Minna Tuppurainen

Technical Research Center of Finland

Group for Technology Studies

P.O.Box 1002, FIN-02044 VTT, FINLAND

Tel. +358 9 456 4249

email: Minna.Tuppurainen@vtt.fi

Key words: longitudinal panel data, enterprise demography, internal and external growth, the evolution of the enterprises and industries

Industries renew over time. New enterprises born while old enterprises are ceased. This turbulence changes industries. Enterprises may also change their size: some grow while others contract. Industries may also renew internally: products and production methods change. Evolution of the industry can be described for example by the number of new enterprises, by growth of the enterprise or by the changes in the enterprise field.

New enterprises reform industry. They may introduce new products, new production methods or for example new business ideas. Growth of the industry can hardly be explained only by the number of new enterprises. Growth of the industry is due to growth of enterprises of the industry. Enterprises may grow internally or externally. Internal growth consists of the growth exclusive of any business arrangements such as merger or acquisition. Both growth and new enterprises change enterprise field. Enterprises by themselves has to grow and change. Enterprises may merge and demerge. Enterprises combine their resources in accordance to get synergic advantages.

The data of this study was collected from a longitudinal worker-establishment database (WEDB). The WEDB includes variables from the Business Register, the Regional Employment Statistics and the Statistics of Bankruptcies. WEDB is a longitudinal panel data and it gives an opportunity to follow up one particular enterprise over time. The uniqueness of the WEDB is the link between the employees, enterprises and establishments.

The main object of this study was to group Finnish manufacturing enterprises by using life-cycle variables. Life-cycle variables turnover, personnel, education, enterprise status, business arrangements and industry were created by using the WEDB. First three life-cycle variable describe change over time and the other three life-cycle variables describe some charasteristic of the enterprise. Based on these life-cycle variables, Finnish manufacturing enterprises were classified into 12 different enterprise groups.

The real growth of an enterprise, i.e. its internal growth, consists of the growth exclusive of any business arrangements, such as mergers or acquisitions. The growth often originates from more effective use of production resources. The factors restricting the growth of an enterprise are its management resources, input and output markets and insecurity about the future. Growth and expansion emanate from elaboration. Elaboration requires management resources, which are limited. Since an enterprise's expertise lies in its management, no additional resources can be hired for the elaboration. The professional competence of the management also affects an enterprise's growth opportunities. The management constitutes an internal restriction, while input and output markets are external restrictions. Insecurity about the future is a combination of internal and external restrictions. How insecurity restricts the activity of an enterprise depends on how its management views the risks involved and what the prevailing external conditions are.

The growth of an enterprise is an internal means of gain. Growth follows from expanded production, made possible by the exploitation of unused resources. The economics of growth exist for an enterprise if, in a given circumstance, its extra costs for producing one extra unit are lower than those of any other competing enterprise in the market. A crucial factor in the economics of growth is the enterprise's ability to exploit its resources more effectively regardless of its size.

An enterprise diversifies when it starts producing new products without deserting totally its old products or production technologies. Diversification follows from changes in an enterprise's internal or external conditions. If an enterprise wants to remain competitive, it has to keep up with changes in production technologies. Diversification may cause an enterprise to change its strategy in various ways, that is: 1. produce a new product for new markets using old technology, 2. produce a new product for old markets using new technology, or 3. produce a new product for new markets using new technology. A fourth alternative would be to produce a new product for itself. This is known as vertical diversification.

In the long run, the best way to compete against direct and indirect competition is to anticipate or at least keep up with the latest innovations, products and technologies. Enterprises want to hold on to their markets and, therefore, want to learn all about the products, markets and technologies available to them, as well as predict the innovations of other enterprises. An enterprise can shield itself against competition and changes in demand by producing as wide a range of products as possible. If there are changes in the competitive situation, e.g. one enterprise reaches monopoly position or the demand for one specific product collapses, an enterprise is protected by the variety of products it produces. An enterprise may reach a point where it feels it has exhausted a certain market. It may then want to diversify to new markets through acquisitions. Diversification to new markets may also become necessary if the present ones cannot satisfy an enterprise's growth potential.

