The Effects of Industrial Systems on Technology Adoption

The Effects of Industrial Systems on Technology Adoption Joung Yeo No Yonsei University

Industrial Systems in Technology Adoption n The advantages of geographical agglomeration on knowledge spillovers and technology adoption n Not only the size of cluster matters n Does the organization of economic activities within a cluster matter as well?

What is Industrial Systems? Relationship between the internal organization of firms and their connections to one another and to the social structures and institutions of their particular localities Industrial system has 3 dimensions: n n n Local institutions and culture Industrial organization Corporate organization

Effect of Industrial Systems on Technology Adoption n The innovativeness of a region depends not only on the size and density of cluster, but also on how the economic activities within a cluster are organized. • Silicon Valley vs. Boston’s Route 128 (Saxenian, 1994) → Silicon Valley - open, flexible and entrepreneurial environment with many small-, medium-sized plants. - decentralized regional network-based system → Route 128 - Rigid and hierarchical with few dominant players. - independent firm-based system

The Objectives 1. How the regional industrial system affects technology adoption by plants 2. How plants respond differently to geographical agglomeration and regional industrial system depending on their internal resources and information networks

Related Literature Three relevant strands: 1. Work on the sources of agglomeration (Rosenthal and Strange 2001; Dumais, Ellison, and Glaeser 2002; Holmes 2002). 2. Work on other types of knowledge spillovers – Patent citations 3. (Jaffe, Trajtenberg, and Henderson 1993). 3. Work on the effects of industrial systems (Saxenian 1996) – Descriptive studies on Silicon Valley and Route 128

Do industrial systems affect plants’ decisions to adopt technologies?

Hypotheses H 1 a: Adoption of advanced manufacturing technologies is more likely with an increase in employment at small plants in the region. H 1 b: Adoption of advanced manufacturing technologies is more likely with an increase in employment at plants that are single-plant firms in the region.

Hypotheses H 2 a: The effect of regional industrial system is greater for small plants than for large plants. H 2 b: The effects of regional industrial system is greater for single-plant firms than for plants that are part of multi-plant firm.

Hypotheses H 3 a: The effect of knowledge spillovers from prior adopters is greater for plants with less internal resources. H 3 b: The effect of regional agglomeration is greater for plants with less internal resources. H 3 c: The effect of knowledge spillovers from prior adopters is greater for plants that are single-plant firms. H 3 d: The effect of regional agglomeration is greater for plants that are single-plant firms.

Main Finding I. Technology adoption is facilitated by the industrial system that are characterized as follow: 1. 2. 3. That are agglomerated with small plants That are agglomerated with single-unit plants That are agglomerated with plants that are similar II. Plants with the following characteristics are more likely to benefit from the regional agglomeration and knowledge spillovers: 1. 2. Plants that are small Plants that are single-unit

DATA 1993 Survey of Innovation and Advanced Technology n Unique, confidential, proprietary data n Adoption of 22 advanced manufacturing technologies at the plant level n 1902 plants covering an entire manufacturing sector across Canada n Panel nature: years of use for each technology (1984 -1993) → Panel of 3 intervals: 1984 -1986, 1987 -1989, and 1990 -1992.

DATA (Cont’d) • Annual Survey of Manufactures - Collects information on the universe of manufacturing plants in Canada. • National Input-Output Table - Input supply and output demand relationships among industries • Census of Population - Demographic information → Sample size: 1, 902 plants, 22 technologies, 3 time periods ⇒ 106, 188 obs.

