their organizational structure in the past three years did so in order to adapt to the new needs of the changing environment. Furthermore, the hypothesis is that companies that are better managed underwent more organizational change.
Hypothesis 8: Firms with higher management scores do more organizational change.[81]
Hypothesis 9: Firms with more organizational change perform better.[82]
3. Data Collection and Method
The data has been obtained from a McKinsey & Company, London School of Economics and Stanford University joint project. The author of the paper was part of a team of 35 international MBA, Ph.d students and professionals from top business schools around the world. Multi - industry, cross firm and cross country data has been obtained through phone interviews, with executives and plant managers of an average length of 53 minutes per interview.[83] In interpreting the data, it is crucial to consider the limitation of this data set. The structure of the data does not address issues of causality.
To analyze the impact of management practices and organizational structure on firm performance, they first need to be measured. The following section discusses these issues.
3.1. Measuring Management Practices
What is good management? To measure management is obviously a hard task, due to the following reasons. First, it is tough to define what good management is. The second challenge is to create evaluation criteria which are applicable across countries and sub industries within the manufacturing sector. To do this, the project used a management practice evaluation tool introduced by McKinsey & Company and modified it according to current academic knowledge. In Appendix 1, the complete management interview guide is documented. The example answers documented in Appendix 1 are those that the interviewer used during the interview with the interviewee, to guide his evaluation. The following Table 1 shows the eighteen dimensions which were considered in the management interview and structures them into three areas lean management, performance management and talent management.
illustration not visible in this excerpt
Table 1: Dimensions and Areas of the Management Interview
These eighteen dimensions were identified by McKinsey & Company in thousands of client projects and are considered to be central in driving firm performance. Similar dimensions have been used by Kao and Hung (2005)[84] and Ichniowski, Shaw and Prenushi (1997).[85] They are an approach to measure management. This approach has already been proven successful by a first McKinsey & Company and London School of Economics pilot study in 2001 and a predecessor project in 2004.[86] Also, see Appendix 3 for a full list of collected variables, including the abbreviations used in Stata, for the log files.[87]
The scaling across each of the eighteen dimensions lies between one (worst practice) and five (best practice). The two extremes were defined according to what McKinsey & Company observed within its client projects. A score of three reflects an industry average practice. Finally, the score allocated to each dimension by the interviewer is based on the applied management practice and matched with the scoring grid.[88] The management score - the variable “Management” for the multivariate regressions under section five and six - is the mean of the individually scored eighteen management dimensions.
In Table 2, the dependent variable logarithm (Sales) was regressed on all eighteen management dimensions. Table 2 shows the significance and the regression coefficients of each of the eighteen management dimensions.
illustration not visible in this excerpt
Table 2: Regression Coefficient and Significance of All Eighteen Management Questions (totaling 3518 firms)[89]
[...]
[1] Dunning (1958), p. 120.
[2] GNI per capita is defined as the Gross National Income converted to U.S. Dollars divided by the midyear population.
[3] See Worldbank (2006b).
[4] See Nissen (1998), p. 411f.
[5] See Bloom / Van Reenen (2007), p. 2ff.
[6] See Bertrand / Schoar (2003), p. 1ff.
[7] See Chandler (1962), Lawrence / Lorsch (1967), Picot (2002).
[8] See Bloom / Van Reenen (2007), p. 1 ff.
[9] See Bloom / Van Reenen (2007), p. 8 f.
[10] See Bertrand / Schoar (2003), p. 7f.
[11] See Bresnahan / Brynjolffson / Hitt (2002), p. 339ff.
[12] See Ichniowski / Shaw / Prenushi (1997), p. 291 ff.
[13] See Kao / Hung, (2005), p. 152 ff
[14] Overall 3518 interviews.
[15] The management score is a quantified measure of the quality of management of a company. Please see section 3.1. for a detailed explanation how the management score was measured and see section 3.3. for how firm performance was measured.
[16] See Hage (1967), p. 503 f.
[17] See Thompson / Wright (1992), p. 11.
[18] See Chandler (1962), p. 2ff.
[19] See Williamson (1975), p. 40 f.
[20] See Jensen / Meckling (1976) p. 306.
[21] See Gupta / Chen / Chiang (1997), p. 512ff.
[22] See Ingham (1992), p. 25.
[23] See Ingham (1992), p.28f.
[24] See Mookherjee (2006), p 369.
[25] See Williamson (1970), p. 170.
[26] See Cable (1988), p. 13.
[27] See Steer / Cable (1978), p. 12 ff, Burton / Obel. (1980), p. 460 ff, Thompson (1980), p. 357 ff, Hill. (1985), p. 731 ff and Hill / Pickering (1986), p. 26 ff.
[28] See Teece (1981), p. 173 ff, Harris (1983) and Cable / Yasuki (1985), p. 401ff.
