Evolutionary economic geography and growth:

technological relatedness and regional branching

Ron Boschma

Utrecht Universityhttp:// econ.geo.uu.nl/boschma/boschma.html

DIMETIC course

Pecs July 5, 2010

structure of lecture

• technological relatedness and related variety

• 4 empirical applications in evolutionary economic geography

1. spatial externalities: related variety2. labour mobility: inflow of related labour3. entrepreneurship: experienced entrepreneurs4. regional diversification: entry and exit of industries

• literature

- Frenken et al. (2007)- Boschma, Eriksson and Lindgren (2009)- Boschma and Wenting (2007)- Neffke, Henning and Boschma (2009)

1. technological relatedness and related variety

• firms differ: firm-specific competences

• variety as key driver of regional growth: the more, the better

• knowledge will spill over to other firms now and then, and geographical proximity is an enabling factor (Jane Jacobs)

• effective learning requires cognitive proximity between firms

• but too much cognitive proximity may be harmful (lock-in)

• need for technological relatedness and related variety to enable knowledge spillovers

• regional level might be crucial: knowledge spillovers are often geographically bounded

2. spatial externalities: related variety

• Jacobs’ externalities and regional growth?

• what matter for regional growth: sectors that are technologically related in a region

• the higher related variety in a region, the higher regional growth: effective knowledge transfer requires some but not too much cognitive proximity between sectors in a region

• Frenken et al. (2007) for the Netherlands, confirmed by studies in other countries (e.g. Italy, Great Britain, Finland)

2. spatial externalities: related variety

• regional growth: may also depend on extra-regional knowledge flows: non-local linkages bring new variety into the region

• Boschma and Iammarino 2009, Economic Geography 85 (3), 289-311: study on related variety, trade linkages and regional growth in Italy

• it is not inflows of extra-regional knowledge per se that matters for regional growth

• inflows of extra-regional knowledge related (but not identical) to the knowledge base in a region that do matter for regional growth

• this concerns new knowledge that can be understood and exploited by related sectors in the region and, thus, be transformed into regional growth

3. labour mobility: inflow of related labour

• labour mobility key mechanism through which knowledge diffuses: however, no attention paid to relatedness

• Boschma, Eriksson and Lindgren (Journal of Economic Geography, vol. 9 (2), 2009): effect of labour mobility on plant performance through relatedness

• (1) at the plant level: the more related the skill portfolio of a plant is, the higher plant productivity growth

• (2) through labour mobility: inflow of new skills that are related to the knowledge base of the plant had a positive effect on plant performance (in contrast to inflows of identical skills)

• (3) extra-regional linkages: labour mobility across regions has only a positive effect on plant performance when this concerns new employees with skills that are related (but not similar) to the existing set of skills at the plant level

relatedness at the plant level

• human capital may matter, but …. does a particular portfolio of skills enhance intra-firm learning and performance of firms?

- need to distinguish between similar, related and unrelated set of skills at firm level

- related set of skills (as opposed to similar and unrelated skill portfolio’s) beneficial for firm performance: not too much cognitive proximity, and not too little cognitive proximity

relatedness and labour mobility

• labour mobility may matter, but …. does the inflow of particular types of skills enhance intra-firm learning and performance of plants?

- too much reliance on intra-firm skills may be harmful, especially when this concerns a portfolio of similar skills (as opposed to a related set of skills at the firm level): need for inflow of new skills to avoid lock-in

- however, absorptive capacity is required to understand and implement new skills at the level of the firm

- therefore, effect of labour mobility depends on the types of skills that are brought into the firm by new employees, and the extent to which these newly recruited skills add to the existing set of skills at the firm level

- hypothesis: inflow of new skills related (but not similar or unrelated) to the skill base of plant enhances intra-firm learning and firm performance

intra- versus inter-regional labour mobility

• labour mobility may matter, but …. does the inflow of skills from the same region enhance intra-firm learning and performance of firms?

- too much reliance on intra-regional labour inflows may also be harmful: need for inter-regional labour inflows to avoid lock-in

- however, this effect depends on inflows of types of skills:

- intra-regional flows: when these concern employees with similar skills, the problem of cognitive lock-in may worsen

- the more unrelated the inflow of new skills, the more need for intra-regional mobility, in order to solve problems of communication

- inter-regional labour mobility will enhance firm performance when these concern inflows of related skills

empirical study of Sweden: data

• data source: micro database ASTRID that connects individuals (e.g. education, skills) to workplaces (e.g. sector, location, employees, value added) for the whole Swedish economy

• excluding young workers (<25 years), part-timers (< half time) and low-skilled job movers: total of 101,093 job moves in 2001

• only workplaces that registered inflows of skilled employees in 2001: total of 17,098 workplaces

• intra- versus inter-regional mobility: 108 labour markets areas

empirical study of Sweden: variables (1)

• dependent variable: labour productivity growth (growth value added per employee) at the plant level 2001-2003

• control variables: plant size, urban size, industries and R&D

• intra-plant data: educational background of employees at 3 digit level (95 categories) as a proxy for skill portfolio

- inhouse similarity: inverted entropy at 3 digit level

- inhouse related variety: weighted sum of entropy at 3 digit level within each 2 digit class in 2001

- inhouse unrelated variety: entropy at 1 digit level

empirical study of Sweden: variables (2)

• inflows of skills at plant level: sector-specific work experience of employees at 5 digit level (753 industries)