Estimating Equation Dependent variable: The probability of technology adoption is a function of: 1. Plant characteristics 2. Local amenities, industry, technology and time fixed effects 3. Regional agglomeration effects 4. Technology spillovers 5. Industrial Systems

Estimating Equation Pr(Adoptionpτirt) =f (Industry. Systemrt, Knoweldge. Spilloverτirt. Regional. Agglomerationrit, Plant. Characteristicsprit, controls)

Technological Dimension 6 technology groups 22 technologies Design & Engineering • CAD/CAE • CAD/CAM • Digital rep. of CAD output used in procurement Fabrication & Assembly • Flexible manuf. cell or system • NC/CNC • Materials working laser • Pick & place robots • Other robots Automated Material Handling • Automated storage and retrieval system • Automated guided vehicle system Manufacturing Inspection Information & System Communication • Automated equip. for inspection of in-process • Automated equip. for inspection of final • LAN for technical data • LAN for factory use • Intercompany computer networks • Programmable controller • Computer for factory floor Integration & Control • Material • Computer requirement integrated planning manufacturing (MRP) (CIM) • Manufacturing • Supervisory resource control & data planning acquisition (MRP II) • Aritifial intelligence & expert systems

Computer Numerically Controlled Machine

Automated Guided Vehicle System

Automated Storage and Retrieval System

Pick and Place Robot Pharmaceutical

Pick and Place Robot Cream cheese

Geographical Dimension 10 provinces 68 Economic Regions 290 Census Divisions Rest of Country Province Economic Region Census Division

Map of Canada

Functional and Industrial Dimension 1. Industrial Dimension based on industry classification 2 -digit (22) and 3 -digit (169) SIC n 2. Functional Dimension n based on similarities in input purchases

A measure of ‘related’ industries: I develop a measure of ‘related’ industries based on the similarity of input purchases across industries. ρij = correlation between industry i and industry j in terms of pattern of input purchases For each industry i, all other industries are classified into three groups: Similar industries : 0. 5 ≤ ρij Moderately similar industries: 0. 2 ≤ ρij < 0. 5 Different industries: ρij < 0. 2

Summary Statistics of Sizes of ‘Related Industries’ Avg. No. of SIC-3 Industries Standard Deviation Similar industries 6. 65 5. 61 Moderately similar industries 8. 89 4. 02 Different industries 92. 46 9. 60 SIC-2 Industry 4. 95 2. 61

Industrial Systems Knowledge Spillovers Factor Conditions Technology Adoption Related and Supporting Industries Organizational Characteristics Demand Conditions

Plant Characteristics is a vector of plant characteristics which includes {Size, No. of commodities, Diversity, Foreign ownership, Single- or Multi-plant firm status}

Fixed Effects • Region • Industry • Technology • Time

Agglomeration Effects Employment in region r at time t-1 Share of scientists & engineers in in the population in region r at time t-1 Value of output of industry i’s input suppliers in region r at time t-1. Value of output of industry i’s output demanders in region r at time t-1.

Technology Spillovers # of adopters of tech τ in Similar industries in region r at time t-1. # of adopters of tech τ in Moderately similar industries in region r at time t-1. # of adopters of tech τ in Different industries in region r at time t-1.

Empirical Results I. Effects of Industrial Systems on Technology Adoption 1. Based on Plant Size 2. Based on Plant Status II. Effects of Regional Agglomeration Conditional on Organizational Capabilities 1. Plant Size 2. Plant Status

1. Main Results Dependent variable: ADOPTIONpτirt Variable Names Coefficient Std. Error Elasticity TECH_Similarτirt-1 . 0388* . 0030 . 0012 TECH_Moderateτirt-1 . 0214* . 0030 . 00065 TECH_Differentτirt-1 -. 0187* . 0043 -. 00057 Regional Empr, t-1 . 0656* (. 0087) . 002 Inputir, t-1 . 0760* (. 0079) . 014 Outputir, t-1 -. 139* (. 0088) -. 013 Engineerr, t-1 4. 39* (. 9700) 1. 07 Observations 16, 188 Log likelihood 68, 172 Notes: 1) * χ2 statistically significant at p < 0. 05 2) Also included are plant characteristics, agglomeration effects, and fixed effects.