[29] See Armour / Teece (1978), p. 106 ff, Roberts / Viscione (1981), p. 285 ff, Cable / Dirrheimer (1983), p. 43 ff, Thompson (1983), p. 297 ff and Hoskisson / Galbraith (1985), p. 55 ff.
[30] This paper gives some support fort his argument under section 3.6.2.
[31] See Gupta / Chen / Chiang (1997), p. 512.
[32] See Lawrence / Lorsch (1967), p. 158.
[33] See Cable (1998), p. 28.
[34] See Bourgeois / McAllister / Mitchell (1978), p. 508 ff, Duncan (1972), p.315 ff; Hrebiniak and Snow (1980), p. 750 ff.
[35] See Gordon / Narayanan (1984), p. 35f.
[36] See Lawrence / Lorsch (1967). p. 158.
[37] See Rueckert / Roehring / Walker (1985) , p. 13 ff.
[38] Please refer to section 3.2. for a detailed description how decentralization was measured and section 3.4. how competition was measured.
[39] See Caroli / Van Reenen (2001), p. 1455.
[40] See Greeenan / Guellec (1997), p. 173 - 197.
[41] See Caroli / Van Reenen (2001), p. 1456.
[42] See Thesmar / Thoenig (2000), p. 1207.
[43] See Askenazy (2001), p.510.
[44] With full decentralization, it is ideally the same point.
[45] See Caroli / Van Reenen (2001), p.1454.
[46] See Caroli /Van Reenen (2001), p.1454.
[47] See Thesmar / Thoenig (2000), p. 1201f.
[48] See OECD Employment Outlook (1999), p. 180.
[49] Firm slope measures the hierarchical shape of a company. Please refer to section 3.2.6. for an explanation how firm slope was measured.
[50] Please refer to section 3.2.2. for an explanation how the decentralization measurement plant autonomy is measured.
[51] Please refer to section 3.2.3. for an explanation how the decentralization measurement worker autonomy is measured.
[52] Please refer section 3.6. for an explanation of measurement errors.
[53] See Gupta / Chen / Chiang (1997), p. 511ff.
[54] See Meredith / Hill (1987), p. 49ff.
[55] See Goldhar / Jelinek (1983), p.141ff.
[56] See Gupta / Chen / Chiang (1997), p. 513.
[57] See Nemetz / Fry (1988), p. 630.
[58] See ebenda, p. 512.
[59] See Gupta / Chen / Chiang (1997), p. 518.
[60] See Gupta / Chen / Chiang (1997), p. 514.
[61] The lean management score is a score which measures the intensity of the use of lean management methods in production of the firm. Please refer to section 3.1. for a detailed description how this score is measured.
[62] Please refer to section 3.1. for a detailed description how the lean management score was measured and to section 3.2. for an explanation of the decentralization measurement at worker level, the worker autonomy.
[63] See Picot / Reichwald / Wigand (2003), p. 56.
[64] See ebenda (2003), p. 55.
[65] See Jensen / Meckling (1976), p. 308.
[66] See Picot / Dietl / Frank (2002), p. 88 ff and see Picot / Reichwald / Wigand (2003), p. 57 ff for more details about principal - agent problem types.
[67] See Jensen / Meckling (1976), p. 308.
[68] Includes the benefit of transactions not undertaken because of agency problems.
[69] See Jensen / Meckling (1976), p. 308.
[70] See Picot / Dietl / Frank (2002), p. 94.
[71] See Picot / Reichwald / Wigand (2003), p. 60.
[72] See Table 1 for an overview over all analyzed management dimensions.
[73] See Kotter (1996), p.18.
[74] See Kotter (1996), p.19.
[75] See Goold / Hampbell (2002), p. 117f.
[76] See Nonaka (1991), p. 21ff and Argyris (1991), p. 81ff.
[77] See Gephart / Marsick (1996), p. 45.
[78] See Krolitzki (2006), p.1.
[79] See Garvin (1993), p. 80.
[80] Ebenda, p. 91.
[81] This paper uses as a quantifiable measure for organizational change, the number of levels changed in the hierarchy of the company. Please refer to section 3.2.5. for a detailed description of how organizational change was measured.
[82] To the best of the author’s knowledge this hypothesis has not been tested yet in the literature.
[83] Phone interviews are widely used for empirical research, see e.g. West (1996).
[84] See Kao / Hung (2005), p. 153 f.
[85] See Ichniowski / Shaw / Hung (1997), p. 300 f.
[86] See Bloom / Van Reenen (2007), p. 2 ff.
[87] Log files of regressions in Stata were handed in on a CD.
[88] See Appendix 1 for the scoring grid.
[89] *significant at 1 % level; **significant at the 5 % level; ***significant at the 10 % level; °not significant. Regression includes full controls. This means inlcuding country dummies, industry dummies and noise cointrols. Noise controls include controls for analyst/interviewer, the average percentage of employees with a college degree, gender, seniority of the interviewer, the interviewees perception of the firm, willingness to reveal information, knowledge and age. See section 2.7 for more details.