Regional Aspects of Innovation and Diffusion Flashcards
Why is the traditional knowledge production function inappropriate?
Traditional knowledge production function
𝑰𝒊 = 𝜶𝑹𝑫𝒊𝜷 𝑯𝑲𝒊𝜸 𝜺𝒊 is inappropriate because it does not account for the geographic dimension of knowledge spillovers (= positive externalities) as it does Jaffe´s specification of the function
-> 𝑰𝒔𝒊 = 𝜶𝑰𝑹𝑫𝜷𝟏 ∗ 𝑼𝑹𝒔𝒊𝜷𝟐 ∗ (𝑼𝑹𝒔𝒊 ∗ 𝑮𝑪𝒔𝒊𝜷𝟑) ∗ 𝜺𝒊.
=> IRD: private corporate R&D input,
UR: university research,
GC: measures the geographic coincidence of university and corporate research,
s is a region,
i is an industry
β1,β2,β3 are coefficients for spillovers
- Innovative activity takes place in s where the knowledge-generating inputs are the greatest
- > the spillovers are the most prevalent
- the traditional knowledge production function resulted to be robust at macro but not micro level, where small firms proved to have a higher ratio of innovative output/inputs.
- Limitation: no understanding of the way in which technological spillovers occur at geographical level
Why is geographic space important for innovation activity?
- due to the presence in geographical regions of positive externalities notoriously enhance the generation and diffusion of tacit knowledge .
- among the positive externalities taking place at local and regional level also thanks to a network of relationships we could mention:
- Spillover of information and technology
- Presence of a common pool of resources
- Presence of pecuniary externalities
- Social , cultural, economic, political and institutional issue
- Proximity among the actors which can also reduce the uncertainty
- the generation of knowledge is peculiar to characteristics of the territory
- > social, cultural, economic, political and institutional issue
- proximity among the actors can also reduce the uncertainty inherent the innovation activity thanks to:
the learning patterns that take place locally
From an economic point of view, what is the motivation for investigating regional aspects of technological diffusion?
- Regional differences in the industrial structure as well as the social, cultural, economic, political and institutional issue -> the adoption of new technologies varies geographically
- Lags behind in the production or adoption of new technology -> may face industrial decline
- Density of sources of knowledge about new technologies is higher
What are clusters, industrial districts and technological parks?
• Industrial districts:
- in the same location of small and medium companies operating in light manufacturing sectors of the economy, involved at different stages and in different ways of the production of a homogenous product
Relational embeddedness of industrial districts refers to the presence of integrated and familial social relationships
Structural embeddedness refers to more formal organizational structures like governance, associativeness and services.
•Clusters
- concept encompassing several possible configurations of companies and institutions;
- it is regional agglomeration of e.g.
high tech firms performing above average on absorption capability, diffusion capacity and access to knowledge
- the way companies compete across locations and the locational choices of multinational companies.
• Technological parks are an agglomeration of research institutes and universities
as well as the services and infrastructures gravitating around them.
- the spatial scale is a crucial element in distinguishing industrial districts and clusters from Regional Innovation Systems (RIS) properly named.
- RISs may embrace numerous clusters and districtsand involve systematic communications and transactions across networks of people and institutions at both intra regional and global level.
- What is the aim of this paper?
Baptista R., Swann P. (1998). Do firms in clusters innovate more?
• The objective of the paper is to determine whether firms located in stronger
clusters are more likely to innovate than other firms
• Use of data on regional employment as a proxy for the strength of the cluster
- What are the benefits and limits of industrial clusters/regions as mentioned in their paper?
Baptista R., Swann P. (1998). Do firms in clusters innovate more?
The benefits can be distinguished between benefits on the demand side an don the supply side.
On the demand side, we should mention externalities (as in Marshall) deriving from:
Labor market pooling
Provision of productive (traded and non traded) inputs deriving from e.g. competitors, firms in related industries, suppliers, customers and research institutions
Knowledge spillovers as well as learning by doing and learning by using processes
Easy access to physical infrastructure
On the supply side, we can mention:
Presence of strong local demand
Visibility
Possibility to exploit sources of good ideas and information coming from key users
Limits of the clustering process are:
Congestion effects
Competition effects
- Which type of data is used?
Baptista R., Swann P. (1998). Do firms in clusters innovate more?
The data adopted is a balanced panel taken from the Science Policy Research Unit (SPRU) innovation database .
This database reports over the count of significant innovation as a measure of innovative output) for 248 companies allocated to an industry (at two digit level) and a region over the period 1975 1982 as well as each firm ´s market share and degree of concentration in their main domestic market.
The second set of data involves employment statistics by region (as collected by the Central Statistical
Office). The rationale behind this choice stands on the fact that the propensity to innovate should be a function of the number of employees in the cluster/region (on the base of the relationship between cluster supply side spillovers and technological infrastructure). Finally, data on total regional population was also collected to account for the regional dimension.
- Which type of analysis do they conduct? Which model do they estimate?
