A contingency based view on deregulated sectors The organisational structure over time:

(1)

Master’s Thesis in Business Administration - track Strategic Innovation Management

The organisational structure over time:

A contingency based view on deregulated sectors

Candidate: Rens Kieft

S3207471

(2)

Abstract

The contingency theory is a well-known theory but has never been studied over a longer period of time. Since the 1960’s the theory has been showed by studying a small number of firms, all studying different contingencies, and the conclusions are applied to a large number of different research fields. This study aims to do a longitudinal study to prove that over time, the findings still hold. Specifically looking at the R&D facilities and the corresponding organisational form after a sector is deregulated. In this case the organisational form is seen as the relative distance between the facility and the headquarters.

To test the hypothesis of this paper, which is that firm’s operating in deregulated sectors adopt more organic organizational structures for their R&D activities, sixteen cases were identified based on regulatory changes in certain sectors by using years prior and after the deregulation. By looking at these cases one by one, and using a fixed effects regression it allowed to show how firms change their structure, but a relatively small group showed they became more organic after a deregulation, but also the statistically significant change in average level between facilities and headquarters.

(3)

1. Introduction

What constitutes the organisational form of a firm? That is a question that has been studied since the nineteen sixties, and so far, there are numerous studies into what makes the organisational form of the firm the way we see them today. For example, Burns and Stalker (1961) argue that the size of the firm is the main reason the firm changes their organisational structure, by either moving to one of two structures. Another influential author, Woodward (1965), argues however that technology is the most important and that a given productional form lays the foundation for the structure of the firm. Mintzberg (1979) however argued that there where as much as five factors that changes the organisational form. However, all of them agree on one thing, there is a best fit between factors in the firm and outside the firm and that if the firm does not change its structure this will lead to lower performance.

When it comes to studying what the most important factor is, or as the literature has since calls the contingency, the studies have all had a few things in common. They study a relatively small number of firms, for one given time period. For example, Burns and Stalker (1961), made use of twenty firms or Lawrence and Lorsch (1967), who were one of the first to make additions to the theory, made use of only six firms to come to their conclusions. Since the theory is compelling, and easy to use, the widespread use of the contingency theory itself in more and more areas of study, but there is a flaw to it. The biggest problem being that it argues that firms change, as something inside their firm or in their environment changes, but we only study the firms for one time period, making it difficult to say whether the firms actually change over time after the change in contingency.

Therefore, this study aims to study a number of firms over time, between the period of 1975 and 1998. To track firms that undergo a change in their surrounding, I study firms that, during this period, experienced a deregulation in their operating sector, which for several reasons changes the contingency the firm finds itself in, for example decreasing stability and increasing hostility. To test this, I make use of a recently made available data set by Png (2019) based on the Cattell Directories, which is fitting as it gives the opportunity to study a large number of firms over time, while also specifying what the organisational structure looked like as it specified the characteristics of the number of facilities and their relative level to the headquarters.

(4)

that firms indeed change structure after changes in their environment but will also provide insights in how we can better study these changes, or how this research can be improved. The paper is organized as follows; section 2 presents the Literature Review, followed by section 3 which is the Hypothesis Development. Section 4 is the Data section, showing the sources, how the final sample was constructed, and the definition of the main variables followed by section 5 explaining the Methodology of this paper. Section 6 describes the Analysis section, subdivided in the descriptive evidence and regressions, and section 7 presents the Discussion, Limitations and Future Research. The numerous tables related to the analysis have been placed in a separate section after the References in order to improve the readability of the paper.

2. Literature review

The contingency theory is an intuitive approach to how a firm is organised. Based on factors inside and outside the firm, the firms must adapt in such a way that it provides highest performance. This is explained clearly by Donaldson (2001) in three core elements. First, there is an association between contingency and the organisational structure. Second, contingency determines the organisational structure, because an organisation that changes its contingency then, in consequence, changes its structure. Third, there is a fit of some level of the organisational structure variables to each level of the contingency which lead to higher performance, whereas misfit leads to lower performance. (Donaldson, 2001) This shows that there is no clear right or wrong for an organisation, but rather a sweet spot for a firm at that moment in time.

(5)

In response to Burns and Stalker (1961), Lawrence and Lorsch (1967) in their study, a comparative study of six firms in the chemical sector, found that modern organisations are being expected to cope with heterogeneous environments that have both highly dynamic and quite stable sector (Lawrence & Lorsch, 1967). Their findings suggest the coexistence of both the mechanistic form as well as the organic form in an organisation as the organisation must continuously focus on both exploration as well as exploitation.

The last most influential paper in the literature is by Mintzberg (1979), who takes a different approach but came to similar results. Instead of basing his findings on empirics he studied the available literature and theoretically found five pure archetypes that companies follow. By setting out eight design parameters and looking at five contingencies he identified the Simple structure, Machine bureaucracy, Professional bureaucracy, Divisionalised form, and Adhocracy.

