39
Recent Research on Industry Clusters ECON 4480 State and Local Economies 1

Recent Research on Industry Clusters ECON 4480 State and Local Economies 1

Embed Size (px)

Citation preview

Recent Research on Industry Clusters

ECON 4480 State and Local Economies

1

Recent Research• “The Economic Performance of Regions”,

Michael Porter, Regional Studies, 2003.• “Is There Evidence to Support Porter-type

Cluster Policies?”, Frank Mcdonald et al, Regional Studies, February 2007.

• “The development of industrial clusters and public policy”, Frank Mcdonald et al, Entrepreneurship & Regional Development, November 2006.

2

“The Economic Performance of Regions”• Examines differences in regional economic

performance.• Assigns industries to one of three categories:

– Local industries: provide goods and services primarily for the local market.

– Resource dependent industries: locate where resources are found; produce for national and international markets.

– Traded industries: sell goods and services across region boundaries and to other countries.

3

BEA Economic Areas

4

“The Economic Performance of Regions”• Identified these industry types for all the BEA

Economic Areas.

5

“The Economic Performance of Regions”• Observations:

– Local industries dominate employment and employment growth: 2/3 of jobs, good job growth rate.

– Traded industries dominate average wages: nearly double that of local industries, much higher rate of growth.

– Average wages for the region are strongly tied to wages in the traded industries.

6

“The Economic Performance of Regions”

• The regional average wage can be boosted by a) increasing wage levels in all industries, or b) shifting the mix of industries towards traded industries.

• Question: how important is the mix effect compared with the level effect?– The difference between regional wages and

national wages can be defined as the sum of the mix effect and the level effect.

7

“The Economic Performance of Regions”• The mix effect measures the contribution to the

regional average wage of the region’s industry mix. To measure the mix effect, hold the average wage constant at the national average and apply the difference between the national mix and the regional mix of traded versus local industries.

• The level effect measures the contribution to the regional average wage of a region’s industry wages compared with the national average. To measure, hold industry mix constant and apply the difference between regional and national wages by industry.

8

“The Economic Performance of Regions”• Results: the level effect dominates, accounting for

79% of difference in average wages across regions, with the mix effect accounting for just 21% of the difference.

• Implications: the key to higher overall wages is to support higher wages in the traded industries. Increasing the mix of traded v local industries is much less important.

9

“The Economic Performance of Regions”• Next, Porter defined sets of industry clusters for all

the Economic Areas.– Used correlations of employment by industry

across areas as a first cut. Strong correlation between two industries suggests common cluster.

– Refined correlations with judgment, knowledge of shared linkages, and finally, the US input-output model.

– Result: defined 41 overlapping clusters for the national economy.

10

“The Economic Performance of Regions”• Question 1: are regional economies becoming

more specialized?• Regional cluster specialization by a state or area

is thought to lead to higher productivity and higher wage growth.

• Cluster specialization with a region can be measured by growing inequality of employment dispersion among industry clusters. The more unequal, the more specialized.

11

“The Economic Performance of Regions”

• Question 1: are regional economies becoming more specialized?

• Answer: greater specialization is associated with higher wage growth rates by state. States with higher wage growth tend to have more specialized industry clusters.

12

“The Economic Performance of Regions”

• Question 2: among clusters, which is more important for determining the regional average wage: the mix effect or the level effect?

• Answer: the level effect is far more important, accounting for 76% of the variation across regions, compared with 24% for the mix effect.

• The level of wages among clusters can differ because of differences in sophistication, productivity, and subcluster structure.

13

“The Economic Performance of Regions”

• Implications: the mix of traded industries cannot be moved enough to make a difference. Rather, competitive standing is more important; increasing productivity will cause higher wages for the given mix of clusters.

• Upgrading the productivity of all important clusters is likely to be much more effective than policies that focus on shifting the mix to more desirable clusters.

14

“The Economic Performance of Regions”

• Other findings:– Lack of cluster diversification does not seem to

affect wage growth and employment growth.– The stronger the cluster, the higher the regional

wages. Cluster strength is measured by the proportion of regional employment in traded clusters with LQs of 0.8 or higher.