Baptista R., Swann P. (1998). Do firms in clusters innovate more?
Model in which the dependent variable is represented by the number of innovations introduced by each company in each time period.
Linear exponential models
𝐼𝑁𝑁𝑂𝑉𝑖𝑡=𝛽𝑖𝑂𝑊𝑁𝐸𝑀𝑃𝑖𝑡+𝛽2𝑂𝑇𝐻𝐸𝑀𝑃𝐼𝑇+𝛽3𝑀𝑆𝑖𝑡+𝛽4𝐶𝑂𝑁𝐶𝑇𝑖𝑡+σ 𝑠𝛾5𝐷5
The best model specification results to be the negative binomial model allowing for heterogeneity on the mean.
- What are the main findings?
Baptista R., Swann P. (1998). Do firms in clusters innovate more?
firms are more likely to innovate if own sector employment in its home region is strong (OWNEMP positive and significant).
level of regional employment in other sectors (OTHEMP) does not seem to affect innovative activity suggesting that congestion effects might outweigh any
benefit coming from diversification within regions.
Market Share (MS) is also significant and positive, thus stressing the importance of market power on propensity to innovate (feedback effect).
When controlling for individual heterogeneity through two variables measuring respectively entry knowledge stock (KNST) and a dummy variable (DKNST) taking the value of 1 if the firm has previously innovated and 0 otherwise the general performance of the models is improved (see table 4 part D).
- What is the aim of this paper?
Baptista R . (2000). Do innovations diffuse faster within geographical clusters?
In particular, the aim of the paper is to test for the
presence of external learning effects acting at regional level on the diffusion of a new industrial technology.
- What is the theoretical background of this study?
Baptista R . (2000). Do innovations diffuse faster within geographical clusters?
The SORE model with particular reference to the epidemic approach for which:
- A potential user will adopt the technology upon learning of its existence
- Information on the existence of the technology is spread by direct contact between a potential user and a user
Uncertainty
- As technological change occurs at an ever increasing pace, the number of potential innovations and subsequent incremental improvements has expanded beyond any single firm’s ability to choose
- Networking is important to reduce uncertainty
Regional aspects of innovation diffusion, including:
- The Local dimension of networks of innovators
- Presence of externalities
- Bandwagon effects
- R&D spillovers and the geographical distribution of innovative performance. R&D show
the following characteristics:
•Are regionally bounded
•Strong correlation between university and corporate R&D at the state level
•Spillovers become more significant as the geographical area becomes smaller
-Geographical clustering of innovative output
• Positive effects
- What is the methodological approach of this study?
Baptista R . (2000). Do innovations diffuse faster within geographical clusters?
Time of adoption of Computer Numerically Controlled (CNC) machine tools
The estimation technique adopted is the duration model focusing on Hazard function , which expresses the probability that a firm i will adopt the innovation at
time t , conditional on not having adopted the innovation before t , as illustrated in the figure below
The rate of diffusion of the CNC machine tools is not uniform over time
Modeling the probability using the Weibull distribution
At each point in time, firm’s benefits from adopting the new technology are assumed to depend on:
- Characteristics specific to the firm and/or industry , representing the rank effect
- The number of firms already using the technology , representing the stock effect
- The order in which previous firms adopted the technology, reflecting the order effect
- The number of previous adopters located in the firm’s own region , capturing the learning effects due to spillovers
For a firm to adopt a technology at a certain time t , the benefits from adoption
must satisfy two conditions:
- Profitability condition and arbitrage condition
- What are the main findings?
Baptista R . (2000). Do innovations diffuse faster within geographical clusters?
Expected reductions in the price of the technology will delay adoption , as will increases in this price at the present moment
Increases in the total number of firms adopting before t should delay adoption via
the stock and order effects -> existence of a global epidemic effect will counteract such outcome
A higher number of regional adopters by t should increase adoption in this period due to localized external learning effects although such effect could also be counteracted by some kind of regional stock effect
The rank effect will depend on which individual firm characteristics are considered
Firm and industry specific factors influencing the adoption decision
New firms , firms with worn out equipment , and firms expanding capacity will be more likely to adopt a new technology
R&D effort at the establishment level can be taken as measure of the firm’s ability
to process new technological information
Establishments that are part of a larger corporation will experience less uncertainty
Different industries will present different risks
The stock of adopters , together with the expected price of technology, should have a negative effect on the adoption rate
Stock of adopters through learning process is more significant at the regional level
The findings revealed the existence of a positive impact of the regional density of early adopters on the innovation diffusion explained through the geographical nature of knowledge spillovers:
Distribution of adoption times are significantly different by region
Regional dimension of diffusion captured by the significant (and positive) regional learning effects on the adoption of CNC machine tools.
The stock of total previous adopters (= at national level) is not significant.
Technology price and expected technology price are significant and have the expected signs
Regional employment (proxy for the regional industry dimension) is not significant suggesting that learning effects do not seem to be sensitive to the presence of firms in closed or related activities.
Ownership status or R&D issues are not significant, while age and size are.