If an enterprise wants to diversify to new markets, acquisitions can be the best way to do it. An acquisition may be considerably more advantageous than the establishment of a new plant. With an acquisition, an enterprise gains skilful management, adequate technology, experienced personnel and new markets. Growth by acquisition does not reduce the responsibility of the management of the acquiring enterprise. To accomplish successful diversification, the management must integrate two enterprises under the same strategy. Diversification to brands the enterprise already knows or brands that support its own activities is logical. Too wide diversification, or diversification to wrong products, could be very disastrous to en enterprise.

Vertical integration is a special type of diversification. In vertical integration, an enterprise increases the production of those products which it uses itself. An enterprise may integrate backward, by beginning to produce products which it previously bought in from somebody else or an enterprise can integrate forward by producing products that are closer to the final consumer. By backward integration an enterprise wants to reduce it costs. Backward integration is gainful if an enterprise can make a product itself more cheaply than the price it would have to pay for it if somebody else made it. In comparing the two alternative types of integration, the resources in alternative use also have to be compared to establish whether there would be any gain if the resources needed to produce the new products already were in alternative use.

Enterprise life-cycle analyses do not have any rules about disruptions in the lifespans of enterprises. The growth of an enterprise may be internal or it may result from an acquisition. If a merger is deemed as an economic way of expansion, it will take place. An enterprise may acquire another enterprise complete, or only one part of its operations. If an acquisition is due to take place, the price must satisfy both parties. The enterprise due to be acquired must be more valuable to the acquirer than to itself. The acquired enterprise may be more valuable to the acquirer due to the different ways enterprises view risks. Alternatively, a small enterprise may reach a point where its management resources are inadequate. It then has to either increase its management resources or merge with another enterprise. A merger may be the only way a small enterprise can solve its problems arising from inadequate management resources.

In a combination merger, two enterprises merge into one and the original enterprises will be wound up. A combination merger will most likely take place if the two enterprises involved are more valuable together than they would be separately. A merger will increase the market power of the two merging enterprises. When two or more enterprises merge, unused resources may emerge, which would be out of the range of any one of the involved enterprises on its own. The joint growth potential of a merger may be far greater than that of any one of the companies involved in it.

The data of the study was collected from a longitudinal worker-establishment database (WEDB). The WEDB consists of panel data covering the years from 1987 to 1995 and it includes variables from the Business Register, the Regional Employment Statistics and the Statistics of Bankruptcies. The first step in forming the WEDB was to collect data on all employees who belonged to the labour force and had a link to an enterprise or establishment in at least one year during the 1987-1993 period. The employees and the respective identifier numbers (social security number, enterprise identifier number and establishment identifier number) were obtained from Statistics Finland's Regional Employment Statistics and the variables characterising this population. Data were collected on variables like age, area of domicile, principal activity, industrial status, level of education and wages. The observation unit of the data was the employee.

The variables of enterprises and establishments were collected from the Business Register. Both the enterprise identifier number and the establishment identifier number were used as the links. The variables characterising enterprises and establishments included home municipality, turnover, total wage bill, principal activity code, type of ownership, legal form and particulars descriptive of status, such as date of setting or winding up of activity, date of transfer or take-over and certain details identifying mergers and de-mergers of units, etc. Information about bankruptcies of enterprises was collected from the Statistics of Bankruptcies.

The data gathered up by linking data on Regional Employment Statistics, the Business Register and the Statistics of Bankruptcies covered approximately 1.5 million employees. The number of enterprises varied between 210,000 and 240,000 and the number of establishments between 260,000 and 300,000 per year. These numbers also included no longer active and wound up enterprises and establishments, which was an absolute prerequisite for linking enterprises and establishments over a period of time.