Since then, there is a vast amount of literature published that studies the matter of organisational fit and the form of this fit and base their theories on the findings of the above-mentioned authors. For example, the study by Hankinson (1999) who found that twice as many Top 100 brand companies in the consumer goods sector had “relatively flat” organisational structures compared to hierarchical structures. Bourgeois III et al. (1978) found that when you ask 24 junior and senior administration students, they will respond more organically to stable environments and more mechanistically to turbulent environments, and shifted to more mechanistic mode when turbulence followed stability (Bourgeois III, McAllister, & Mitchell, 1978)

(6)

all. (McAdam, Miller, & McSorley, 2019) Further, Sunder and Prashar (2020) found that when surveying 213 business units operating in the manufacturing and service sector, the environment is an important factor in continuous improvement. To conclude Gnizy et al. (2017) used a cross-sectional mail survey from 225 UK exporters to look at export marketing decisions and found that greater export dispersion levels are advantageous when the export customer environment is more turbulent and, concurrently, the export technological environment is more stable and the firm employs lesser levels of export information sharing. (Gnizy, Cadogan, Oliveira, & Nizam, 2017)

So far there have been few articles that are based on large quantitative datasets. One of the few papers that used a large data set is by Fourné et al. who did a meta-analysis using 20 panel data and 93 cross-sectional research papers to combine them into a data set consisting of 33,492 organizations studying the ambidexterity within an organisation. They find that organizations in high technology environments can improve their ability to balance exploration and exploitation by adopting structural separation (Fourné, Rosenbusch, Heyden, & Jansen, 2019). The problem is that most of these papers describe the given organisational structure at a given point in time. Therefore, there exists a gap in the literature that studies this structure over time and preferably based on many firms. Or to put in the words of the original authors of an influential theory: Given that the authors do not present theory testing research in The

Management of Innovation, and have conducted no such research since, it becomes necessary

to look elsewhere for evidence as to the validity of mechanistic and organic systems theory. Ideally the hypothesis on technological change would be tested longitudinally. (Burns & Stalker, Mechanistic and Organic Systems, 2006)

3. Theory development

(7)

3.1 Environment

As shown above one of the important contingencies is the surrounding of the firm often referred to as the environment. The firm’s environment can be divided in the internal and the external environment, but the scope of this paper is the external environment. A key advocate of this idea is Donaldson. In his book (2001), he identified four characteristics that the literature identifies as key to describing the environment of the firm. These characteristics are stability (which ranges from stable to dynamic), complexity (ranging from simple to complex), market diversity (ranging from integrated to diversified), and hostility (ranging from munificent to hostile). (Donaldson, 2001) Mintzberg (1979) in his book previously paid attention to the environmental characteristics as used by Donaldson. According to him the difference in stability can be compared to armies as far away from the battlefield the organizations is highly bureaucratic, but on the battlefield, there is a need for greater flexibility. But the main causes for a dynamic environment are unstable government, unpredictable shifts in the economy, unexpected changes in customer demand or competitor supply, client demands for creativity or frequent novelty, and a rapid changing technology or knowledge base. Complexity is more predictable in his view, as an environment is complex to the extent that it requires the organization to have a great deal of sophisticated knowledge about products, customers, or whatever. However, Heydebrand and Noell (1973) found that when knowledge becomes simple when knowledge can be rationalized and broken down into easily comprehended components. Market diversity is rather simple as the diversity can be based on either three things, a broad range of clients, broad range of products or services, or a broad range of geographical areas. Hostility is a special characteristic, according to Mintzberg, as it could be part of stability, as hostile environments are by definition dynamic. Hostility is influenced by competition, the organizations relationship with, governments, outside groups and unions, and the availability of resources.

Apart from environment, size is another influential contingency, with the main hypothesis that the larger the organization the, more elaborate the structure of that firm (Mintzberg, 1979). The other influential contingency is that of technology, with the most important hypothesis that based on the type of production process that is used, the structure of the firm changes (Woodword, 1965)

3.2 Regulation

(8)

operating core. This view is rather limited as in his view regulation describes the influence of the technical system on the work of the operators. However, when we look at the Cambridge dictionary regulation is an “official rule or the act of controlling something,” which of course can be part of how the company does something, but a government body can also define how things should be done. Which brings us to regulated sectors. As we have seen in the previous paragraph, the key characteristics of a firm’s external environment are stability, complexity, market diversity and hostility. The theory of economic liberalisation dictates what happens when governments privatise or intervene in a sector. For example, Amadeo and Banuri (1991) show that policy intervention leads to stability and that the income of the financial sector derives precisely from its anarchic nature, from its quickness and flexibility, its ability to take advantage of transient profit opportunities without being bogged down by longer-run considerations and obligations. The complexity of a firm’s environment is not necessarily affected by being either regulated or not regulated, as a space exploration firm can be state owned, but the postal sector can also be governmentally regulated. Market diversity can be influenced by regulation in the fact that it works limiting to that given country, as it limits the range of customers the firm serves and limits the number of geographical locations to the country instead of to the surrounding countries or the world. Hostility is also limited by governmental regulation as regulated sectors often have a monopoly position in the market, by default not having competitors. Also, the relation to the government, is quite self-explanatory, good and the government has an interest in securing the needed resources for the firm to operate. The regulation also has effect on the ability to participate in R&D, as the new products or services a firm introduces first has to be certified and validated by the regulator, which increases the risk to innovate. Moreover, in regulated services such as energy or telecommunications, there is a high degree of interoperability that forces firms to standardize their new products so that they can be coupled with those of other firms. These two inherent features of regulated sectors impose bureaucracy on innovation activities that mechanistic organizations are better equipped to comply with rather than organic firms.

