Geography Reference
In-Depth Information
temporary peak, or it could mean they experience a cluster ef ect for R&D expenditure
if R&D expenditure increase levels remain high over time. Yet a further interpretation is
that they are inei cient spenders of R&D investment that is only associated with margin-
ally better overall performance than is the case for non-collaborators.
Focusing on the collaborators' side, Table 11.1 shows that for some indicators of eco-
nomic performance, collaborators in clusters perform better than collaborators in non-
clusters. Collaborators in clusters tend to have superior performance regarding higher
R&D as a share of turnover in 2003 (19 per cent compared to just 13 per cent), a higher
number of i rms recording an increase in R&D expenditure between 2000 and 2003, more
i rms announcing patents, and a greater number of patents announced in both 2000 and
2003. Clustered collaborators tend to be bigger than non-clustered collaborators
and have a higher number of i rms that increased their employment size between 2000
and 2003. However, while mean turnover of the two sub-groups is similar (£5 million
in 2003), 76 per cent of non-clustered collaborators increased their turnover between
2000 and 2003. It can be concluded that collaborators in clusters tend to be superior on
the R&D and patenting input side. However, with regard to the output side the higher
investments in inputs do not benei t i rm performance.
Turning attention to the non-collaborators, it is perhaps surprising that a signii cant
number of non-collaborators consciously locate in clusters. However, the data reveal
that non-collaborators in clusters perform better than their counterparts in non-clusters
in just some innovation indicators (employment increase and new products/services). In
contrast, non-collaborators in non-clusters spend more on R&D, and there is a higher
proportion of i rms that increased their R&D expenditure and turnover growth in the
2000-03 period. It can be argued that clustering can provide competitive advantage to
non-collaborators as the non-collaborators that co-locate in geographical proximity are
smaller in size (an average of 19 employees compared to 41 for non-clustered i rms and
£2 million average turnover compared to £5 million for the isolated non-collaborators).
This seems to be a coni rmation that economic spillovers are available as even neo-
classical literature predicts in cluster settings. However it must be questioned whether
notions like 'collaborators' and 'non-collaborators' would enter their econometric radar.
Nevertheless it would be consistent with their 'knowledge spillovers' attraction ef ect
since small i rms would be more rational in utility-maximisation terms to seek out such
spillovers than more self-contained larger ones. In this respect, diseconomies of scale
ef ects might be discounted. However, as shown in Table 11.1 and, later, Table 11.2, this
may help innovation but does not necessarily benei t i rms' performance.
Table 11.2 presents correlations measuring associations between performance vari-
ables and collaboration ef ects and perceptions. Thus, in a novel way we compare
expected outcomes of collaboration with realised i rm performance on these variables.
This helps us get at motivations for collaboration. The results show the following: i rst,
collaborators display relatively high and statistically signii cantly improved market share,
wholly new product or service (to i rm and, crucially, market) innovation, and to a lesser
extent, turnover and patenting improvements (consistent with Table 11.1). Interestingly,
performance indicator expectations from collaboration coincided somewhat with the
actuality, measured on the variable 'capacity to introduce new products or services', but
actual performance on that variable was lower and not statistically signii cant. This may
be interpreted as a kind of 'over-optimism' variable where reality produced less from