The data for this study were collected from the WEDB. The panel data covered the period of time between 1987 and 1995 and contained about 20 variables for almost 80,000 manufacturing enterprises. Some new variables were created using the panel data variables. These new variables were primarily produced to describe industrial renewal.

Life-cycle variables were created to describe the lifespan of an enterprise. The data and the life-cycle variables covered the years between 1987 and 1995. The first and the last year of the examination need further discussion. When I speak about the first year, or starting point, I mean the year in which an enterprise first enters the study. For enterprises born before, or in, 1987, the first year is 1987. For enterprises born after 1987, the first year is their year of birth. Respectively, the last year, or the ending point, is the last year the enterprise was included in our study. For enterprises alive in, or after, 1995, the last year is 1995. For enterprises wound up before 1995, the last year is the year in which they were wound up. In investigating the life-cycle variables one must recognise that they describe the situation from the point of view of one industry only, i.e. if an enterprise changes its industry, its activities in the new industry have not been examined. The life-cycle variables of turnover, personnel and education describe change over a period of time. The other three variables of enterprise status, industry and business arrangements describe certain features in an enterprise's character.

The life-cycle variable of turnover describes how the turnover of an enterprise has developed between 1987 and 1995. Enterprises were classified into ten different groups. The largest group comprised enterprises with no turnover, or turnover in one year only, between 1987 and 1995, where the development of turnover could, therefore, not be studied. The main direction of the development of turnover was ascertained by comparing the turnover of the last year against that of the first year. Development was deemed as stable if the change varied within the range of ± 12 per cent. If turnover increased by more than 12 per cent, it was classified as "increased". In contrast, if turnover decreased by more than 12 per cent, it was classified as "decreased".

Once the main trend was established, the turnover was examined more closely. If the annual increase exceeded 30 per cent, turnover was classified as "increased with a boost". If turnover increased or decreased by more than 70 per cent, it was classified as "rapidly increased" or "rapidly decreased". If turnover increased by more than 70 per cent and the annual rate in any one year exceeded 30 per cent, it was classified as "rapidly increased with a boost". All the life-cycle variables of increased turnover are illustrated in Figure 1, while those of decreased turnover are shown in Figure 2.

If the change in turnover was within ± 12 per cent, its development was examined more closely by studying its minimum and maximum points. If the maximum was more than 30 per cent above the average of the values for the first and last year, turnover was classified as "convex". If the minimum was more than 30 per cent below the average of the values for the first and last year, turnover was classified as "concave". If turnover did not exceed or fall below the average by more than 30 per cent, it was classified as "stable". These three life-cycle variables of turnover are illustrated in Figure 3.

Figure 1: Life-cycle variables of increased turnover

Figure 2: Life-cycle variables of decreased turnover

Figure 3: Life-cycle variables of turnover: convex, concave and stable

The classes of turnover are described in Table 1. Letter f implies the first and letter l the last year of examination. Turnover is abbreviated as to.

Table 1: Classes of turnover

The life-cycle variable personnel describes the development of the number of employees. It has been formed in almost the same way as the variable turnover. Many enterprises did not have any employees during the examination period or they only had employees in one year and thus the development of the number of employees could not be investigated. As before, the first concern was to examine the overall change between the starting and ending points. If the total change was within ± 5 per cent, an enterprise was classified either as "stable", "convex" or "concave". If the maximum number of employees exceeded the average of the starting and ending points by over 10 per cent, an enterprise was classified as "convex" and, respectively, if the minimum number of employees fell by over 10 per cent below the average of the starting and ending points, an enterprise was classified as 'concave'. If the number of employees increased by more than 5 per cent, the personnel variable was classified as "increased" and if, in addition, the number of employees increased by more than 50 per cent in any one year, then the personnel variable was classified as "increased with a boost". If the number of employees decreased by more than 5 per cent, the personnel variable was classified as "decreased" and if, in addition, the number of employees decreased by more than 50 per cent in any one year, the personnel variable was classified as "rapidly decreased".