3.3 Organisational form

(9)

market-based units will be, and where extreme hostility drives the organization to temporarily centralize the structure. Lastly, disparities in the environment encourage the organization to decentralize selectively to differentiated work constellations. (Donaldson, 2001) Based on these hypotheses, Donaldson identified four basic types of structures that derive from a combination of either a stable or a dynamic environment and either a complex or a simple environment. The four combinations that he found are Decentralized Bureaucratic (stable and complex environment), Decentralized Organic (dynamic and complex environment), Centralized Bureaucratic (stable and simple environment), and Centralized Organic (dynamic and simple environment).

When we look at what regulation did to the firm’s environment in the structures identified by Donaldson, we see that the regulated sector firms are expected to have either a decentralized bureaucratic or centralized bureaucratic. However, when the sector the firm finds itself in deregulates, we expect them to move to a more organic variation such as decentralized organic or centralized organic.

Based on this view the following hypothesis can be derived:

H1: Firms operating in deregulated sectors adopt more organic organizational structures for

their R&D activities whereas firms operating in heavily regulated sectors adopt more mechanistic organizational forms for their R&D activities.

An interesting addition to these theories is that Greenan (2003) found that during the 1988 to 1993 period a large number of French firms started to reorganise their firms to modernise. The main goal was to create greater flexibility, like we have seen in the organic variant of firms structures and doing so by implementing more decentralisation and integration. In these reorganized firms, the evolution of the allocation of responsibilities shows a pattern where operators and specialists intervene in a growing number of tasks. What Greenan found is that this decentralization leads to “the growth of the specialist’s area of responsibility” and this causes an “increasing need for experts” which they describe as a movement towards increased

technical expertise. (Greenan, 2003) Coming from a more mechanistic organisational form and

moving towards a more organic form I can expect to see the same result as found by Greenan. From this we can formulate the following hypothesis:

H2: Organisations that operate in recently deregulated sectors will have an increasing number

(10)

4.Data Section

4.1 Sources

The main source of data for this paper are the digitalized Cattell Directories, by Ivan Png (2019), which used the data of US manufacturing firms that were involved in R&D during the period 1975-1998. This dataset provides information on hierarchical levels, number of facilities involved in R&D, number of professionals and specialists. Therefore, it allows to study a large group of firms and their organizational form focusing on their R&D units. By doing so we can study the firms on a corporate level and see whether there are differences after deregulation and whether the contingency theory holds. The data has been collected for the years 1975,1979, 1981, 1983, 1985, 1988, 1991, 1994, and 1998.

To define which industries are prone to regulation we use the sectors identified by Alesina et al. (2003) who found seven sectors that are heavily regulated which they combined in three broader sectors utilities (electricity and gas), communication (telecommunications and post), and transportation (airlines, road freight and railways).

In the OECD review of regulatory reform concerning the united states (1999) we find that during the period of our data set there have been several deregulations in the following sectors: railroads (1976), air cargo (1977), airlines and natural gas (1978), satellite communications (1979), trucking, railroads again, financial institutions, cable television (1980), petroleum, radio (1981) and buses and communications equipment (1982). The deregulation consisted of the replacement of price and entry controls with pro-competitive regulatory regimes, backed up by strong competition policies, which effectively changes the stability the firm finds itself in. 4.2 Construction of final sample

(11)

observations. However, we are mostly interested in the number of observations for our sectors and that fit the years prior and after the deregulation and, reducing the number of feasible firms to study, but the number of observable firms are described in table 1 in the appendix. This left our group of firms and corresponding observation to a total of eighty-four. Next to this I also created a control group with available firms that were in somewhat similar sectors, which proved to be difficult but with some imagination we can see how these sectors are similar to our sectors as alternatives would have been more different from our sectors. The sectors being telephone and telegraph apparatus (similar to communication equipment), oil and gas exploration services (similar to petroleum), miscellaneous products of petroleum, industrial inorganic chemistry (similar to petroleum) and household audio and video equipment (similar to communication equipment). As from our deregulated sectors, eighty-four observations were added providing the final data set with 168 observations.

4.3 Definition of main variables of interest

For this paper there are two dependent variables of interest. For the first hypothesis the dependent variable is the hierarchical level of the facility compared to the headquarters, which is studied on total level and mean level. For the second hypothesis the dependent variable is the number of professionals, which is studied on total and mean level.

The level of the facilities is the dependent variable for hypothesis one as a higher level is an indicator of a more mechanistic and bureaucratic organisational form. To have a more in-depth view of this dependent variable we will study it on both the total level, which is the accumulated number of levels that all facilities have from the headquarters, meaning that if the firm has one facility with level two and one facility with level one the total level will be three and the mean level, which is average level a facility has from the headquarters. So, if we take the previous example, two and one equal three, between two facilities, so the mean level is one and a halve. These two variables, total level, and mean level were generated.