Plant Characteristics Dependent variable: ADOPTIONpτirt Variable Names Coefficient Std. Error Elasticity Size . 557* . 0076 . 017 Age -. 0752* . 009 -. 0023 Segment (# of 4 -sic) . 100* . 0085 . 003 Commodity -. 115* . 0091 -. 0035 Small(=1) -. 407* . 020 Foreign(=1) . 086* . 015 Single(=1) -. 182* . 016 Observations 16, 188 Log likelihood 67, 936

I. Effect of Industrial Structure on Technological Adoption Based on Plant Size Main Variable Name Emp_regionirt-1 . 0656* (. 0087) Regional Employment Related Industry Employment . 0015 Small_regionirt-1 . 0616* (. 0102) . 0019 Large_regionirt-1 . 0055 (. 0072) . 00017 Small_relatedirt-1 . 0426* (. 0074) . 0013 Large_relatedirt-1 . 0114 (. 0033) . 00035 Observation 105, 902 Log Likelihood 67, 937 67, 948 67, 960

I. Effect of Industrial Structure on Technological Adoption Based on Plant Status Main Variable Name Emp_regionirt-1 . 0656* (. 0087) Regional Employment Related Industry Employment . 0015 Single_regionirt-1 . 1167* (. 0122) . 0035 Multi_regionirt-1 -. 0430* (. 0112) -. 0013 Single_relatedirt-1 . 0403* (. 0067) . 0012 Multi_relatedirt-1 -. 0241* (. 0035) -. 00073 Observation 105, 902 Log Likelihood 67, 937 67, 999 68, 003

I. Effect of Industrial Structure on Technological Adoption Small vs. Large Main Small Plants Only Large Plants Only . 08866* (. 0138) . 0843* (. 0146) Variable Name Emp_regionirt-1 . 0656* (. 0087) Small_regionirt-1 . 0616* (. 0102) . 0169* (. 0153) -. 1125* (. 0259) Large_regionirt-1 . 0055 (. 0072) -. 0024 (. 0081) . 1978* (. 0300) Observation 105, 902 77, 153 28, 749 Log Likelihood 67, 937 67, 948 40, 318 39, 832 21, 420 21, 323

I. Effect of Industrial Structure on Technological Adoption Single vs. Multi Main Single Plants Only Multi Plants Only . 0741* (. 0125) . 0491* (. 0149) Variable Name Emp_regionirt-1 . 0656* (. 0087) Small_regionirt-1 . 1167* (. 0122) . 0168* (. 0178) . 0092 (. 0231) Large_regionirt-1 -. 0430* (. 0112) -. 597* (. 141) . 0457 (. 0261) Observation 105, 902 71, 527 34, 375 Log Likelihood 67, 937 67, 999 47, 981 48, 041 20, 900 20, 902

II. Effects of Regional Agglomeration Conditional on Plant Size Variable Name (1) (2) (3) (4) (5) Tech Users in Related Ind . 0355* . 0280* . 0336* . 0360* . 0297* (. 0030) (. 0032) Regional Employment . 0656* (. 0087) . 0670* (. 0088) . 0661* (. 0088) . 0041* (. 0088) . 0980* (. 0156) Interaction Terms SMALL* Tech users_rel . 0177* (. 0036) SIZE*Tech users_rel -. 0051* (. 0013) SMALL*EMP_REGION . 0120* (. 0045) SIZE*EMP_REGION -. 0069* (. 0027) Observations 105, 902 105, 902 Log Likelihood 67, 937 67, 962 68, 025 67, 957 67, 945

II. Effects of Regional Agglomeration Conditional on Plant Status Variable Name Regional Employment (2) (4) (5) . 0355* . 0280* . 0360* . 0297* (. 0030) Tech Users in Related Ind (1) (. 0032) (. 0030) (. 0032) . 0656* (. 0087) . 0680* (. 0088) . 0041* (. 0088) . 0980* (. 0156) Interaction Terms SINGLE* Tech users_rel . 0214* (. 0038) SINGLE*EMP_REGION . 0161* (. 0040). 500* (. 0081) . 405* (. 0084) Observations 105, 902 Log Likelihood 67, 937 67, 970 67, 976 67, 993

Conclusion I. Technology adoption is facilitated by the industrial system that are characterized as follow: 1. 2. 3. That are agglomerated with small plants That are agglomerated with single-unit plants That are agglomerated with plants that are similar II. Plants with the following Characteristics are more likely to benefit from the regional agglomeration and knowledge spillovers: 1. 2. Plants that are small Plants that are single-unit