Table 2: Classes of life-cycle variable personnel

The life-cycle variable education describes how an enterprise values education. In this study we constructed an educational level variable (EL) to describe the development of the educational level within enterprises and industry between 1987 and 1995. Equation 1. shows how the educational level was calculated.

f_i=number of persons in the enterprise or industry

x_i=level code of the Finnish Standard Classification of Education

2=education at the second level, second stage other than engineering, natural science, transport or communication

2.5=education at the second level, second stage in engineering, natural science, transport or communication

3=education at the third level, first stage other than engineering, natural science, transport or communication

3.5=education at the third level, first stage in engineering, natural science, transport or communication

5=masters degree but not in engineering, natural science, transport or communication

6=masters degree in engineering, natural science, transport communication

8=post graduate degree but not in engineering, natural science, transport or communication

10=post graduate degree in engineering, natural science, transport or communication

The educational level variable describes the development of the educational level within an enterprise or industry during the examination period.

Enterprises with employees were also classified by the most educated person. It describes the level of education of the highest educated employee in an enterprise. The classification was created using UNESCO's International Standard Classification of Education.

The educational level variable was constructed in three different parts. First, the educational level of an enterprise was compared with the average for the industry in each year. After that the development of the educational level of an enterprise was examined in relation to the average of the industry. The educational level of an enterprise could be about the same, higher or lower than the average, or it could dissect the average value in some year. The second part of the educational level variable was constructed using the most educated person in the enterprise. If the level of education of the most educated person went up, it was interpreted as a jump upwards in the educational level. Conversely, there could also be a jump downwards. The third part of the educational level variable was constructed by examining the development of the educational level of the enterprise independently, in order to establish whether it had increased or decreased. Using these three single factors the enterprises were classified according to how they viewed education. The five classes applied were "investor in education", "non-investor in education", "strong investor in education", "education neutral" and "stable".

Figure 4: Life-cycle variable education

Table 3: Classes of life-cycle variable education

The life-cycle variable enterprise status describes the activity of enterprises in the 1987-1995 period. An enterprise may be born and/or wound up during the examination period. An enterprise may also live right through the whole examination period. An enterprise that was born before 1987 and had not been wound up by 1995, is classified as "normal". The birth of an enterprise can either be real or administrative. An administrative birth occurs if an enterprise continues the activity of another enterprise or if it is born as a result of a merger or de-merger. Enterprise closures were also classified as either real or administrative. An administrative closure occurs when another enterprise continues the activity of a wound up enterprise or if an enterprise merges with another enterprise.

The life-cycle variable business arrangements describes the extent to which an enterprise engages in external arrangements. An external arrangement can be an acquisition, disposal, establishment or winding up of an establishment. On the basis of these activities enterprises were classified as either "diversifiers", "reducers" or "steady". If an enterprise experiences either diversification or reduction more than once, it is classed as "intense diversifier" or "intense reducer".

The life-cycle variable industry indicates whether an enterprise changes its industry during the examination period. A change may take place due to changed production patterns. Since the industry is defined according to the value of production, multi-product enterprises are classified into the industry in which their production is relatively most valuable. So, if dramatic changes occur in the production pattern of an enterprise due to, for example, an acquisition, the industry of the enterprise may change. An enterprise is classed as "faithful" if does not change its industry. An enterprise that enters an industry is classed as an "entrant" and an enterprise that leaves an industry is classed as a "departurer". An entrant does not leave, and a departurer does not return to, the examined industry. If an enterprise changes its position relative to the examined industry, it is classed as a "changer".

Table 4: Classes of life-cycle variable industry

The growth of an enterprise can be divided into internal and external growth. Internal growth does not involve any external activity. It is created by exploiting the unused resources of the enterprise itself. The life-cycle variables of turnover, personnel and education describe this internal growth. External growth is growth by acquisition or merger. It does not necessarily involve any unused resources. The life-cycle variable business arrangements describes the relative external activity.