• inflow similar skills: number of new employees with background in the same 5-digit sector as the plant

• inflow related skills: number of new employees with background in all 5-digit sectors that share the same 3-digit sector as the plant, excluding those with a background in the same 5-digit sector as the plant

• inflow unrelated skills: number of new employees with background in all other 5-digit sectors

• inflow intra-regional versus inter-regional (108 labour market areas)

Inflow related skills

plant inflow labour

3 digit 4 digit 5 digit

311

3111

31111 31112 31113

311

3111

31112 3111331111

empirical study of Sweden: main results (1)

• control variables: strongest effect of plant size (negative)

• skill portfolio plant:

- educational level per se: positive effect, but …- inhouse similarity: not significant- inhouse related variety: strong positive effect- inhouse unrelated variety: not significant

• inflow types of skills:

- labour mobility of skilled people per se: negative effect, but ….- inflow of similar skills: negative effect- inflow of related skills: strong positive effect- inflow of unrelated skills: not significant

empirical study of Sweden: main results (2)

• intra- versus inter-regional flows of types of skills:

- intra-regional labour mobility: not significant- inter-regional labour mobility: a negative effect

- (1) inflow of similar skills: negative effect, no matter whether it concerns intra-regional or inter-regional labour mobility

- (2) inflow of related skills: positive effect, no matter whether it concerns intra-regional or inter-regional labour mobility

- thus, inter-regional labour mobility turns from a negative into a positive effect when it concerns inflows of related skills. Inter-regional labour mobility has only a positive effect on productivity growth when it concerns new skills that are complementary to the plant

- (3) inflow of unrelated skills: the effect of inflows of unrelated skills turns into a positive effect when it concerns intra-regional mobility, while inter-regional flows of unrelated skills have a negative impact on productivity growth

relatedness and labour mobility: research agenda

• need for dynamic analysis: skill portfolio of plants

• to get more sophisticated measures of relatedness

• to link this study to topic of regional branching: what kinds of labour mobility are required to diversify regions:

(1) related labour(2) places with related industries(3) mostly local labour

• need for certain types of labour may differ between different stages of industry/cluster formation: e.g. how to avoid lock-in?

4. entrepreneurship: experienced entrepreneurs

• through entrepreneurship, new industries emerge, but these do not start from scratch: relatedness is again crucial

• empirical study on the spatial evolution of British automobile sector 1895-1968 (Boschma and Wenting, Industrial and Corporate Change, 2007, 16 (2): 213-238):

- experienced entrepreneurs (with relevant knowledge from related industries) are crucial for first stage of the industry lifecycle

- related knowledge and skills are transferred from old sectors (engineering, cycle making, coach making) to the new (automobile) sector: this increased their survival rate, in comparison to other types of entrepreneurs

- British regions endowed with these related industries also had a higher probability to develop the new automobile industry: effective knowledge transfer from the old to the new requires relatedness: regional diversification or branching

5. regional diversification: entry and exit of industries

• many case studies on regional diversification or branching, but no systematic study

• creative destruction (Schumpeter): rise and fall of industries in regions

• depending on the degree of technological relatedness with existing industries in a region?

• two research questions

(1) rise of new industries: do industries have a higher probability to enter a region when these are technologically related to pre-existing industries in a region?

(2) fall of existing industries: do industries have a lower probability to exit a region when these are technologically related to pre-existing industries in a region?

regional diversification

• long-term analysis of 70 Swedish regions 1969-2002 (Neffke, Henning and Boschma, 2009)

• first, we needed an indicator to measure the degree of technological relatedness between industries

• co-occurrence analysis to assess inter-industry relatedness

• co-occurrence of products that belong to different industries in the portfolios of manufacturing plants: economies of scope

• control for profitability and size of industries

• Sweden: data on manufacturing plants concerning the products produced 1969-2002: product data assigned to standard industry codes (a total of 174 6-digit industries)

• we calculated the average degree of technological relatedness between each pair of industries in the period 1969-2002

regional diversification

regional diversification

• in order to determine whether entries and exits of industries in a region are affected by the degree of technological relatedness with other industries in the region, we aggregated our data to 70 Swedish regions

• then, we developed an indicator to determine how close an industry is to all other industries in a region: technological closeness

• defined as the number of industries in a region that an industry is technologically related to above a certain threshold

illustration of closeness indicator

regional diversification

• we estimated the entry probability of an industry in a region, and how that is affected by the closeness of the new industry to a region’s portfolio

• as expected, industries that are technologically related to pre-existing sectors in a region have a higher probability to enter the region

• we also estimated the exit probability of an industry in a region, and how that is affected by its closeness to the region’s portfolio

• as expected, unrelated sectors have a higher probability to exit the region

entry and exit probabilities

average industry space in Sweden 1969-2002, with the evolution of the production structure of Linkoping

regional diversification

• thus, systematic evidence: regions tend to diversify by branching into new industries that are related to their current industries: path dependent process

• the same applies to the destruction process: existing industries tend to disappear from regions when these are not embedded in current industries

• many open questions: e.g. through which mechanisms (entrepreneurship, labour mobility, pipelines, etc.) do new industries connect to existing industries?

• what are the effects of the entry of new industries that bridge two technology clusters in a region?

• what are the regional implications of changes in industry space?

• does industry space differ from country to country?

regional diversification

Evolution of Swedish regional production structures with closeness measure (the number of related industries with a RR over 0.25)