15

“The Economic Performance of Regions”• Conclusions:

– Traded clusters drive regional wage growth.– Raising productivity in traded clusters is the key to sustained

regional wage growth. Cluster mix is much less important.– Traded cluster strength strongly effects regional performance.– Regional policies should focus on traded clusters, as these

clusters drive wage growth and employment growth in the local clusters.

– States should focus on upgrading the productivity of all traded clusters that are important locally, rather than to emphasize some clusters over others.

16

“Is There Evidence to Support Porter-type Cluster Policies?”

• Frank McDonald et al, Regional Studies, February 2007.• Study examines evidence related to Porter’s work. • Examines Porter-type hypotheses:

– 1) Clusters that are deep are more likely to be associated with employment growth. A deep cluster has a large number of connections with other firms in the cluster and with supporting agencies, mostly focused on distributing information and knowledge.

– 2) Established clusters are more likely than embryonic and mature clusters to be associated with employment growth.

17

“Is There Evidence to Support Porter-type Cluster Policies?”

• Dependent variable: employment growth by cluster (1=growth, 0=no growth)

• Independent variables (predictors):– Stage of development (embryonic, established, and mature)– Cluster depth (shallow, deep, and unknown)– Industrial sector (manufacturing, services, and other)

• Results: – Cluster depth does not significantly explain employment growth.– Established clusters ARE more likely to be associate with employment growth

than other stages of development.– Clusters in non-manufacturing sectors more likely to show employment growth

than in other sectors.

18

“Is There Evidence to Support Porter-type Cluster Policies?”

• Support for Porter’s hypotheses is mixed:– Cluster depth has little impact on cluster performance (employment

growth), contrary to Porter’s conclusions.– Stage of development does have an impact.

• Implications:– Local networks may not be as important as national and international

networks.– Improving competitiveness is clearly important, but promoting deep

local supply chains may be less important than developing national and international linkages.

19

“The development of industrial clusters and public policy”

• Frank McDonald et al, Entrepreneurship & Regional Development, November 2006.

• The literature has established that the benefits of clusters arise from technological externalities, the ability to increase innovation and learning, and increased flexibility and effectiveness of production and distribution systems.

• If these are the benefits, has public policy been effective in encouraging the development of clusters?

20

“The development of industrial clusters and public policy”

• Clusters have several important characteristics, but deep and extensive networks are usually associated with good economic performance.

• Deep networks – a large number of connections with other firms in the cluster and with supporting agencies, mostly focused on distributing information and knowledge.

• Extensive networks – well-developed flows of goods, services, and information in local supply chains.

21

“The development of industrial clusters and public policy”

• The clusters literature suggests an important role for public policy: – Encourage cooperation by supporting networking between firms, and– Support firm development by providing information on technological

and market developments.

• Hypotheses:– 1) Public policies can affect the degree of firm-to-firm cooperation; the

larger the number of policies and agencies, the higher the cooperation.– 2) Cooperation is associated with economic growth.

22

“The development of industrial clusters and public policy”

• Data gathered from public sources and a survey of firms in the European Union (EU).

• Includes 43 EU industrial clusters.• Questionnaires administered to key players in these clusters.

23

“The development of industrial clusters and public policy”

• Dependent variable: level of cooperation• Independent variables (predictors):

– The number of policies applied to the cluster,– The number of government agencies involved with the cluster,– Controlling for age of the cluster (older clusters more likely to have

developed cooperation)

• Results: Policies appear to have no effect on cooperation. Age (the control variable) is the only significant predictor.

• Uses logistic regression due to qualitative (categorical) dependent variable.

24

“The development of industrial clusters and public policy”

• Dependent variable: growth in the number of firms• Independent variables (predictors):

– The number of policies applied to the cluster,– The number of government agencies involved with the cluster,– Cooperation between firms,– Controlling for age of the cluster (older clusters more likely to have

developed cooperation)

• Results: Policies appear to have no effect on growing the number of firms. Age (the control variable) is the only significant predictor.

25

“The development of industrial clusters and public policy”

• The notion that wide-ranging public policies by many government agencies are helpful for developing industry clusters is not supported by the evidence.