The life-cycle variables were used to classify enterprises into 12 different groups. The classification was implemented using the deduction tree, illustrated in Figure 5. The life-cycle variable of enterprise status was examined first. On the basis of their status, enterprises were then classified into three different groups: born, wound up or normal. Special effort was made to distinguish between internal and external growth in the grouping of the normal enterprises.

Figure 5: Deduction tree

An enterprise may be born in a particular year, but may not start its activities until a few years later. Beginners and active beginners are enterprises which were born in the period between 1987 and 1995. Beginners may have had either turnover or employees, but not both. Beginners did not really start their activities during the examination period. One might call them "desk drawer firms". Active beginners, on the contrary, did start their activities. They had both turnover and employees.

Closurers and active closurers were wound up during the examination period. Closurers may have had some turnover, or employees - but not both - during the examination period. Their activities had been run down some years earlier and they did not have any significant activity left. Active closurers ran their activities down quite rapidly during the examination period. Prior to closing, they had both turnover and employees.

If a normal enterprise grew or contracted during the examination period, we made a distinction between internal and external growth and internal and external contraction. Enterprises which grew internally had no business arrangements. Their turnover and jobs grew internally, without any external arrangements. Respectively, enterprises whose turnover and jobs contracted without any external arrangements were classified as internally contracted. If growth or contraction of turnover or jobs happened at the same time as external arrangements took place, the growth or contraction was classified as external.

The rest of the normal firms were classified either as boosters, flabbies, stables or inoperative. If an enterprise implemented no business arrangements but its turnover per employee increased, the enterprise was classified as a booster. Respectively, an enterprise whose turnover per employee decreased was classified as a flabby. Enterprises which did execute business arrangements but these had no impact on turnover or employees were classified as stable. The last group were enterprises which had neither turnover nor employees during the examination period. These enterprises were classified as inoperative.

In the 1987 to 1995 period, almost 80,000 firms operated in at least one year in the Finnish manufacturing sector. These enterprises were classified into 12 different groups. Table 5 shows the distribution of the manufacturing enterprises. About 62 per cent of the enterprises were born during the examination period, but only about one quarter of the firms born also really started operating. Almost 17 per cent of the manufacturing enterprises were wound up in the 1987 to 1995 period. About 3 per cent of the enterprises were internally growing, while only 1.3 per cent of them grew externally. Respectively, almost 4 per cent of the enterprises contracted internally, while nearly 2 per cent contracted externally. About 5 per cent of the Finnish manufacturing enterprises intensified their operations while a little over 3 per cent operated inefficiently. Only 1.4 per cent of the enterprises were stable, whereas 2.2 per cent were inoperative.

Table 5: Enterprise groups

Measured by the number of enterprises, the most important group would appear to be the beginners. But the beginners did not have any activity, so their impact on the Finnish economy was insubstantial: if either turnover or jobs are considered instead of the number of enterprises, the picture changes quite dramatically. Percentages of turnover, jobs and number of enterprises by enterprise group in 1987 and 1995 are shown in table 6.

Table 6: Percentages of turnover, employment and number of enterprises by enterprise group 1987 and 1995

In 1987, the most important enterprise group measured by the number of enterprises was that of active beginners. But if we look at turnover and jobs, the most important group was that of boosters. Externally contracting and growing enterprises also accounted for a considerable proportion of turnover and employment in 1987. In 1995, active beginners accounted for over 45 per cent of employment. Boosters' share remained considerable, although it was slightly reduced. As the boosters' share diminished, externally growing enterprises caught up with them.

Table 7 shows how the turnover and employment per enterprise group developed from 1987 to 1995. In 1987, the average turnover of a Finnish manufacturing enterprise was about FIM 6.5 million and they employed, on average, almost 12 persons. In nine years the average turnover of a manufacturing enterprise increased to FIM 8.5 million but the average number of persons employed decreased. In 1995, a Finnish manufacturing enterprise employed, on average, 8.7 persons.

