Persistence and Change of Regional Industrial Activities: The Impact of Diversification in the German Machine Tool Industry

Abstract

The paper investigates the stability and change of regional economic activities in the long run. As the unit of analysis, we selected the machine tool industry in West Germany for the years 1953–2002. We spot a strong variance in the activities between different regions. These differences are relatively stable over time, and the regional activities are rather path-dependent. Nevertheless, the paper also identifies changes in the level of activities. As the main driving factors for these developments, we examine the effect of changing regional degrees of diversification over time. We find that those regions which generally broaden their scope of activities have a higher likelihood to grow than regions which are specializing. Furthermore, diversification into totally new technological and product fields is only beneficial under specific circumstances based on technological and market developments. Hence, in most cases, a broad diversification is superior to one focusing on new state-of-the-art technological fields.

Notes

We also examined the effect of related variety, but for different reasons mentioned later in the paper, we are not able to follow this line of research.

For the years 1949 and 1950, only one catalogue was issued; therefore, the values for those 2 years are identical. The catalogue for 1952 is not accessible; therefore, the values for 1952 are approximated by the respective values in 1951. Because of these missing data, our analysis starts in 1953.

The location corresponds to the headquarter and the single production site of the firms, respectively. Given the small- and medium-sized company structure and a plant per firm ratio of 1.1 within the broader classification of “manufacturing of metal working machines” (SIC/SYPRO-No 3220; Fleischer, 1997), we minimize the risk of wrongly assigning the firms' activities.

“Raumordnungsregionen” represents regional planning districts and hence functional regions which cover commuter flows and consist of several “Landkreise” (counties).

This classification goes beyond the mere differentiation of metal forming and metal cutting products. Products are hierarchically categorized into six technology groups on the first level and another 39 product classes on the second level based on the VDMA scheme in the buyer's guide.

These factors include, for example, the collapse of the Soviet Union (Schwab, 1996) as one of the major customers, the inability to act as a global player because of the small- and medium-sized company structure within the industry and the adherence to “over-engineered” machines (Carlsson, 1989; Hirsch-Kreinsen, 2000).

(Computerized) NCs operate the tools automatically with the help of computerized control tapes which contain pre-programmed sets of commands. All process relevant actions, such as machine feed and speed, selection of a particular tool, distance and direction are carried out automatically as opposed to a skilled worker continuously supervising the production processes. By changing the set of commands and thus the processing instructions, an entirely different operation can be initiated (Arnold, 2003).

An FMS is defined as “a grouping of CNC machines (often including machining centers) which is fed by an automatic materials handling and transfer system meeting the needs to manufacture small and medium batches of a variety of products” (WS Atkins Management Consultants, 1990, p. 53). FMCs and machining centres are also based on CNC technology, but contain fewer machine tools than FMS.

This development was accompanied by significant price drops, which made the technology feasible also for smaller jobshops (Schwab, 1996). Moreover, the given economic situation of rebuilding the European industrial equipment after WW2 held back the immediate shift towards numerically controlled machinery (Guenther, 2009).

For a detailed description of the adaptation process and how numerical controls were chosen in favour of alternative concepts, see Guenther (2009).

Even though the new control technique was also integrated into single-function machinery, we concentrate our analysis on multi-functional machinery as the implementation of the changing market requirements is most remarkable and complete in these machining concepts.

Schwab (1996) explicitly mentions Stuttgart and the Rhine-Main area as two highly active regions.

Fleischer (1997) chose a similar approach by investigating whether the rank positioning of the top firms remains constant over time. His analysis is though limited to the time period 1990–1994, but he nevertheless found a high mobility of firms within the ranking during these years.

In addition to the analysis based on the absolute number of firms, we also calculated all rank correlations based on the number of firms weighted by the population density within the region. The results remain the same and, hence, only the results for the absolute number of firms are presented in the following.

The overall correlation coefficient between every time period t and the subsequent period t + 1 even amounts to 0.9876.

We did the same rank correlation analysis for each individual year and observed a minimum value of 0.77 (comparing the rankings of 1954 and 2002) for the correlation coefficient, which is significant on the 1% level.

For a similar approach based on start-up data, see Fritsch and Mueller (2005).

We also applied Tobit and truncated regressions (with −1 and 1 as lower and upper limits for regressions with RCA as dependent variable and 0 and 64 for regressions with rank change as dependent variable) and generalized least square regressions. We receive identical signs for all coefficients and some slight deviations with regard to the significance of the variables, but these results do not contradict the findings of the robust OLS regressions. We only report the results of the robust OLS regressions in the following.

Despite some problems in calculating related and unrelated varieties, we tested them as well, but related variety is never significant and unrelated variety delivers the same results as change in product categories and change in HHI. We omitted the results here because of the problems in calculating these variables (many zeros as nominator or denominator).

Even though numerical controls required a change within the entire product architecture, specialized knowledge in individual processing procedures was nonetheless still valuable.

Entry into MFPs before the second period was not possible, because these categories simply did not exist before the 1960s.