(12)

The explanatory variable for this paper is the deregulation of industries. As we have seen in section 4.1 a number of industries have been deregulated during the period of the dataset. In order to study those and link it to our dataset I used the Standardized Industrial Classification. To have the needed SIC numbers I looked which matched the deregulated sectors, and these numbers can be found in table 2 in the Appendix. The deregulated sectors are linked to the years and therefore a dummy variable has been created, where the dummy variable gets value one after the year in which the industry the firm operates in has been deregulated and value zero in the years preceding the deregulation (or if there is no deregulation at all).

4.4 Descriptive statistics

Table three in the appendix describes the descriptive statistics of our four main variables of interest. In addition, the frequencies tables of these variables can be found in the appendix in tables 4 through table 8.

5. Methods

To test the two hypotheses, I opted to study them in two ways, by looking at the descriptive evidence, meaning studying the individual firms and how they change over time, and these changes in individual firms can be found in the tables in the appendix. Second, I use a fixed effect regression, with the following formula:

𝑦𝑖𝑡 = 𝛽𝑑𝑒𝑟𝑒𝑔𝑢𝑙𝑎𝑡𝑒𝑑𝑖𝑡+ 𝜃𝑖+ 𝜃𝑡+ 𝜀𝑖𝑡

Where i indexes firms, t indexes years, 𝑑𝑒𝑟𝑒𝑔𝑢𝑙𝑎𝑡𝑒𝑑_𝑖𝑡 is a dummy variable that switches from zero to one as the industry becomes deregulated, 𝜃𝑖 are firm fixed effects and 𝜃𝑡 are year fixed

effects. 𝛽 is the parameter of interest that tests the hypothesis of interest which is either total level, mean level, total professionals, or mean professionals. For hypothesis one this means that if 𝛽 is negative the number of total levels or mean levels is decreasing, while for hypothesis two, if 𝛽 is positive the total number of professionals or the mean number of professionals is increasing. This is estimated on a sample of deregulated firms and non-deregulated controls. For the non-regulated controls, 𝑑𝑒𝑟𝑒𝑔𝑢𝑙𝑎𝑡𝑒𝑑𝑖𝑡 will always be zero. For the deregulated firms,

(13)

6. Results

6.1 Descriptive evidence

When looking for evidence that supports hypothesis 1, that after deregulation the firms would opt for more organic firm structure, we want to look at the total level and mean level variables. The observed changes can be found in table 9 and table 10.

Table 9. Observed changes in Total level.

Total level Case(s)

Decreases Cases 3, 7, 12

Remains equal Cases 2, 4, 8, 10, 11, 13, 14, and 16 Increases then decreases Cases 9, and 15

Decreases then increases Case 1

Fluctuates (General trend downwards) Cases 5, and 6

Table 10. Observed changes in Mean level.

Mean level Case(s)

Decreases Cases 3, 7, and 12

Remains equal Cases 2, 4, 8, 9, 10, 11, 13, 14, and 16

Increases Case 9

Fluctuates (General trend upwards) Case 1

Fluctuates (General trend downwards) Cases 5, 6, and 15

We see that in only three companies the total level in relation to the headquarters becomes more organic, but also that for 2 firms that the total level first increases and then decreases. In two more firms the trend is more fluctuating, but in the end still lower than where the firms started. In the other cases the total level remains constant or becomes more mechanistic. Cases 7, 12, and fifteen operate in the same industry and the same is true for cases 3, 5, and 6.

(14)

For hypothesis 2, that after the deregulation the number of professionals increases, we look at the total number of professionals on the corporate level but also look at the mean of the professionals per facility. The observed changes can be found in table 11 and 12.

Table 11. Observed change in total number of professionals.

Total number of professionals Case(s)

Decreases Cases 7, and 12

Remains equal Cases 8, and 16

Increases Cases 1, 9, and 10

Fluctuates (General trend upwards) Case 4, 6, and 8 Fluctuates (General trend downwards) Case 3

Fluctuates (End year equal to first year) Case 14

Unknown Cases 2, 5, 11, 13, and 15

Table 12. Observed changes in average number of professionals.

Average number of professionals Case(s)

Decreases Cases 7

Remains equal Cases 8, and 16

Increases Cases 1, 4, and 10

Fluctuates (General trend upwards) Cases 3, 6, 9, and 12 Fluctuates (End year equal to first year) Case 14

Unknown Cases 2, 5, 11, 13, and 15

(15)

A flaw which will be later discussed in the limitations is that for five firms we have missing values for either one year or multiple years which makes it harder to study, which is the case for five firms for both the total and average number of professionals.

All tables concerning the specific case firms can be found in the appendix in table 13 through table 28.

6.2 Formal econometric results

For the results of the fixed effect regression on the total level we find the following. The β is -1.143349 but has a t-score of -1.8 and a p>t score of 0.075, meaning that it is not significant. Also, the model itself is not correct as the prob>f is 0.3533 as it lies above 0.05 indicating that a large number of coefficients in the model are zero, which we have also seen in the descriptive evidence section.