Table 7: Turnover and number of persons employed per enterprise by enterprise group

The high share of new enterprises of both turnover and employment is partly explained by the high number of new enterprises. But, as Table 7 shows, new enterprises have also grown very fast. In 1995 their average turnover was almost FIM 14 million and they employed, on average, just under 14 persons.

The largest enterprises, on average, were the externally growing and contracting enterprises. It is reasonable to assume that an enterprise has to be of a certain size to be able to engage in any significant business arrangements. Internally growing and contracting enterprises were considerably smaller than the enterprises which engaged in business arrangements. Business arrangements appear to have a substantial impact on the size and volume of an enterprise. Externally growing enterprises grew significantly, doubling their turnover and jobs. The average number of employees of externally contracting enterprises decreased drastically, while their average turnover contracted only slightly. It seems that externally contracting enterprises were rationalising their activities by reducing the number of employees.

The average size of booster enterprises diminished, while their average turnover increased. Nevertheless, the average size of a booster enterprise was still bigger than the average size of a Finnish manufacturing enterprise.

Enterprise groups by industry are shown in Appendix 2. Our data comprised all enterprises that operated in the manufacturing sector during the 1987-1995 period. Examined by industry, an enterprise that changed its industry occurs more than once in our data, i.e. in all the manufacturing industries in which it had operated. That is why the number of enterprises in Appendix 2 is greater than that in Table 5.

Looking at the distribution of enterprises between enterprise groups, the Finnish manufacturing sector appears quite homogeneous. The picture the enterprise grouping gives of the Finnish manufacturing industry is quite coherent. The comparison between industries is based purely on enterprise grouping and should, therefore, be regarded with caution. This applies, in particular, to oil refining. Oil refining is a very minor industry in Finland, with only a few enterprises operating in it, so the indicators for this industry became distorted due to the small number of enterprises. That is why our study ignored oil refining industry. I will next discuss a couple of interesting industries in more detail.

The structure of the enterprise field changed more rapidly in food manufacture than in manufacturing general. The numbers of active beginners and closurers were greater in food manufacture than the corresponding average numbers for the entire manufacturing field. This may indicate that the wastage of enterprises was high in the food industry.

Textile and clothing manufacture is one of the most regressive industries in Finland. Since the former Soviet Union collapsed and Europe faced a recession in the late eighties, Finnish textile and clothing industry has been declining steadily. Fewer than average new enterprises were born - and also started activities - in the textile and clothing industry. The proportion of enterprises that might grow, either externally or internally, was smaller than the average for the manufacturing field. Respectively, the proportion of enterprises that were actively wound up was higher than in manufacturing in general.

Pulp, paper and paper products manufacture has traditionally been the backbone of Finnish manufacturing industry. Examined by enterprise group, paper industry seemed to be doing fine. There were more active beginners, boosters and internally and externally growing enterprises than in manufacturing in general.

There were substantially more externally growing enterprises in the manufacture of chemicals and chemical products than in the manufacturing sector in general. The chemical industry underwent rapid growth during the 1987-1995 period. However, most of this growth was attributable to the expansion of a few large enterprises through business arrangements. These large enterprises dominated the markets and possessed considerable market power in relation to the general development of the industry.

The manufacture of electrical products and instruments is one of the most important exporting sectors in Finland. Finland can claim to be among the leading countries in the world in the manufacture of electronics, particularly telecommunications equipment. There were more internally and externally growing enterprises in the electronics industry than in the manufacturing sector in general. This could indicate that the electronics industry will go on growing and its strong impact on the Finnish economy will continue.

