combination of growth in inputs and efficiency improvements (Solow, 1957). The underlying
qualitative nature of economic development in terms of the variety of sectors or the variety of
technologies has been addressed only rarely. One can distinguish three types of relationships
between variety and economic development (Frenken et al., 2005, 2006). The first approach
centres on variety, knowledge spillovers and growth, which has become a central theme in what
is called ‘new growth theory’. It has been argued that, apart from spillovers occurring between
firms within a sector, spillovers also occur between sectors, which are commonly referred to as
‘Jacobs externalities’, after Jacobs (1969). A second way to relate variety to regional economic
development is to view variety as a portfolio strategy to protect a region from external shocks in
demand (Essletzbichler, forthcoming). In this context, one also speaks of regional diversification
analogous to corporate diversification as a risk-spreading strategy. A third type of relationship
between variety and economic development concerns the long-term effect of variety on the
economic system. An economy that does not increase the variety of sectors over time, will suffer
from structural unemployment, and will ultimately stagnate. In this view, the development of new
technologies and sectors in an economy is required to absorb labour that has become redundant in
existing sectors (Pasinetti, 1981, 1993; Saviotti and Pyka, 2004). This process underlying long-
term growth has major geographical implications, when new sectors emerge in other areas than
the ones where old sectors are located. This would imply that labour becomes redundant
primarily in areas where the old sectors are concentrated, while new employment is primarily
created in new areas. This imbalance may be counteracted by labour migration from old to new
areas and by firm migration in the opposite direction.
Although many empirical studies have analysed the effects of variety on regional growth in
the past decade or so, some methodological issues in empirical research remain. First, the
measurement of variety is not trivial. For example, one would like to distinguish between related
variety underlying spillovers and unrelated variety underlying the portfolio effects (Frenken et
al., 2005; Essletzbichler, forthcoming). Second, explaining regional phenomena requires a careful
econometric specification so as to allow different effects to take place at different spatial levels of
aggregation. For example, the rate of regional growth or the rate of regional information
technology adoption can be made dependent on the rate of growth in neighbouring regions
through the use of spatial autocorrelation econometrics (Essletzbichler, forthcoming; Bonaccorsi
et al., forthcoming).
14