For a long time, there has been interest in where business sector research and development (R&D) is located and in factors that drive localization. The reason for this is that localization can be expected to increase access to new knowledge, promote growth and create highly skilled, well-paid jobs. This means that for policy it is important to know why business R&D is located in a certain country or region.

To be able to describe the pattern and development of business R&D at an appropriate regional level, R&D activities within business are distributed in functional labor market regions (FA-regions), i.e., economically integrated regions where people tend to both be residents and work. The main purpose of this study is to analyze which factors are important for the localization of business R&D within enterprise groups at the FA regional level in Sweden. The study is thus a complement to Tillväxtanalys (2022), which studied determinants of the localization of R&D within Swedish-owned multinational enterprises at the national level, that is, to different countries (including Sweden).

Business R&D in Sweden is strongly concentrated in a few enterprise groups and industries

R&D within the Swedish business sector is carried out by a small number of enterprise groups. In most cases, these enterprise groups are multinationals, Swedish multinational enterprises (MNEs) or foreign-owned firms; as much as 93 percent of the R&D that takes place in the Swedish business sector is conducted by MNEs. Among the enterprise groups that report R&D expenditures, the expenses are also concentrated in a few enterprise groups; in 2019, one percent of the enterprise groups with the largest R&D expenditures (roughly six enterprise groups) carried out 57 percent of the total R&D expenditures in the Swedish business sector.

A few industries contribute to the bulk of the R&D investments conducted in the Swedish business sector. Most R&D is carried out in the motor vehicles and computer programming and data consulting industries and in recent years significant increases in both industries have been observed. The sharply rising R&D expenses in computer programming and data consulting are mainly a result of the very R&D-intensive telecom company Ericsson being reclassified in 2015 from an industrial company to a service company. Apart from this, however, it is an industry where there has been a considerable increase in R&D expenditure.

. . . and to a few regions

Even at the regional level, R&D expenditures are highly concentrated. The two largest FA-regions, Stockholm-Uppsala and Gothenburg, account for over 64 percent of the R&D investments conducted in the Swedish business sector. If we also add Malmö-Lund and Linköping-Norrköping, we reach almost 83 percent. Notably, is also that the concentration in a smaller number of regions has been relatively stable over time. An interesting and remarkable change is the significant increase in R&D spending that we can see in the Gothenburg region which is driven by the automotive industry. Currently, R&D expenditure in Gothenburg is almost on par with that in Stockholm-Uppsala, where we actually observe a decrease. In the four largest FA-regions, there are considerable R&D investments in several industries, while in other regions, a single industry usually dominates. It is noteworthy that within the business sector there is no region in northern Sweden with particularly significant R&D expenditures.

Potential knowledge transfer between enterprise groups within the same region is an important localization factor

Proximity between different activities makes the transfer of knowledge between adjacent enterprise groups easier. This is because certain knowledge can be difficult to communicate; it is implicit and cannot be codified (tacit knowledge). However, there are transfer benefits from proximity. A channel for knowledge transfer is when employees move between enterprise groups and bring the knowledge with them to the new employer. Such knowledge transfer is facilitated if enterprise groups are in the same FA-region. The econometric analysis shows that enterprise groups are eager to locate their R&D in the same region where other enterprise groups have located their R&D. This applies in particular to cases when such groups are active in industries other than their own (interindustry spillovers).

Potential knowledge transfer of R&D from academia to the business sector within the same region is another

A corresponding reasoning as above can be used in regard to knowledge transfer between academia and the business sector. In regard to R&D expenditure in academia, Stockholm-Uppsala dominates with just under 45 percent of the total R&D expenditure within academia, followed by Gothenburg and Malmö-Lund with just over 15 percent each. The pattern that R&D is predominantly carried out at large universities has been stable over time although after the turn of the millennium there has been a strong expansion in the number of students, above all at smaller and medium-sized colleges and universities. In the empirical analysis, we find that enterprise groups that conduct R&D in the Swedish business sector have a tendency to locate their R&D in regions where there are universities with significant R&D, especially if these universities conduct R&D in natural sciences and technology.

Colocation of R&D and manufacturing within the same enterprise group is an additional one

Colocation of R&D and manufacturing is expected to lead to better coordination, control and overview as well as to facilitate communication and knowledge sharing between these operations. Our analysis shows that manufacturing enterprise groups tend to place both their R&D and their manufacturing in the same FA-region, indicating that proximity is important for a mutual exchange of knowledge between these activities.

. . and so is the availability of skilled labor

A broad knowledge base and availability of qualified R&D personnel are factors often noted as essential for the localization of R&D. R&D investments in the business sector are expected to be attracted to regions that have good access to qualified labor with the ability to perform R&D (potential R&D employees), for example civil engineers in natural sciences and technology or in information technology or research graduates. The econometric analysis in this study confirms this. In the analysis we use different measures of potential R&D employees and it turns out that enterprise groups are particularly inclined to locate their R&D activities in regions with a high proportion of such workers. Universities that carry out considerable R&D, especially in natural sciences and technology, provide the business sector with basic and applied research as well as competent research-trained R&D personnel, not least in the region where they are located.

In sum, we conclude that knowledge transfer between enterprise groups within the same region and between enterprise groups and universities within the same region (external agglomeration) appears to play a significant role in the localization of business R&D. In addition, knowledge transfer between manufacturing and R&D within the same enterprise group seems to be facilitated by whether these operations are located in the same region (internal agglomeration). This is indicated by the fact that colocation of such activities is common within manufacturing enterprise groups in the Swedish business sector. Finally, the endowments of potential R&D employees in FA-regions seem to be important for the regional localization of R&D activities in the business sector.

How the analysis was carried out

In the report, a model is estimated on data on R&D expenditure at the enterprise group level in the Swedish business sector in 60 FA-regions between 2001 and 2019. The R&D data come from the survey that Statistics Sweden conducts every two years and where firms are asked how much expenditure they have on R&D at the county level. To obtain data at the FA-region level, we have to distribute the data at the county level to the FA-region level. For this purpose, we use data on employees' occupations (R&D occupations) at the enterprise group level in various FA-regions from the occupational register. We also use the occupational register to determine the extent to which there are manufacturing operations (share of the employees employed in manufacturing occupations) and R&D activities within an enterprise group in the same FA-region.