• There appear to be no identifiable generic factors that lead to the development of clusters.

• Consequently, policy makers should determine cluster needs by working with clusters on a case-by-case basis.

26

Questions for Review

• 1) Why is it that wages are higher in traded industries compared with local industries?

• 2) What are the mix effect and the level effect? According to Porter, which one is more important?

• 3) Describe how Porter identifies clusters?

27

Questions for Review

• 4) According to Porter, which is more important: targeting resources to certain clusters, or raising productivity in all important clusters?

• 5) What is meant by the term ‘deep cluster’?• 6) Macdonald (‘Evidence’) tested Porter

hypotheses regarding deep clusters and established clusters. Discuss the evidence.

28

Questions for Review

• 7) Macdonald (‘Entrepreneurship’) examined how the degree of cooperation depends on the number of government policies, the number of government agencies, and the age of firms in the cluster. Which explanatory variables did he find significant?

• 8) What are Macdonald’s policy conclusions? Explain why you agree or disagree?

29

Digression: Determining the Mix Effect and the Level Effect

• The difference between regional wages and national wages can be decomposed into the mix effect and the level effect.

• The mix effect gets at the impact of the regional industry mix on the regional wage, and

• The level effect examines the magnitude of wages by industry, given the industry mix.

• The sum of the two effects adds to the difference between the regional wage and the national wage.

30

Digression: Mix Effect and Level Effect

• Average wage for private-sector industries:– Nashville MSA: $44,366– United States: $45,155– Difference: $789– Question: how much of the difference is due to the Mix

Effect and how much to the Level Effect?

31

Digression: Mix Effect and Level Effect• Level Effect:

– The portion of the wage difference due to higher or lower wages in particular industries compared with the U.S. average for those industries.

– Calculation: Σ(EMPiMSA * (WAGEi

MSA – WAGEius)) / EMPT

MSA

• EMPiMSA = MSA employment for industry i

• WAGEiMSA = MSA wage for industry i

• WAGEiUS = US wage for industry i

• EMPTMSA = Total MSA employment

32

Digression: Mix Effect and Level Effect• Mix Effect: The portion of the wage difference due to the

MSA’s particular distribution of employment by industry. An industry with above average employment in a nationally high wage industry will raise its average wage.– Calculation:

Σ((EMPiMSA – PEMPi

MSA)* (WAGEius – WAGET

us)) / EMPTMSA

• EMPiMSA = MSA employment for industry i

• PEMPiMSA = ‘Predicted’ MSA employment for industry I; MSA

employment in the industry assuming the industry has the same share of employment as does the US. Obtained by multiplying the US industry employment share by MSA total employment.

• WAGEiUS = US wage for industry i

• WAGETUS = US wage across all industries

33

Digression: Mix Effect and Level Effect

34

• Example: Tulsa MSA• Data: 2009 data from the Quarterly Census of

Employment and Wages (QCEW).– www.bls.gov/cew– Census of all employers with a payroll.– Includes employment, payroll, and average wages.

Digression: Mix Effect and Level Effect

35

• Example: Tulsa MSA• Average wage data for 2009:

– Tulsa MSA: $40,717– United States: $45,155

• The difference is -$4,438.• How much of the difference is due to industry

mix? How much of it is due to wage levels by industry?

Digression: Mix Effect and Level Effect

36

•Tulsa MSA wages by industry are lower than the US for every industry except Natural Resources and Trade.•The Level Effect is -$5,144.

Digression: Mix Effect and Level Effect

37

Digression: Mix Effect and Level Effect

38

• Tulsa MSA has a faintly favorable mix of industries.• A higher proportion of well-paying construction,

manufacturing, and natural resources jobs compared with the US.

• But a lower proportion of good-paying Finance jobs and a high proportion of low-paying Trade jobs.

• Mix effect: +$709

Digression: Mix Effect and Level Effect

39

• Results:– Level Effect: -$5,144– Mix Effect: +$ 709– Total difference: -$4,435

• The Level Effect dominates. The Tulsa MSA needs to improve productivity in several industries, particularly Financial Activities, Information, and Professional and Business Services.