For the results of the fixed effect regression on the mean level we find the following. The β is -0.1740071 and has a t-score of -2.81, which is above -1.96 meaning it is significant at a 95% confidence interval and a p>t value of 0.006, which is below 0.05 meaning that it is significant at 95% confidence interval. Also, the model itself is correct as the prob>f is 0.0378 as it lies below 0.05.

For the results of the fixed effect regression on the total number of professionals we find the following. The β is 29.58358 and has a t-score of 0.5, which is below 1.96 meaning it is not significant and a p>t value of 0.0609, which is above 0.05 meaning that it is not significant. The model itself is correct as the prob>f is 0.0466 as it lies below 0.05.

For the results of the fixed effect regression on the mean number of professionals we find the following. The β is 39.45026 and has a t-score of 1.59, which is below 1.96 meaning it is not significant and a p>t value of 0.116, which is above 0.05 meaning that it is not significant. The model itself is not correct as the prob>f is 0.0577 as it lies above 0.05.

The entire fixed regression tables can be found in the appendix under tables 29 through 32.

7. Conclusion, discussion, and limitations

(16)

facilities and the headquarters, and the change in mean level has also been proved using the econometric results of a fixed effects regression.

For hypothesis 2, which tested the idea that when firms move to a more organic form, the number of professionals increases, we found, based on the descriptive evidence, that one firm indeed increase the total number of professionals, and three firms increased the average number of professionals per facility. These findings were not statistically proven by the econometric test of the fixed effects regressions.

Limitations and future research

This paper seems to have little impact on the existing literature and started out by claiming to make use of a large quantitative dataset, which turned out to be smaller than expected, but it did answer to a call from the original authors of the contingency theory, Burns and Stalker, to do this type of study. Using the current data set, there are a number of limitations.

For example, the depleted final dataset. The Cattell data set in itself is a rich dataset as it recorded for a large number of firms all the fields that a firm performed R&D in. For example, one facility can have as many as fifteen sub-R&D fields that it works in, which can all be accounted to different sectors. But since we want to study the firm on a corporate level, this collapse causes a lot of data to get lost. The problem is it is hard to track down what the firm’s main line of business was, as that is the most important sector the firm operates in. In order to counter this, I made use of a Compustat match, to use their available data and match it to Cattell dataset, but this resulted in the loss of about 80% of the observations. This might be caused by the fact that Compustat only has data on publicly traded firms where the Cattell Directories were collecting their data on all known non-government facilities currently active in any commercially applicable basic and applied research, including development of products and processes. This raises the problem that often, regulated firms are not publicly traded meaning they might have been left out in this research. Future studies could make use of the field and sub field data that is in the Cattell files in order to circumvent this issue of underrepresentation of the firms that are not publicly traded.

(17)

that a large number of headquarters is not represented in this sub-data set. It could therefore very well be that many these observations can be ascribed to the headquarters, but it could also be that in the original Cattell Directories this has not been recorded. Further research could dive deeper into this issue and come up with a way to resolve this issue.

With regard to the limitation mentioned in the descriptive evidence, which was absence of some data for the total and average number of professionals, the problem lies in the fact that in the original Cattell Directories there may have been problems with registering the number of professionals per facility or doing so in a continuous way, meaning that for every Directory the correct number was registered. An example of this is firm 9, which has now been presented as an increase as it increases from 6 to 1667 to 1673, but 6 professionals divided over 9 facilities seems odd, and an increase of 1661 professionals in a year also seams highly unlikely. The findings in this paper suggest that the findings by Greenan, can be proven, but this dataset might not have been the most appropriate to do so.

(18)

References

Alesina, A., Ardagna, S., Nicoletti, G., & Schiantarelli, F. (2003). Regulation and Investment.

NBER Working Paper Series.

Blau, P., McHugh Falbe, C., McKinley, W., & Tracy, P. (Mar. 1976, Vol. 21, No. 1). Technology and Organization in Manufacturing. Administrative Science Quarterly , 20-40.

Bourgeois III, L., McAllister, D., & Mitchell, T. (1978). The Effects of Different Organizational Environments upon Descicions about Organizational Structure. The Academy of

Management Journal, 508-514.

Burns, T., & Stalker, G. (1961). The Management of Innovation. Oxford University Press. Burns, T., & Stalker, G. (2006). Mechanistic and Organic Systems. In J. Miner, Organizational

Behavior 2: Esential Theories of Process and Structure (pp. 214-226). London:

Routledge.

Donaldson, L. (2001). The Contingency Theory of Organizations. SAGE Publication, Inc. Fourné, S., Rosenbusch, N., Heyden, M., & Jansen, J. (2019). STRUCTURAL AND

CONTEXTUAL APPROACHES TO AMBIDEXTERITY: A META-ANALYSIS OF ORGANIZATIONAL AND ENVIONMENTAL CONTINGENCIES. European

Manangement Journal, forthcoming.

Gnizy, I., Cadogan, J., Oliveira, J., & Nizam, A. (2017). The empirical link between export dispersion and export performance: A contingency-based approach. International

Business review, 239-249.