The industries serving the manufacturing sector are technical services, data processing services and research and development. These three industries differ from the rest of the manufacturing sector. In Finland, like in the rest of Europe, the services sector is growing rapidly. Its importance to the economy as a whole is increasing. The services sector is already a significant employer and there are hopes that it will employ even more people in Finland in the future. However, the enterprise grouping indicates that there were a lot beginners unable to start their activities in this sector. These may be have been one-man enterprises. There were fewer externally growing enterprises in the services sector than in manufacturing in general. This is not surprising, but what is confusing is that there were also fewer internally growing enterprises and boosters in the services sector than in manufacturing in general. Service enterprises did not wind up their activities, either, since the number of contracting and closing service enterprises was smaller than the average in manufacturing. The beginners in the service industries did have the potential to grow, but for some reason they did not manage to do so.

In this study we had an exellent opportunity to explore an unique database. While it gave us many chances it also raised several problems. The WEDB was extremely challenging to work with. The streght of the WEDB was the opportunity to follow up one particular enterprise over time. At longest we were able to follow up the development of enterprises for nine years. By following the development of the enterprises of the industry we were able to get deeper picture of the development of the industry. The WEDB let us behind the aggregated figures. Most of the Finnish enterprises are small- or medium-sized enterprises. But when one examines the development of the whole industry we often see only the development of the few large enterprises, which dominate aggregated numbers. With the WEDB we were able to solve this problem.

The life-cycle variables were result from the experiment of the WEDB. We did not have any remarkable theoretical base nor did we have many references. The lack of references was due to fact that there has not been such a databases available. Our main goal was to find out what opportunities this unique data could provide. We have not proved any of these results and therefore one should regard the results with caution. In the future it would be interesting to make some case studies around the life-cycle variables.

We tested different methods for creating the enterprise groups like factor-, cluster- and correlation anylyzes, but none of them were appropriate for this matter. One of the obstacles was the nature of the WEDB variables. The WEDB consisted only few quantitative variables, such as turnover, number of employees and establishments. Most of the variables of the WEDB were qualititative. Therefore all the life-cycle variables were qualitative not quantative.

Problems will raise if one does not keep in mind the nature of the WEDB. For examle the significance of the new enterprises may be overrestimated, since we did not make any difference between the real new enterprises and administrative openings. Administrative openings distort the examination of the enterprise groups. If the opening of the new enterprise was administrative, then it propably had already sustainably activities. While really new enterprise had to create its activities; hire employees, introduce them to the new assignment, create markets, etc.

In this study we only scratch the surface of the WEDB. We tried to exploit WEDB as many ways as possible. In addition to the life-cycle variables and enterprise grouping formulation we focused on new enterprises and their development. We made a cohort study for new enterprises born in years between 1987 and 1991. We were able to follow up these enterprises for five years. We examined the birth and death rates and the development of different cohorts. Since WEDB has the employee link, we were able to find out from where the employees of the new enterprise came from. We knew their age and education and principal activity. With the help of the employee link we defined on what the knowledge of the new enterprises was based on.

One may consider why we did not include such a variables like investments, R&D expenditurers and innovations in to our study. Indeed these factors would give more precise picture about the evolution of the industry. But we did not have comprehensive data about these factors. Statistics Finland do combile statistics about the R&D expenditurer and innovations, but only for the limited group of enterprises. This information was available only for high-tech industries and large enterprises, while our interest was focused on SME's.

Appendix 1: Manufacturing industries and industries that support them.

11 Food, beverage and tobacco manufacture

12 Textiles manufacture

13 Wearing apparel, leather goods and footwear manufacture

14 Wood and wood products manufacture

15 Pulp, paper and paper products manufacture

16 Publishing and printing

17 Furniture manufacture

18 Chemicals and chemical products manufacture

19 Petroleum and coal products and nuclear fuel manufacture

21 Rubber and plastic products manufacture

22 Glass, clay and stone products manufacture

23 Basic metal industries

24 Fabricated metal products manufacture

25 Machinery and equipment manufacture

26 Electrical products and instruments manufacture

27 Transport equipment manufacture

71 Technical services

72 Data processing services

86 Research and development

Appendix 2: Enterprise groups by industry