Greenan, N. (2003). Organisational change, technology, employment and skills: an emprical study fo French manufacturing. Cambridge Journal of Economics, 287-316.

Hankinson, P. (1999). An empirical study which compares the organisational strucutres of companies mangaging the World's Top 100 brands with thsoe managing Outsider brands. Journal of Product & Brand Management, 402-414.

Heydebrand, W., & Noell, J. (1973). Task Structure and Innovation in Professional Oranzations. In W. V. Heydebrand, Comparative Organizations (pp. 294-322). Englewood Cliffs: Prentice-Hall.

Lam, A. (2010). Innovative Organizations: structure, learning and adaptation. Innovation

Perspectives for the 21st century, Madrid BBVA, 163-177.

Lawrence, P., & Lorsch, J. (1967). Differentiation and Integration in Complex Organizations.

Administrative Science Quarterly, 1-47.

Maduenyi, S., Oke, A., Fadeyi, O., & Ajagbe, M. (2015). Impact of Organisational Structure on Organisational Performance. International Conference on African Development

(19)

McAdam, R., Miller, K., & McSorley, C. (2019). Towards a contingency theory perspective of quality management in enabling strategic alignment. Int. J. Production Economics, 195-209.

MIles, R., Snow, C., Meyer, A., & Coleman, H. J. (1978). Organizational Strategy, Structure and Process. The Academy of Management Review, 546-562.

Mintzberg, H. (1979). The structuring of Organizations. Englewood Cliffs.

OECD. (1999). Government Capacity To Assure High Quality Regulation In The United States. Paris: Organisation for Economic Co-Operation and Development.

Pleshko, L. (2006). Strategic orientation, organisational structure, and the associated effects on performance . Journal of Finciancial Services Marketing, 53-64.

Png, I. (2019). U.S. R&D, 1975-1998: A New Dataset. Strategic Management Journal, 715-735.

Pugh, D., Hickson, D., Hinings, C., & Turner, C. (1968). Dimensioins of Organization Structure. Administrative Science Quarterly, 65-105.

Reinman, B. C. (1973). On the Dimensions of Bureaucratic Structure: An Emperical Reappraisal. Administrative Science Quarterly, 462-476.

Sunder, V., & Prashar, A. (2020). Empirical examination of critical failure factors of continuous improvement deployments: stage-wise results and a contingency theory perspective.

International Journal of Production Research, 4894-4915.

(20)

Appendix

Table 1. Deregulated sectors

Sector (year) Corresponding SIC code (either two, three or four number code)

(21)

Table 2. Deregulated sector and corresponding SIC code.

Deregulated sector (year) Corresponding SIC code (either two, three or four number code)

Railroad (1976 &1980) 40 and 3743 Air cargo (1977) 45 Natural gas (1978) 492 Satellite communication (1979) 489 Trucking (1980) 4212, 4213, 4214 Cable television (1980) 4833, 4841 Petroleum (1981) 1311, 2911, 2999 Radio (1981) 4832 Communication equipment (1982) 3663,3669

Table 3. Descriptive statistics main variables of interest

N Minimum Maximum Mean Std. Deviation

Total level 166 0 16 2.73 3.809

Mean Level 166 0 1.56 0.4647 0.47329

Total professionals 166 0 1673 129.66 293.013 Mean Professionals 126 2 502.5 62.8976 101.73743

Table 4. Frequencies year

Frequency Percent Valid Percent Cumulative Percent

(22)

Table 5. Frequencies Total level

0 76 45,8 45,8 45,8 1 11 6,6 6,6 52,4 2 21 12,7 12,7 65,1 3 12 7,2 7,2 72,3 4 4 2,4 2,4 74,7 5 10 6,0 6,0 80,7 6 8 4,8 4,8 85,5 7 4 2,4 2,4 88,0 8 5 3,0 3,0 91,0 9 3 1,8 1,8 92,8 10 1 ,6 ,6 93,4 11 3 1,8 1,8 95,2 12 1 ,6 ,6 95,8 13 1 ,6 ,6 96,4 14 4 2,4 2,4 98,8 16 2 1,2 1,2 100,0 Total 166 100,0 100,0

Table 6. Frequencies mean level.

(23)

1,13 1 ,6 ,6 89,2 1,14 2 1,2 1,2 90,4 1,17 2 1,2 1,2 91,6 1,20 1 ,6 ,6 92,2 1,22 1 ,6 ,6 92,8 1,25 2 1,2 1,2 94,0 1,29 1 ,6 ,6 94,6 1,33 3 1,8 1,8 96,4 1,38 1 ,6 ,6 97,0 1,40 3 1,8 1,8 98,8 1,44 1 ,6 ,6 99,4 1,56 1 ,6 ,6 100,0 Total 166 100,0 100,0

Table 7. Frequencies total professionals

(24)

(25)

800 1 ,6 ,6 93,4 841 2 1,2 1,2 94,6 857 1 ,6 ,6 95,2 879 1 ,6 ,6 95,8 925 1 ,6 ,6 96,4 944 1 ,6 ,6 97,0 975 1 ,6 ,6 97,6 983 1 ,6 ,6 98,2 1005 1 ,6 ,6 98,8 1667 1 ,6 ,6 99,4 1673 1 ,6 ,6 100,0 Total 166 100,0 100,0

Table 8. Frequency mean professionals.

(26)

(27)

196,60 1 ,6 ,8 89,7 210,25 1 ,6 ,8 90,5 231,25 1 ,6 ,8 91,3 236,00 1 ,6 ,8 92,1 247,00 1 ,6 ,8 92,9 285,67 1 ,6 ,8 93,7 295,50 2 1,2 1,6 95,2 334,60 1 ,6 ,8 96,0 392,00 1 ,6 ,8 96,8 400,00 1 ,6 ,8 97,6 416,75 1 ,6 ,8 98,4 420,50 1 ,6 ,8 99,2 502,50 1 ,6 ,8 100,0 Total 126 75,9 100,0 Table 13. Firm 1

newid firm year gvkey sic nfacy tlevel level tprof prof dummy

172 1 1979 1122 4841 7 6 0.857143 0 0

172 1 1981 1122 4841 7 6 0.857143 0 1

172 1 1983 1122 4841 6 5 0.833333 16 16 1

172 1 1985 1122 4841 7 8 1.142857 79 19.75 1

Table 14. Firm 2

newid firm year gvkey sic nfacy tlevel level tprof prof Dummy

5584 2 1979 4363 1311 1 0 0 0 0

5584 2 1981 4363 1311 1 0 0 0 1

5584 2 1983 4363 1311 1 0 0 0 1

Table 15. Firm 3

1622 3 1985 1848 2911 12 16 1.333333 983 196.6 0

1622 3 1988 1848 2911 7 8 1.142857 857 285.6667 1

1622 3 1991 1848 2911 4 4 1 494 247 1

1622 3 1994 1848 2911 5 5 1 591 295.5 1

(28)

Table 16. Firm 4

2819 4 1979 2639 2911 3 2 0.666667 6 3 0 2819 4 1981 2639 2911 3 2 0.666667 6 3 1 2819 4 1983 2639 2911 3 2 0.666667 3 3 1 2819 4 1985 2639 2911 3 2 0.666667 8 4 1 2819 4 1988 2639 2911 3 2 0.666667 16 5.333334 1 Table 17. Firm 5

9845 5 1979 7008 2911 3 3 1 12 12 0 9845 5 1981 7008 2911 2 1 0.5 3 3 1 9845 5 1983 7008 2911 2 1 0.5 6 3 1 9845 5 1985 7008 2911 3 3 1 4 4 1 9845 5 1997 7008 2911 2 1 0.5 0 1 Table18. Firm 6

15407 6 1979 10482 2911 9 14 1.555556 304 76 0 15407 6 1981 10482 2911 9 13 1.444444 585 146.25 1 15407 6 1983 10482 2911 10 14 1.4 584 146 1 15407 6 1985 10482 2911 10 14 1.4 648 129.6 1 15407 6 1988 10482 2911 6 7 1.166667 841 210.25 1 15407 6 1991 10482 2911 6 8 1.333333 925 231.25 1 15407 6 1994 10482 2911 7 9 1.285714 944 236 1 15407 6 1997 10482 2911 8 11 1.375 670 167.5 1 Table19. Firm 7

6825 7 1981 5138 3663 4 3 0.75 21 7 0

6825 7 1983 5138 3663 4 3 0.75 19 6.333334 1

6825 7 1985 5138 3663 4 3 0.75 19 6.333334 1

6825 7 1988 5138 3663 3 2 0.666667 11 5.5 1

Table 20. Firm 8

10308 8 1981 7268 3663 1 0 0 16 16 0

10308 8 1983 7268 3663 1 0 0 16 16 1

(29)

Table 21. Firm 9

10866 9 1981 7585 3663 9 9 1 6 6 0

10866 9 1983 7585 3663 9 11 1.222222 1667 416.75 1

10866 9 1985 7585 3663 8 10 1.25 1673 334.6 1

Table 22. Firm 10

12903 10 1981 8917 3663 1 0 0 10 10 0 12903 10 1983 8917 3663 1 0 0 10 10 1 12903 10 1985 8917 3663 1 0 0 10 10 1 12903 10 1988 8917 3663 1 0 0 10 10 1 12903 10 1991 8917 3663 1 0 0 11 11 1 Table 23. Firm 11

13808 11 1981 9493 3663 1 0 0 0 0 13808 11 1983 9493 3663 1 0 0 0 1 13808 11 1985 9493 3663 1 0 0 0 1 13808 11 1988 9493 3663 1 0 0 5 5 1 13808 11 1991 9493 3663 1 0 0 5 5 1 Table 24. Firm 12

13842 12 1981 9503 3663 6 5 0.833333 128 25.6 0

13842 12 1983 9503 3663 3 2 0.666667 100 33.33333 1

13842 12 1985 9503 3663 1 0 0 31 31 1

Table 25. Firm 13

(30)

Table 26. Firm 14

15160 14 1981 10364 3663 1 0 0 4 4 0 15160 14 1983 10364 3663 1 0 0 4 4 1 15160 14 1985 10364 3663 1 0 0 7 7 1 15160 14 1988 10364 3663 1 0 0 7 7 1 15160 14 1991 10364 3663 1 0 0 10 10 1 15160 14 1994 10364 3663 1 0 0 10 10 1 15160 14 1997 10364 3663 1 0 0 4 4 1 Table 27. Firm 15

16562 15 1981 11312 3663 7 6 0.857143 0 0 16562 15 1983 11312 3663 7 6 0.857143 20 20 1 16562 15 1985 11312 3663 7 6 0.857143 40 20 1 16562 15 1988 11312 3663 8 8 1 77 19.25 1 16562 15 1991 11312 3663 6 7 1.166667 50 25 1 16562 15 1994 11312 3663 4 3 0.75 0 1 16562 15 1997 11312 3663 4 3 0.75 0 1 Table 28. Firm 16

3932 16 1981 3481 3669 1 0 0 8 8 0

3932 16 1983 3481 3669 1 0 0 8 8 1

(31)

Table 29. Fixed effect regression results Total level.

Table 30. Fixed effect regression results mean level.

F test that all u_i=0: F(32, 124) = 11.44 Prob > F = 0.0000 rho .70820656 (fraction of variance due to u_i)

sigma_e .25178342 sigma_u .39225577 _cons .4699524 .1285538 3.66 0.000 .2155084 .7243963 1997 .0931342 .1482246 0.63 0.531 -.2002437 .3865122 1994 .0894029 .1465665 0.61 0.543 -.2006933 .3794991 1991 .1181053 .1399144 0.84 0.400 -.1588246 .3950351 1988 .1141721 .1346131 0.85 0.398 -.1522649 .3806091 1985 .1461976 .1312685 1.11 0.268 -.1136195 .4060147 1983 .0975412 .1300461 0.75 0.455 -.1598565 .354939 1981 -.2575065 .1298701 -1.98 0.050 -.5145557 -.0004572 1979 -.2926373 .1354411 -2.16 0.033 -.5607133 -.0245614 year tlevel -.0221737 .0123465 -1.80 0.075 -.0466108 .0022634 dummy Coef. Std. Err. t P>|t| [95% Conf. Interval] corr(u_i, Xb) = -0.0479 Prob > F = 0.0000 F(9,124) = 7.80 overall = 0.1252 max = 9 between = 0.0558 avg = 5.0 within = 0.3616 min = 2 R-sq: Obs per group:

Group variable: newid Number of groups = 33 Fixed-effects (within) regression Number of obs = 166

sigma_e .24727968 sigma_u .41932105 _cons .5990528 .1398959 4.28 0.000 .3221596 .875946 1997 .0730343 .1456837 0.50 0.617 -.2153146 .3613831 1994 .0748914 .1436868 0.52 0.603 -.2095051 .3592878 1991 .1049079 .1374933 0.76 0.447 -.1672299 .3770456 1988 .0949931 .1325079 0.72 0.475 -.1672771 .3572633 1985 .1179577 .129514 0.91 0.364 -.1383869 .3743022 1983 .0668545 .1285232 0.52 0.604 -.187529 .321238 1981 -.3010807 .1291308 -2.33 0.021 -.5566667 -.0454948 1979 -.3367843 .1346091 -2.50 0.014 -.6032135 -.0703551 year level -.3442857 .1224758 -2.81 0.006 -.5866996 -.1018717 dummy Coef. Std. Err. t P>|t| [95% Conf. Interval] corr(u_i, Xb) = -0.2222 Prob > F = 0.0000 F(9,124) = 8.60 overall = 0.0852 max = 9 between = 0.0157 avg = 5.0 within = 0.3842 min = 2 R-sq: Obs per group:

(32)

Table 31. Fixed effect regression results total number of professionals.

Table 32. Fixed effect regression results mean number of professionals.

sigma_e .25476743 sigma_u .37243347 _cons .3851492 .1215844 3.17 0.002 .1444997 .6257988 1997 .1154424 .1502476 0.77 0.444 -.1819397 .4128245 1994 .1143925 .1487054 0.77 0.443 -.1799372 .4087222 1991 .1268474 .1426383 0.89 0.376 -.1554737 .4091685 1988 .1188536 .1376678 0.86 0.390 -.1536295 .3913367 1985 .145368 .1342485 1.08 0.281 -.1203475 .4110835 1983 .1153745 .131522 0.88 0.382 -.1449443 .3756933 1981 -.2350399 .1307429 -1.80 0.075 -.4938167 .0237368 1979 -.2641137 .1360673 -1.94 0.055 -.533429 .0052016 year tprof .0000714 .0001394 0.51 0.609 -.0002045 .0003473 dummy Coef. Std. Err. t P>|t| [95% Conf. Interval] corr(u_i, Xb) = 0.1058 Prob > F = 0.0000 F(9,124) = 7.30 overall = 0.1927 max = 9 between = 0.2088 avg = 5.0 within = 0.3464 min = 2 R-sq: Obs per group: